Cosmos/source2/Users/Orvid/Orvid.Graphics/ImageFormats/LibJpeg.cs

/****************************************************************************
 * 
 * LibJpeg.Net
 * Copyright (c) 2008-2011, Bit Miracle
 * 
 * All rights reserved.
 * 
 * Redistribution and use in source and binary forms, with or without 
 * modification, are permitted provided that the following conditions are 
 * met: 
 * Redistributions of source code must retain the above copyright notice,
 * this list of conditions and the following disclaimer. 
 * 
 * Redistributions in binary form must reproduce the above copyright 
 * notice, this list of conditions and the following disclaimer in the 
 * documentation and/or other materials provided with the distribution. 
 * 
 * Neither the name of the Bit Miracle nor the names of its contributors 
 * may be used to endorse or promote products derived from this software 
 * without specific prior written permission. 
 * 
 * THIS Software IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS 
 * IS" AND Any EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED 
 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A 
 * PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL BIT MIRACLE BE 
 * LIABLE FOR Any DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 
 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 
 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 
 * INTERRUPTION) HOWEVER CAUSED AND ON Any THEORY OF LIABILITY, WHETHER IN 
 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 
 * ARISING IN Any WAY OUT OF THE USE OF THIS Software, EVEN IF ADVISED OF 
 * THE POSSIBILITY OF SUCH DAMAGE. 
 ****************************************************************************/
using System;
using System.Collections.Generic;
using System.Text;
using System.IO;
using System.Drawing;
using System.Collections.ObjectModel;

namespace BitMiracle.LibJpeg
{
    #region JpegImage
    /// <summary>
    /// Main class for work with JPEG images.
    /// </summary>
    public sealed class JpegImage : IDisposable
    {
        private bool m_alreadyDisposed;

        /// <summary>
        /// Description of image pixels (samples)
        /// </summary>
        private List<SampleRow> m_rows = new List<SampleRow>();

        private int m_width;
        private int m_height;
        private byte m_bitsPerComponent;
        private byte m_componentsPerSample;
        private ColorSpace m_colorspace;

        // Fields below (m_compressedData, m_decompressedData, m_bitmap) are not initialized in constructors necessarily.
        // Instead direct access to these field you should use corresponding properties (compressedData, decompressedData, bitmap)
        // Such agreement allows to load required data (e.g. compress image) only by request.

        /// <summary>
        /// Bytes of jpeg image. Refreshed when m_compressionParameters changed.
        /// </summary>
        private MemoryStream m_compressedData;

        /// <summary>
        /// Current compression parameters corresponding with compressed data.
        /// </summary>
        private CompressionParameters m_compressionParameters;

        /// <summary>
        /// Bytes of decompressed image (bitmap)
        /// </summary>
        private MemoryStream m_decompressedData;

        /// <summary>
        /// .NET bitmap associated with this image
        /// </summary>
        private Bitmap m_bitmap;

        /// <summary>
        /// Creates <see cref="JpegImage"/> from <see cref="System.Drawing.Bitmap">.NET bitmap</see>
        /// </summary>
        /// <param name="bitmap">Source .NET bitmap.</param>
        public JpegImage(System.Drawing.Bitmap bitmap)
        {
            createFromBitmap(bitmap);
        }

        /// <summary>
        /// Creates <see cref="JpegImage"/> from stream with an arbitrary image data
        /// </summary>
        /// <param name="imageData">Stream containing bytes of image in 
        /// arbitrary format (BMP, Jpeg, GIF, PNG, TIFF, e.t.c)</param>
        public JpegImage(Stream imageData)
        {
            createFromStream(imageData);
        }

        /// <summary>
        /// Creates <see cref="JpegImage"/> from file with an arbitrary image
        /// </summary>
        /// <param name="fileName">Path to file with image in 
        /// arbitrary format (BMP, Jpeg, GIF, PNG, TIFF, e.t.c)</param>
        public JpegImage(string fileName)
        {
            if (fileName == null)
                throw new ArgumentNullException("fileName");

            using (FileStream input = new FileStream(fileName, FileMode.Open))
                createFromStream(input);
        }

        /// <summary>
        /// Creates <see cref="JpegImage"/> from pixels
        /// </summary>
        /// <param name="sampleData">Description of pixels.</param>
        /// <param name="colorspace">ColorSpace of image.</param>
        /// <seealso cref="SampleRow"/>
        public JpegImage(SampleRow[] sampleData, ColorSpace colorspace)
        {
            if (sampleData == null)
                throw new ArgumentNullException("sampleData");

            if (sampleData.Length == 0)
                throw new ArgumentException("sampleData must not be empty");

            if (colorspace == ColorSpace.Unknown)
                throw new ArgumentException("Unknown colorspace");

            m_rows = new List<SampleRow>(sampleData);

            SampleRow firstRow = m_rows[0];
            m_width = firstRow.Length;
            m_height = m_rows.Count;

            Sample firstSample = firstRow[0];
            m_bitsPerComponent = firstSample.BitsPerComponent;
            m_componentsPerSample = firstSample.ComponentCount;
            m_colorspace = colorspace;
        }

        /// <summary>
        /// Creates <see cref="JpegImage"/> from <see cref="System.Drawing.Bitmap">.NET bitmap</see>
        /// </summary>
        /// <param name="bitmap">Source .NET bitmap.</param>
        /// <returns>Created instance of <see cref="JpegImage"/> class.</returns>
        /// <remarks>Same as corresponding <see cref="M:BitMiracle.LibJpeg.JpegImage.#ctor(System.Drawing.Bitmap)">constructor</see>.</remarks>
        public static JpegImage FromBitmap(Bitmap bitmap)
        {
            return new JpegImage(bitmap);
        }

        /// <summary>
        /// Frees and releases all resources allocated by this <see cref="JpegImage"/>
        /// </summary>
        public void Dispose()
        {
            Dispose(true);
            GC.SuppressFinalize(this);
        }

        private void Dispose(bool disposing)
        {
            if (!m_alreadyDisposed)
            {
                if (disposing)
                {
                    // dispose managed resources
                    if (m_compressedData != null)
                        m_compressedData.Dispose();

                    if (m_decompressedData != null)
                        m_decompressedData.Dispose();
                    if (m_bitmap != null)
                        m_bitmap.Dispose();
                }

                // free native resources
                m_compressionParameters = null;
                m_compressedData = null;
                m_decompressedData = null;
                m_bitmap = null;
                m_rows = null;
                m_alreadyDisposed = true;
            }
        }

        /// <summary>
        /// Gets the width of image in <see cref="Sample">samples</see>.
        /// </summary>
        /// <value>The width of image.</value>
        public int Width
        {
            get
            {
                return m_width;
            }
            internal set
            {
                m_width = value;
            }
        }

        /// <summary>
        /// Gets the height of image in <see cref="Sample">samples</see>.
        /// </summary>
        /// <value>The height of image.</value>
        public int Height
        {
            get
            {
                return m_height;
            }
            internal set
            {
                m_height = value;
            }
        }

        /// <summary>
        /// Gets the number of color components per <see cref="Sample">sample</see>.
        /// </summary>
        /// <value>The number of color components per sample.</value>
        public byte ComponentsPerSample
        {
            get
            {
                return m_componentsPerSample;
            }
            internal set
            {
                m_componentsPerSample = value;
            }
        }

        /// <summary>
        /// Gets the number of bits per color component of <see cref="Sample">sample</see>.
        /// </summary>
        /// <value>The number of bits per color component.</value>
        public byte BitsPerComponent
        {
            get
            {
                return m_bitsPerComponent;
            }
            internal set
            {
                m_bitsPerComponent = value;
            }
        }

        /// <summary>
        /// Gets the colorspace of image.
        /// </summary>
        /// <value>The colorspace of image.</value>
        public ColorSpace Colorspace
        {
            get
            {
                return m_colorspace;
            }
            internal set
            {
                m_colorspace = value;
            }
        }


        /// <summary>
        /// Retrieves the required row of image.
        /// </summary>
        /// <param name="rowNumber">The number of row.</param>
        /// <returns>Image row of samples.</returns>
        public SampleRow GetRow(int rowNumber)
        {
            return m_rows[rowNumber];
        }

        /// <summary>
        /// Writes compressed JPEG image to stream.
        /// </summary>
        /// <param name="output">Output stream.</param>
        public void WriteJpeg(Stream output)
        {
            WriteJpeg(output, new CompressionParameters());
        }

        /// <summary>
        /// Compresses image to JPEG with given parameters and writes it to stream.
        /// </summary>
        /// <param name="output">Output stream.</param>
        /// <param name="parameters">The parameters of compression.</param>
        public void WriteJpeg(Stream output, CompressionParameters parameters)
        {
            compress(parameters);
            compressedData.WriteTo(output);
        }

        /// <summary>
        /// Writes decompressed image data as bitmap to stream.
        /// </summary>
        /// <param name="output">Output stream.</param>
        public void WriteBitmap(Stream output)
        {
            decompressedData.WriteTo(output);
        }

        /// <summary>
        /// Retrieves image as .NET Bitmap.
        /// </summary>
        /// <returns>.NET Bitmap</returns>
        public Bitmap ToBitmap()
        {
            return bitmap.Clone() as Bitmap;
        }

        private MemoryStream compressedData
        {
            get
            {
                if (m_compressedData == null)
                    compress(new CompressionParameters());
                return m_compressedData;
            }
        }

        private MemoryStream decompressedData
        {
            get
            {
                if (m_decompressedData == null)
                    fillDecompressedData();
                return m_decompressedData;
            }
        }

        private Bitmap bitmap
        {
            get
            {
                if (m_bitmap == null)
                {
                    long position = compressedData.Position;
                    m_bitmap = new Bitmap(compressedData);
                    compressedData.Seek(position, SeekOrigin.Begin);
                }

                return m_bitmap;
            }
        }

        /// <summary>
        /// Needs for DecompressorToJpegImage class
        /// </summary>
        internal void addSampleRow(SampleRow row)
        {
            if (row == null)
                throw new ArgumentNullException("row");

            m_rows.Add(row);
        }

        /// <summary>
        /// Checks if imageData contains jpeg image
        /// </summary>
        private static bool isCompressed(Stream imageData)
        {
            if (imageData == null)
                return false;

            if (imageData.Length <= 2)
                return false;

            imageData.Seek(0, SeekOrigin.Begin);
            int first = imageData.ReadByte();
            int second = imageData.ReadByte();
            return (first == 0xFF && second == (int)JpegMarkerType.SOI);
        }

        private void createFromBitmap(Bitmap bitmap)
        {
            initializeFromBitmap(bitmap);
            compress(new CompressionParameters());
        }

        private void createFromStream(Stream imageData)
        {
            if (imageData == null)
                throw new ArgumentNullException("imageData");

            if (isCompressed(imageData))
            {
                m_compressedData = Utils.CopyStream(imageData);
                decompress();
            }
            else
            {
                createFromBitmap(new Bitmap(imageData));
            }
        }

        private void initializeFromBitmap(Bitmap bitmap)
        {
            if (bitmap == null)
                throw new ArgumentNullException("bitmap");

            m_bitmap = bitmap;
            m_width = m_bitmap.Width;
            m_height = m_bitmap.Height;
            processPixelFormat(bitmap.PixelFormat);
            fillSamplesFromBitmap();
        }

        private void compress(CompressionParameters parameters)
        {

            if (m_rows == null)
                throw new Exception("'m_rows' Cannot be null!");

            if (m_rows.Count == 0)
                throw new ArgumentException("'m_rows' Cannot be empty!");
            

            RawImage source = new RawImage(m_rows, m_colorspace);
            compress(source, parameters);
        }

        private void compress(IRawImage source, CompressionParameters parameters)
        {
            if (source == null)
                throw new ArgumentNullException("'source' Cannot be null!");

            if (!needCompressWith(parameters))
                return;

            m_compressedData = new MemoryStream();
            m_compressionParameters = new CompressionParameters(parameters);

            Jpeg jpeg = new Jpeg();
            jpeg.CompressionParameters = m_compressionParameters;
            jpeg.Compress(source, m_compressedData);
        }

        private bool needCompressWith(CompressionParameters parameters)
        {
            return m_compressedData == null ||
                   m_compressionParameters == null ||
                   !m_compressionParameters.Equals(parameters);
        }

        private void decompress()
        {
            Jpeg jpeg = new Jpeg();
            jpeg.Decompress(compressedData, new DecompressorToJpegImage(this));
        }

        private void fillDecompressedData()
        {
            if (m_decompressedData != null)
                throw new Exception("'m_decompressedData' Is not null!");

            m_decompressedData = new MemoryStream();
            BitmapDestination dest = new BitmapDestination(m_decompressedData);

            Jpeg jpeg = new Jpeg();
            jpeg.Decompress(compressedData, dest);
        }

        private void processPixelFormat(System.Drawing.Imaging.PixelFormat pixelFormat)
        {
            //See GdiPlusPixelFormats.h for details

            if (pixelFormat == System.Drawing.Imaging.PixelFormat.Format16bppGrayScale)
            {
                m_bitsPerComponent = 16;
                m_componentsPerSample = 1;
                m_colorspace = ColorSpace.Grayscale;
                return;
            }

            byte formatIndexByte = (byte)((int)pixelFormat & 0x000000FF);
            byte pixelSizeByte = (byte)((int)pixelFormat & 0x0000FF00);

            if (pixelSizeByte == 32 && formatIndexByte == 15) //PixelFormat32bppCMYK (15 | (32 << 8))
            {
                m_bitsPerComponent = 8;
                m_componentsPerSample = 4;
                m_colorspace = ColorSpace.CMYK;
                return;
            }

            m_bitsPerComponent = 8;
            m_componentsPerSample = 3;
            m_colorspace = ColorSpace.RGB;

            if (pixelSizeByte == 16)
                m_bitsPerComponent = 6;
            else if (pixelSizeByte == 24 || pixelSizeByte == 32)
                m_bitsPerComponent = 8;
            else if (pixelSizeByte == 48 || pixelSizeByte == 64)
                m_bitsPerComponent = 16;
        }

        private void fillSamplesFromBitmap()
        {
            if (m_bitmap == null)
                throw new Exception("Field 'm_bitmap' Cannot be null!");

            for (int y = 0; y < Height; ++y)
            {
                short[] samples = new short[Width * 3];
                for (int x = 0; x < Width; ++x)
                {
                    Color color = m_bitmap.GetPixel(x, y);
                    samples[x * 3] = color.R;
                    samples[x * 3 + 1] = color.G;
                    samples[x * 3 + 2] = color.B;
                }
                m_rows.Add(new SampleRow(samples, m_bitsPerComponent, m_componentsPerSample));
            }
        }
    }
    #endregion

    #region BitmapDestination
    class BitmapDestination : IDecompressor
    {
        /// <summary>
        /// Target file spec; filled in by djpeg.c after object is created.
        /// </summary>
        private Stream m_output;
        private byte[][] m_pixels;
        /// <summary>
        /// physical width of one row in the BMP file
        /// </summary>
        private int m_rowWidth;
        /// <summary>
        /// next row# to write to virtual array
        /// </summary>
        private int m_currentRow;
        private LoadedImageAttributes m_parameters;

        public BitmapDestination(Stream output)
        {
            m_output = output;
        }

        public override Stream Output
        {
            get
            {
                return m_output;
            }
        }

        public override void SetImageAttributes(LoadedImageAttributes parameters)
        {
            if (parameters == null)
                throw new ArgumentNullException("parameters");

            m_parameters = parameters;
        }

        /// <summary>
        /// Startup: normally writes the file header.
        /// In this module we may as well postpone everything until finish_output.
        /// </summary>
        public override void BeginWrite()
        {
            //Determine width of rows in the BMP file (padded to 4-byte boundary).
            m_rowWidth = m_parameters.Width * m_parameters.Components;
            while (m_rowWidth % 4 != 0)
                m_rowWidth++;

            m_pixels = new byte[m_rowWidth][];
            for (int i = 0; i < m_rowWidth; i++)
                m_pixels[i] = new byte[m_parameters.Height];

            m_currentRow = 0;
        }

        /// <summary>
        /// Write some pixel data.
        /// </summary>
        public override void ProcessPixelsRow(byte[] row)
        {
            if (m_parameters.Colorspace == ColorSpace.Grayscale || m_parameters.QuantizeColors)
            {
                putGrayRow(row);
            }
            else
            {
                if (m_parameters.Colorspace == ColorSpace.CMYK)
                    putCmykRow(row);
                else
                    putRgbRow(row);
            }

            ++m_currentRow;
        }

        /// <summary>
        /// Finish up at the end of the file.
        /// Here is where we really output the BMP file.
        /// </summary>
        public override void EndWrite()
        {
            writeHeader();
            writePixels();

            /* Make sure we wrote the output file OK */
            m_output.Flush();
        }


        /// <summary>
        /// This version is for grayscale OR quantized color output
        /// </summary>
        private void putGrayRow(byte[] row)
        {
            for (int i = 0; i < m_parameters.Width; ++i)
                m_pixels[i][m_currentRow] = row[i];
        }

        /// <summary>
        /// This version is for writing 24-bit pixels
        /// </summary>
        private void putRgbRow(byte[] row)
        {
            /* Transfer data.  Note destination values must be in BGR order
             * (even though Microsoft's own documents say the opposite).
             */
            for (int i = 0; i < m_parameters.Width; ++i)
            {
                int firstComponent = i * 3;
                byte red = row[firstComponent];
                byte green = row[firstComponent + 1];
                byte blue = row[firstComponent + 2];
                m_pixels[firstComponent][m_currentRow] = blue;
                m_pixels[firstComponent + 1][m_currentRow] = green;
                m_pixels[firstComponent + 2][m_currentRow] = red;
            }
        }

        /// <summary>
        /// This version is for writing 24-bit pixels
        /// </summary>
        private void putCmykRow(byte[] row)
        {
            /* Transfer data.  Note destination values must be in BGR order
             * (even though Microsoft's own documents say the opposite).
             */
            for (int i = 0; i < m_parameters.Width; ++i)
            {
                int firstComponent = i * 4;
                m_pixels[firstComponent][m_currentRow] = row[firstComponent + 2];
                m_pixels[firstComponent + 1][m_currentRow] = row[firstComponent + 1];
                m_pixels[firstComponent + 2][m_currentRow] = row[firstComponent + 0];
                m_pixels[firstComponent + 3][m_currentRow] = row[firstComponent + 3];
            }
        }

        /// <summary>
        /// Write a Windows-style BMP file header, including colormap if needed
        /// </summary>
        private void writeHeader()
        {
            int bits_per_pixel;
            int cmap_entries;

            /* Compute colormap size and total file size */
            if (m_parameters.Colorspace == ColorSpace.Grayscale || m_parameters.QuantizeColors)
            {
                bits_per_pixel = 8;
                cmap_entries = 256;
            }
            else
            {
                cmap_entries = 0;

                if (m_parameters.Colorspace == ColorSpace.RGB)
                    bits_per_pixel = 24;
                else if (m_parameters.Colorspace == ColorSpace.CMYK)
                    bits_per_pixel = 32;
                else
                    throw new InvalidOperationException();
            }

            byte[] infoHeader = null;
            if (m_parameters.Colorspace == ColorSpace.RGB)
                infoHeader = createBitmapInfoHeader(bits_per_pixel, cmap_entries);
            else
                infoHeader = createBitmapV4InfoHeader(bits_per_pixel);

            /* File size */
            const int fileHeaderSize = 14;
            int infoHeaderSize = infoHeader.Length;
            int paletteSize = cmap_entries * 4;
            int offsetToPixels = fileHeaderSize + infoHeaderSize + paletteSize; /* Header and colormap */
            int fileSize = offsetToPixels + m_rowWidth * m_parameters.Height;

            byte[] fileHeader = createBitmapFileHeader(offsetToPixels, fileSize);

            m_output.Write(fileHeader, 0, fileHeader.Length);
            m_output.Write(infoHeader, 0, infoHeader.Length);

            if (cmap_entries > 0)
                writeColormap(cmap_entries, 4);
        }

        private static byte[] createBitmapFileHeader(int offsetToPixels, int fileSize)
        {
            byte[] bmpfileheader = new byte[14];
            bmpfileheader[0] = 0x42;    /* first 2 bytes are ASCII 'B', 'M' */
            bmpfileheader[1] = 0x4D;
            PUT_4B(bmpfileheader, 2, fileSize);
            /* we leave bfReserved1 & bfReserved2 = 0 */
            PUT_4B(bmpfileheader, 10, offsetToPixels); /* bfOffBits */
            return bmpfileheader;
        }

        private byte[] createBitmapInfoHeader(int bits_per_pixel, int cmap_entries)
        {
            byte[] bmpinfoheader = new byte[40];
            fillBitmapInfoHeader(bits_per_pixel, cmap_entries, bmpinfoheader);
            return bmpinfoheader;
        }

        private void fillBitmapInfoHeader(int bitsPerPixel, int cmap_entries, byte[] infoHeader)
        {
            /* Fill the info header (Microsoft calls this a BITMAPINFOHEADER) */
            PUT_2B(infoHeader, 0, infoHeader.Length);   /* biSize */
            PUT_4B(infoHeader, 4, m_parameters.Width); /* biWidth */
            PUT_4B(infoHeader, 8, m_parameters.Height); /* biHeight */
            PUT_2B(infoHeader, 12, 1);   /* biPlanes - must be 1 */
            PUT_2B(infoHeader, 14, bitsPerPixel); /* biBitCount */
            /* we leave biCompression = 0, for none */
            /* we leave biSizeImage = 0; this is correct for uncompressed data */

            if (m_parameters.DensityUnit == DensityUnit.DotsCm)
            {
                /* if have density in dots/cm, then */
                PUT_4B(infoHeader, 24, m_parameters.DensityX * 100); /* XPels/M */
                PUT_4B(infoHeader, 28, m_parameters.DensityY * 100); /* XPels/M */
            }
            PUT_2B(infoHeader, 32, cmap_entries); /* biClrUsed */
            /* we leave biClrImportant = 0 */
        }

        private byte[] createBitmapV4InfoHeader(int bitsPerPixel)
        {
            byte[] infoHeader = new byte[40 + 68];
            fillBitmapInfoHeader(bitsPerPixel, 0, infoHeader);

            PUT_4B(infoHeader, 56, 0x02); /* CSType == 0x02 (CMYK) */

            return infoHeader;
        }

        /// <summary>
        /// Write the colormap.
        /// Windows uses BGR0 map entries; OS/2 uses BGR entries.
        /// </summary>
        private void writeColormap(int map_colors, int map_entry_size)
        {
            byte[][] colormap = m_parameters.Colormap;
            int num_colors = m_parameters.ActualNumberOfColors;

            int i = 0;
            if (colormap != null)
            {
                if (m_parameters.ComponentsPerSample == 3)
                {
                    /* Normal case with RGB colormap */
                    for (i = 0; i < num_colors; i++)
                    {
                        m_output.WriteByte(colormap[2][i]);
                        m_output.WriteByte(colormap[1][i]);
                        m_output.WriteByte(colormap[0][i]);
                        if (map_entry_size == 4)
                            m_output.WriteByte(0);
                    }
                }
                else
                {
                    /* Grayscale colormap (only happens with grayscale quantization) */
                    for (i = 0; i < num_colors; i++)
                    {
                        m_output.WriteByte(colormap[0][i]);
                        m_output.WriteByte(colormap[0][i]);
                        m_output.WriteByte(colormap[0][i]);
                        if (map_entry_size == 4)
                            m_output.WriteByte(0);
                    }
                }
            }
            else
            {
                /* If no colormap, must be grayscale data.  Generate a linear "map". */
                for (i = 0; i < 256; i++)
                {
                    m_output.WriteByte((byte)i);
                    m_output.WriteByte((byte)i);
                    m_output.WriteByte((byte)i);
                    if (map_entry_size == 4)
                        m_output.WriteByte(0);
                }
            }

            /* Pad colormap with zeros to ensure specified number of colormap entries */
            if (i > map_colors)
                throw new InvalidOperationException("Too many colors");

            for (; i < map_colors; i++)
            {
                m_output.WriteByte(0);
                m_output.WriteByte(0);
                m_output.WriteByte(0);
                if (map_entry_size == 4)
                    m_output.WriteByte(0);
            }
        }

        private void writePixels()
        {
            for (int row = m_parameters.Height - 1; row >= 0; --row)
                for (int col = 0; col < m_rowWidth; ++col)
                    m_output.WriteByte(m_pixels[col][row]);
        }


        private static void PUT_2B(byte[] array, int offset, int value)
        {
            array[offset] = (byte)((value) & 0xFF);
            array[offset + 1] = (byte)(((value) >> 8) & 0xFF);
        }

        private static void PUT_4B(byte[] array, int offset, int value)
        {
            array[offset] = (byte)((value) & 0xFF);
            array[offset + 1] = (byte)(((value) >> 8) & 0xFF);
            array[offset + 2] = (byte)(((value) >> 16) & 0xFF);
            array[offset + 3] = (byte)(((value) >> 24) & 0xFF);
        }
    }
    #endregion

    #region BitStream
    class BitStream : IDisposable
    {
        private bool m_alreadyDisposed;

        private const int bitsInByte = 8;
        private Stream m_stream;
        private int m_positionInByte;

        private int m_size;

        public BitStream()
        {
            m_stream = new MemoryStream();
        }

        public BitStream(byte[] buffer)
        {
            if (buffer == null)
                throw new ArgumentNullException("buffer");

            m_stream = new MemoryStream(buffer);
            m_size = bitsAllocated();
        }

        public void Dispose()
        {
            this.Dispose(true);
            GC.SuppressFinalize(this);
        }

        protected virtual void Dispose(bool disposing)
        {
            if (!m_alreadyDisposed)
            {
                if (disposing)
                {
                    if (m_stream != null)
                        m_stream.Dispose();
                }

                m_stream = null;
                m_alreadyDisposed = true;
            }
        }

        public int Size()
        {
            return m_size;
        }

        public Stream UnderlyingStream
        {
            get
            {
                return m_stream;
            }
        }

        public virtual int Read(int bitCount)
        {
            if (Tell() + bitCount > bitsAllocated())
                throw new ArgumentOutOfRangeException("bitCount");

            return read(bitCount);
        }

        public int Write(int bitStorage, int bitCount)
        {
            if (bitCount == 0)
                return 0;

            const int maxBitsInStorage = sizeof(int) * bitsInByte;
            if (bitCount > maxBitsInStorage)
                throw new ArgumentOutOfRangeException("bitCount");

            for (int i = 0; i < bitCount; ++i)
            {
                byte bit = (byte)((bitStorage << (maxBitsInStorage - (bitCount - i))) >> (maxBitsInStorage - 1));
                if (!writeBit(bit))
                    return i;
            }

            return bitCount;
        }

        public void Seek(int pos, SeekOrigin mode)
        {
            switch (mode)
            {
                case SeekOrigin.Begin:
                    seekSet(pos);
                    break;

                case SeekOrigin.Current:
                    seekCurrent(pos);
                    break;

                case SeekOrigin.End:
                    seekSet(Size() + pos);
                    break;
            }
        }

        public int Tell()
        {
            return (int)m_stream.Position * bitsInByte + m_positionInByte;
        }

        private int bitsAllocated()
        {
            return (int)m_stream.Length * bitsInByte;
        }

        private int read(int bitsCount)
        {
            //Codes are packed into a continuous bit stream, high-order bit first. 
            //This stream is then divided into 8-bit bytes, high-order bit first. 
            //Thus, codes can straddle byte boundaries arbitrarily. After the EOD marker (code value 257), 
            //any leftover bits in the final byte are set to 0.

            if (bitsCount < 0 || bitsCount > 32)
                throw new ArgumentOutOfRangeException("bitsCount");

            if (bitsCount == 0)
                return 0;

            int bitsRead = 0;
            int result = 0;
            byte[] bt = new byte[1];
            while (bitsRead == 0 || (bitsRead - m_positionInByte < bitsCount))
            {
                m_stream.Read(bt, 0, 1);

                result = (result << bitsInByte);
                result += bt[0];

                bitsRead += 8;
            }

            m_positionInByte = (m_positionInByte + bitsCount) % 8;
            if (m_positionInByte != 0)
            {
                result = (result >> (bitsInByte - m_positionInByte));

                m_stream.Seek(-1, SeekOrigin.Current);
            }

            if (bitsCount < 32)
            {
                int mask = ((1 << bitsCount) - 1);
                result = result & mask;
            }

            return result;
        }

        private bool writeBit(byte bit)
        {
            if (m_stream.Position == m_stream.Length)
            {
                byte[] bt = { (byte)(bit << (bitsInByte - 1)) };
                m_stream.Write(bt, 0, 1);
                m_stream.Seek(-1, SeekOrigin.Current);
            }
            else
            {
                byte[] bt = { 0 };
                m_stream.Read(bt, 0, 1);
                m_stream.Seek(-1, SeekOrigin.Current);

                int shift = (bitsInByte - m_positionInByte - 1) % bitsInByte;
                byte maskByte = (byte)(bit << shift);

                bt[0] |= maskByte;
                m_stream.Write(bt, 0, 1);
                m_stream.Seek(-1, SeekOrigin.Current);
            }

            Seek(1, SeekOrigin.Current);

            int currentPosition = Tell();
            if (currentPosition > m_size)
                m_size = currentPosition;

            return true;
        }

        private void seekSet(int pos)
        {
            if (pos < 0)
                throw new ArgumentOutOfRangeException("pos");

            int byteDisplacement = pos / bitsInByte;
            m_stream.Seek(byteDisplacement, SeekOrigin.Begin);

            int shiftInByte = pos - byteDisplacement * bitsInByte;
            m_positionInByte = shiftInByte;
        }

        private void seekCurrent(int pos)
        {
            int result = Tell() + pos;
            if (result < 0 || result > bitsAllocated())
                throw new ArgumentOutOfRangeException("pos");

            seekSet(result);
        }
    }
    #endregion

    #region CoefControllerImpl
    class CoefControllerImpl : JpegCompressorCoefController
    {
        private BufferMode m_passModeSetByLastStartPass;
        private JpegCompressor m_cinfo;

        private int m_iMCU_row_num;    /* iMCU row # within image */
        private int m_mcu_ctr;     /* counts MCUs processed in current row */
        private int m_MCU_vert_offset;        /* counts MCU rows within iMCU row */
        private int m_MCU_rows_per_iMCU_row;  /* number of such rows needed */

        /* For single-pass compression, it's sufficient to buffer just one MCU
        * (although this may prove a bit slow in practice).  We allocate a
        * workspace of CompressorMaxBlocksInMCU coefficient blocks, and reuse it for each
        * MCU constructed and sent.  (On 80x86, the workspace is FAR even though
        * it's not really very big; this is to keep the module interfaces unchanged
        * when a large coefficient buffer is necessary.)
        * In multi-pass modes, this array points to the current MCU's blocks
        * within the virtual arrays.
        */
        private JpegBlock[][] m_MCU_buffer = new JpegBlock[JpegConstants.CompressorMaxBlocksInMCU][];

        /* In multi-pass modes, we need a virtual block array for each component. */
        private JpegVirtualArray<JpegBlock>[] m_whole_image = new JpegVirtualArray<JpegBlock>[JpegConstants.MaxComponents];

        public CoefControllerImpl(JpegCompressor cinfo, bool need_full_buffer)
        {
            m_cinfo = cinfo;

            /* Create the coefficient buffer. */
            if (need_full_buffer)
            {
                /* Allocate a full-image virtual array for each component, */
                /* padded to a multiple of samp_factor DCT blocks in each direction. */
                for (int ci = 0; ci < cinfo.m_num_components; ci++)
                {
                    m_whole_image[ci] = JpegCommonBase.CreateBlocksArray(
                        JpegUtils.jround_up(cinfo.Component_info[ci].Width_in_blocks, cinfo.Component_info[ci].H_samp_factor),
                        JpegUtils.jround_up(cinfo.Component_info[ci].height_in_blocks, cinfo.Component_info[ci].V_samp_factor));
                    m_whole_image[ci].ErrorProcessor = cinfo;
                }
            }
            else
            {
                /* We only need a single-MCU buffer. */
                JpegBlock[] buffer = new JpegBlock[JpegConstants.CompressorMaxBlocksInMCU];
                for (int i = 0; i < JpegConstants.CompressorMaxBlocksInMCU; i++)
                    buffer[i] = new JpegBlock();

                for (int i = 0; i < JpegConstants.CompressorMaxBlocksInMCU; i++)
                {
                    m_MCU_buffer[i] = new JpegBlock[JpegConstants.CompressorMaxBlocksInMCU - i];
                    for (int j = i; j < JpegConstants.CompressorMaxBlocksInMCU; j++)
                        m_MCU_buffer[i][j - i] = buffer[j];
                }

                /* flag for no virtual arrays */
                m_whole_image[0] = null;
            }
        }

        // Initialize for a processing pass.
        public virtual void start_pass(BufferMode pass_mode)
        {
            m_iMCU_row_num = 0;
            start_iMCU_row();

            switch (pass_mode)
            {
                case BufferMode.PassThru:
                    if (m_whole_image[0] != null)
                        throw new Exception("Bogus buffer control mode");
                    break;

                case BufferMode.SaveAndPass:
                    if (m_whole_image[0] == null)
                        throw new Exception("Bogus buffer control mode");
                    break;

                case BufferMode.CrankDest:
                    if (m_whole_image[0] == null)
                        throw new Exception("Bogus buffer control mode");
                    break;

                default:
                    throw new Exception("Bogus buffer control mode");
            }

            m_passModeSetByLastStartPass = pass_mode;
        }

        public virtual bool compress_data(byte[][][] input_buf)
        {
            switch (m_passModeSetByLastStartPass)
            {
                case BufferMode.PassThru:
                    return compressDataImpl(input_buf);

                case BufferMode.SaveAndPass:
                    return compressFirstPass(input_buf);

                case BufferMode.CrankDest:
                    return compressOutput();
            }

            return false;
        }

        /// <summary>
        /// Process some data in the single-pass case.
        /// We process the equivalent of one fully interleaved MCU row ("iMCU" row)
        /// per call, ie, v_samp_factor block rows for each component in the image.
        /// Returns true if the iMCU row is completed, false if suspended.
        /// 
        /// NB: input_buf contains a plane for each component in image,
        /// which we index according to the component's SOF position.
        /// </summary>
        private bool compressDataImpl(byte[][][] input_buf)
        {
            int last_MCU_col = m_cinfo.m_MCUs_per_row - 1;
            int last_iMCU_row = m_cinfo.m_total_iMCU_rows - 1;

            /* Loop to write as much as one whole iMCU row */
            for (int yoffset = m_MCU_vert_offset; yoffset < m_MCU_rows_per_iMCU_row; yoffset++)
            {
                for (int MCU_col_num = m_mcu_ctr; MCU_col_num <= last_MCU_col; MCU_col_num++)
                {
                    /* Determine where data comes from in input_buf and do the DCT thing.
                     * Each call on forward_DCT processes a horizontal row of DCT blocks
                     * as wide as an MCU; we rely on having allocated the MCU_buffer[] blocks
                     * sequentially.  Dummy blocks at the right or bottom edge are filled in
                     * specially.  The data in them does not matter for image reconstruction,
                     * so we fill them with values that will encode to the smallest amount of
                     * data, viz: all zeroes in the AC entries, DC entries equal to previous
                     * block's DC value.  (Thanks to Thomas Kinsman for this idea.)
                     */
                    int blkn = 0;
                    for (int ci = 0; ci < m_cinfo.m_comps_in_scan; ci++)
                    {
                        JpegComponent componentInfo = m_cinfo.Component_info[m_cinfo.m_cur_comp_info[ci]];
                        int blockcnt = (MCU_col_num < last_MCU_col) ? componentInfo.MCU_width : componentInfo.last_col_width;
                        int xpos = MCU_col_num * componentInfo.MCU_sample_width;
                        int ypos = yoffset * JpegConstants.DCTSize;

                        for (int yindex = 0; yindex < componentInfo.MCU_height; yindex++)
                        {
                            if (m_iMCU_row_num < last_iMCU_row || yoffset + yindex < componentInfo.last_row_height)
                            {
                                m_cinfo.m_fdct.forward_DCT(componentInfo.Quant_tbl_no, input_buf[componentInfo.Component_index],
                                    m_MCU_buffer[blkn], ypos, xpos, blockcnt);

                                if (blockcnt < componentInfo.MCU_width)
                                {
                                    /* Create some dummy blocks at the right edge of the image. */
                                    for (int i = 0; i < (componentInfo.MCU_width - blockcnt); i++)
                                        Array.Clear(m_MCU_buffer[blkn + blockcnt][i].data, 0, m_MCU_buffer[blkn + blockcnt][i].data.Length);

                                    for (int bi = blockcnt; bi < componentInfo.MCU_width; bi++)
                                        m_MCU_buffer[blkn + bi][0][0] = m_MCU_buffer[blkn + bi - 1][0][0];
                                }
                            }
                            else
                            {
                                /* Create a row of dummy blocks at the bottom of the image. */
                                for (int i = 0; i < componentInfo.MCU_width; i++)
                                    Array.Clear(m_MCU_buffer[blkn][i].data, 0, m_MCU_buffer[blkn][i].data.Length);

                                for (int bi = 0; bi < componentInfo.MCU_width; bi++)
                                    m_MCU_buffer[blkn + bi][0][0] = m_MCU_buffer[blkn - 1][0][0];
                            }

                            blkn += componentInfo.MCU_width;
                            ypos += JpegConstants.DCTSize;
                        }
                    }

                    /* Try to write the MCU.  In event of a suspension failure, we will
                     * re-DCT the MCU on restart (a bit inefficient, could be fixed...)
                     */
                    if (!m_cinfo.m_entropy.encode_mcu(m_MCU_buffer))
                    {
                        /* Suspension forced; update state counters and exit */
                        m_MCU_vert_offset = yoffset;
                        m_mcu_ctr = MCU_col_num;
                        return false;
                    }
                }

                /* Completed an MCU row, but perhaps not an iMCU row */
                m_mcu_ctr = 0;
            }

            /* Completed the iMCU row, advance counters for next one */
            m_iMCU_row_num++;
            start_iMCU_row();
            return true;
        }

        /// <summary>
        /// Process some data in the first pass of a multi-pass case.
        /// We process the equivalent of one fully interleaved MCU row ("iMCU" row)
        /// per call, ie, v_samp_factor block rows for each component in the image.
        /// This amount of data is read from the source buffer, DCT'd and quantized,
        /// and saved into the virtual arrays.  We also generate suitable dummy blocks
        /// as needed at the right and lower edges.  (The dummy blocks are constructed
        /// in the virtual arrays, which have been padded appropriately.)  This makes
        /// it possible for subsequent passes not to worry about real vs. dummy blocks.
        /// 
        /// We must also emit the data to the entropy encoder.  This is conveniently
        /// done by calling compress_output() after we've loaded the current strip
        /// of the virtual arrays.
        /// 
        /// NB: input_buf contains a plane for each component in image.  All
        /// components are DCT'd and loaded into the virtual arrays in this pass.
        /// However, it may be that only a subset of the components are emitted to
        /// the entropy encoder during this first pass; be careful about looking
        /// at the scan-dependent variables (MCU dimensions, etc).
        /// </summary>
        private bool compressFirstPass(byte[][][] input_buf)
        {
            int last_iMCU_row = m_cinfo.m_total_iMCU_rows - 1;

            for (int ci = 0; ci < m_cinfo.m_num_components; ci++)
            {
                JpegComponent componentInfo = m_cinfo.Component_info[ci];

                /* Align the virtual buffer for this component. */
                JpegBlock[][] buffer = m_whole_image[ci].Access(m_iMCU_row_num * componentInfo.V_samp_factor,
                    componentInfo.V_samp_factor);

                /* Count non-dummy DCT block rows in this iMCU row. */
                int block_rows;
                if (m_iMCU_row_num < last_iMCU_row)
                {
                    block_rows = componentInfo.V_samp_factor;
                }
                else
                {
                    /* NB: can't use last_row_height here, since may not be set! */
                    block_rows = componentInfo.height_in_blocks % componentInfo.V_samp_factor;
                    if (block_rows == 0)
                        block_rows = componentInfo.V_samp_factor;
                }

                int blocks_across = componentInfo.Width_in_blocks;
                int h_samp_factor = componentInfo.H_samp_factor;

                /* Count number of dummy blocks to be added at the right margin. */
                int ndummy = blocks_across % h_samp_factor;
                if (ndummy > 0)
                    ndummy = h_samp_factor - ndummy;

                /* Perform DCT for all non-dummy blocks in this iMCU row.  Each call
                 * on forward_DCT processes a complete horizontal row of DCT blocks.
                 */
                for (int block_row = 0; block_row < block_rows; block_row++)
                {
                    m_cinfo.m_fdct.forward_DCT(componentInfo.Quant_tbl_no, input_buf[ci],
                        buffer[block_row], block_row * JpegConstants.DCTSize, 0, blocks_across);

                    if (ndummy > 0)
                    {
                        /* Create dummy blocks at the right edge of the image. */
                        Array.Clear(buffer[block_row][blocks_across].data, 0, buffer[block_row][blocks_across].data.Length);

                        short lastDC = buffer[block_row][blocks_across - 1][0];
                        for (int bi = 0; bi < ndummy; bi++)
                            buffer[block_row][blocks_across + bi][0] = lastDC;
                    }
                }

                /* If at end of image, create dummy block rows as needed.
                 * The tricky part here is that within each MCU, we want the DC values
                 * of the dummy blocks to match the last real block's DC value.
                 * This squeezes a few more bytes out of the resulting file...
                 */
                if (m_iMCU_row_num == last_iMCU_row)
                {
                    blocks_across += ndummy;    /* include lower right corner */
                    int MCUs_across = blocks_across / h_samp_factor;
                    for (int block_row = block_rows; block_row < componentInfo.V_samp_factor; block_row++)
                    {
                        for (int i = 0; i < blocks_across; i++)
                            Array.Clear(buffer[block_row][i].data, 0, buffer[block_row][i].data.Length);

                        int thisOffset = 0;
                        int lastOffset = 0;
                        for (int MCUindex = 0; MCUindex < MCUs_across; MCUindex++)
                        {
                            short lastDC = buffer[block_row - 1][lastOffset + h_samp_factor - 1][0];
                            for (int bi = 0; bi < h_samp_factor; bi++)
                                buffer[block_row][thisOffset + bi][0] = lastDC;

                            thisOffset += h_samp_factor; /* advance to next MCU in row */
                            lastOffset += h_samp_factor;
                        }
                    }
                }
            }

            /* NB: compress_output will increment iMCU_row_num if successful.
             * A suspension return will result in redoing all the work above next time.
             */

            /* Emit data to the entropy encoder, sharing code with subsequent passes */
            return compressOutput();
        }

        /// <summary>
        /// Process some data in subsequent passes of a multi-pass case.
        /// We process the equivalent of one fully interleaved MCU row ("iMCU" row)
        /// per call, ie, v_samp_factor block rows for each component in the scan.
        /// The data is obtained from the virtual arrays and fed to the entropy coder.
        /// Returns true if the iMCU row is completed, false if suspended.
        /// </summary>
        private bool compressOutput()
        {
            /* Align the virtual buffers for the components used in this scan.
             */
            JpegBlock[][][] buffer = new JpegBlock[JpegConstants.MaxComponentsInScan][][];
            for (int ci = 0; ci < m_cinfo.m_comps_in_scan; ci++)
            {
                JpegComponent componentInfo = m_cinfo.Component_info[m_cinfo.m_cur_comp_info[ci]];
                buffer[ci] = m_whole_image[componentInfo.Component_index].Access(
                    m_iMCU_row_num * componentInfo.V_samp_factor, componentInfo.V_samp_factor);
            }

            /* Loop to process one whole iMCU row */
            for (int yoffset = m_MCU_vert_offset; yoffset < m_MCU_rows_per_iMCU_row; yoffset++)
            {
                for (int MCU_col_num = m_mcu_ctr; MCU_col_num < m_cinfo.m_MCUs_per_row; MCU_col_num++)
                {
                    /* Construct list of pointers to DCT blocks belonging to this MCU */
                    int blkn = 0;           /* index of current DCT block within MCU */
                    for (int ci = 0; ci < m_cinfo.m_comps_in_scan; ci++)
                    {
                        JpegComponent componentInfo = m_cinfo.Component_info[m_cinfo.m_cur_comp_info[ci]];
                        int start_col = MCU_col_num * componentInfo.MCU_width;
                        for (int yindex = 0; yindex < componentInfo.MCU_height; yindex++)
                        {
                            for (int xindex = 0; xindex < componentInfo.MCU_width; xindex++)
                            {
                                int bufLength = buffer[ci][yindex + yoffset].Length;
                                int start = start_col + xindex;
                                m_MCU_buffer[blkn] = new JpegBlock[bufLength - start];
                                for (int j = start; j < bufLength; j++)
                                    m_MCU_buffer[blkn][j - start] = buffer[ci][yindex + yoffset][j];

                                blkn++;
                            }
                        }
                    }

                    /* Try to write the MCU. */
                    if (!m_cinfo.m_entropy.encode_mcu(m_MCU_buffer))
                    {
                        /* Suspension forced; update state counters and exit */
                        m_MCU_vert_offset = yoffset;
                        m_mcu_ctr = MCU_col_num;
                        return false;
                    }
                }

                /* Completed an MCU row, but perhaps not an iMCU row */
                m_mcu_ctr = 0;
            }

            /* Completed the iMCU row, advance counters for next one */
            m_iMCU_row_num++;
            start_iMCU_row();
            return true;
        }

        // Reset within-iMCU-row counters for a new row
        private void start_iMCU_row()
        {
            /* In an interleaved scan, an MCU row is the same as an iMCU row.
             * In a noninterleaved scan, an iMCU row has v_samp_factor MCU rows.
             * But at the bottom of the image, process only what's left.
             */
            if (m_cinfo.m_comps_in_scan > 1)
            {
                m_MCU_rows_per_iMCU_row = 1;
            }
            else
            {
                if (m_iMCU_row_num < (m_cinfo.m_total_iMCU_rows - 1))
                    m_MCU_rows_per_iMCU_row = m_cinfo.Component_info[m_cinfo.m_cur_comp_info[0]].V_samp_factor;
                else
                    m_MCU_rows_per_iMCU_row = m_cinfo.Component_info[m_cinfo.m_cur_comp_info[0]].last_row_height;
            }

            m_mcu_ctr = 0;
            m_MCU_vert_offset = 0;
        }
    }
    #endregion

    #region ColorConverter
    /// <summary>
    /// ColorSpace conversion
    /// </summary>
    class ColorConverter
    {
        private const int SCALEBITS = 16;  /* speediest right-shift on some machines */
        private const int CBCR_OFFSET = JpegConstants.MediumSampleValue << SCALEBITS;
        private const int ONE_HALF = 1 << (SCALEBITS - 1);

        // We allocate one big table and divide it up into eight parts, instead of
        // doing eight alloc_small requests.  This lets us use a single table base
        // address, which can be held in a register in the inner loops on many
        // machines (more than can hold all eight addresses, anyway).
        private const int R_Y_OFF = 0;           /* offset to R => Y section */
        private const int G_Y_OFF = (1 * (JpegConstants.MaxSampleValue + 1));  /* offset to G => Y section */
        private const int B_Y_OFF = (2 * (JpegConstants.MaxSampleValue + 1));  /* etc. */
        private const int R_CB_OFF = (3 * (JpegConstants.MaxSampleValue + 1));
        private const int G_CB_OFF = (4 * (JpegConstants.MaxSampleValue + 1));
        private const int B_CB_OFF = (5 * (JpegConstants.MaxSampleValue + 1));
        private const int R_CR_OFF = B_CB_OFF;        /* B=>Cb, R=>Cr are the same */
        private const int G_CR_OFF = (6 * (JpegConstants.MaxSampleValue + 1));
        private const int B_CR_OFF = (7 * (JpegConstants.MaxSampleValue + 1));
        private const int TABLE_SIZE = (8 * (JpegConstants.MaxSampleValue + 1));

        private JpegCompressor m_cinfo;

        private bool m_useNullStart;

        private bool m_useCmykYcckConvert;
        private bool m_useGrayscaleConvert;
        private bool m_useNullConvert;
        private bool m_useRgbGrayConvert;
        private bool m_useRgbYccConvert;

        private int[] m_rgb_ycc_tab;     /* => table for RGB to YCbCr conversion */

        public ColorConverter(JpegCompressor cinfo)
        {
            m_cinfo = cinfo;

            /* set start_pass to null method until we find out differently */
            m_useNullStart = true;

            /* Make sure input_components agrees with in_color_space */
            switch (cinfo.m_in_color_space)
            {
                case ColorSpace.Grayscale:
                    if (cinfo.m_input_components != 1)
                        throw new Exception("Bogus input colorspace!");
                    break;

                case ColorSpace.RGB:
                case ColorSpace.YCbCr:
                    if (cinfo.m_input_components != 3)
                        throw new Exception("Bogus input colorspace!");
                    break;

                case ColorSpace.CMYK:
                case ColorSpace.YCCK:
                    if (cinfo.m_input_components != 4)
                        throw new Exception("Bogus input colorspace!");
                    break;

                default:
                    /* JCS_UNKNOWN can be anything */
                    if (cinfo.m_input_components < 1)
                        throw new Exception("Bogus input colorspace!");
                    break;
            }

            /* Check num_components, set conversion method based on requested space */
            clearConvertFlags();
            switch (cinfo.m_jpeg_color_space)
            {
                case ColorSpace.Grayscale:
                    {
                        if (cinfo.m_num_components != 1)
                            throw new Exception("Bogus Jpeg colorspace!");


                        if (cinfo.m_in_color_space == ColorSpace.Grayscale)
                        {
                            m_useGrayscaleConvert = true;
                        }
                        else if (cinfo.m_in_color_space == ColorSpace.RGB)
                        {
                            m_useNullStart = false; // use rgb_ycc_start
                            m_useRgbGrayConvert = true;
                        }
                        else if (cinfo.m_in_color_space == ColorSpace.YCbCr)
                        {
                            m_useGrayscaleConvert = true;
                        }
                        else
                        {
                            throw new Exception("Unsupported color conversion request.");
                        }

                        break;
                    }
                case ColorSpace.RGB:
                    {
                        if (cinfo.m_num_components != 3)
                            throw new Exception("Bogus Jpeg colorspace!");


                        if (cinfo.m_in_color_space == ColorSpace.RGB)
                        {
                            m_useNullConvert = true;
                        }
                        else
                        {
                            throw new Exception("Unsupported color conversion request.");
                        }

                        break;
                    }
                case ColorSpace.YCbCr:
                    {
                        if (cinfo.m_num_components != 3)
                            throw new Exception("Bogus Jpeg colorspace!");


                        if (cinfo.m_in_color_space == ColorSpace.RGB)
                        {
                            m_useNullStart = false; // use rgb_ycc_start
                            m_useRgbYccConvert = true;
                        }
                        else if (cinfo.m_in_color_space == ColorSpace.YCbCr)
                        {
                            m_useNullConvert = true;
                        }
                        else
                        {
                            throw new Exception("Unsupported color conversion request.");
                        }

                        break;
                    }
                case ColorSpace.CMYK:
                    {
                        if (cinfo.m_num_components != 4)
                            throw new Exception("Bogus Jpeg colorspace!");


                        if (cinfo.m_in_color_space == ColorSpace.CMYK)
                        {
                            m_useNullConvert = true;
                        }
                        else
                        {
                            throw new Exception("Unsupported color conversion request.");
                        }

                        break;
                    }
                case ColorSpace.YCCK:
                    {
                        if (cinfo.m_num_components != 4)
                            throw new Exception("Bogus Jpeg colorspace!");


                        if (cinfo.m_in_color_space == ColorSpace.CMYK)
                        {
                            m_useNullStart = false; // use rgb_ycc_start
                            m_useCmykYcckConvert = true;
                        }
                        else if (cinfo.m_in_color_space == ColorSpace.YCCK)
                        {
                            m_useNullConvert = true;
                        }
                        else
                        {
                            throw new Exception("Unsupported color conversion request.");
                        }

                        break;
                    }

                default:
                    /* allow null conversion of JCS_UNKNOWN */
                    if (cinfo.m_jpeg_color_space != cinfo.m_in_color_space || cinfo.m_num_components != cinfo.m_input_components)
                        throw new Exception("Unsupported color conversion request.");

                    m_useNullConvert = true;
                    break;
            }
        }

        public void start_pass()
        {
            if (!m_useNullStart)
                rgb_ycc_start();
        }

        /// <summary>
        /// Convert some rows of samples to the JPEG colorspace.
        /// 
        /// Note that we change from the application's interleaved-pixel format
        /// to our internal non-interleaved, one-plane-per-component format.
        /// The input buffer is therefore three times as wide as the output buffer.
        /// 
        /// A starting row offset is provided only for the output buffer.  The caller
        /// can easily adjust the passed input_buf value to accommodate any row
        /// offset required on that side.
        /// </summary>
        public void color_convert(byte[][] input_buf, int input_row, byte[][][] output_buf, int output_row, int num_rows)
        {
            if (m_useCmykYcckConvert)
                cmyk_ycck_convert(input_buf, input_row, output_buf, output_row, num_rows);
            else if (m_useGrayscaleConvert)
                grayscale_convert(input_buf, input_row, output_buf, output_row, num_rows);
            else if (m_useRgbGrayConvert)
                rgb_gray_convert(input_buf, input_row, output_buf, output_row, num_rows);
            else if (m_useRgbYccConvert)
                rgb_ycc_convert(input_buf, input_row, output_buf, output_row, num_rows);
            else if (m_useNullConvert)
                null_convert(input_buf, input_row, output_buf, output_row, num_rows);
            else
                throw new Exception("Unsupported color conversion request.");
        }

        private void clearConvertFlags()
        {
            m_useCmykYcckConvert = false;
            m_useGrayscaleConvert = false;
            m_useNullConvert = false;
            m_useRgbGrayConvert = false;
            m_useRgbYccConvert = false;
        }

        private static int FIX(double x)
        {
            return (int)(x * (1L << SCALEBITS) + 0.5);
        }

        #region RGB to YCC
        /// <summary>
        /// Initialize for RGB->YCC colorspace conversion.
        /// </summary>
        private void rgb_ycc_start()
        {
            /* Allocate and fill in the conversion tables. */
            m_rgb_ycc_tab = new int[TABLE_SIZE];

            for (int i = 0; i <= JpegConstants.MaxSampleValue; i++)
            {
                m_rgb_ycc_tab[i + R_Y_OFF] = FIX(0.29900) * i;
                m_rgb_ycc_tab[i + G_Y_OFF] = FIX(0.58700) * i;
                m_rgb_ycc_tab[i + B_Y_OFF] = FIX(0.11400) * i + ONE_HALF;
                m_rgb_ycc_tab[i + R_CB_OFF] = (-FIX(0.16874)) * i;
                m_rgb_ycc_tab[i + G_CB_OFF] = (-FIX(0.33126)) * i;

                /* We use a rounding fudge-factor of 0.5-epsilon for Cb and Cr.
                 * This ensures that the maximum output will round to MaxSampleValue
                 * not MaxSampleValue+1, and thus that we don't have to range-limit.
                 */
                m_rgb_ycc_tab[i + B_CB_OFF] = FIX(0.50000) * i + CBCR_OFFSET + ONE_HALF - 1;

                /*  B=>Cb and R=>Cr tables are the same
                    rgb_ycc_tab[i+R_CR_OFF] = FIX(0.50000) * i    + CBCR_OFFSET + ONE_HALF-1;
                */
                m_rgb_ycc_tab[i + G_CR_OFF] = (-FIX(0.41869)) * i;
                m_rgb_ycc_tab[i + B_CR_OFF] = (-FIX(0.08131)) * i;
            }
        }

        /// <summary>
        /// RGB -&gt; YCbCr conversion: most common case
        /// YCbCr is defined per CCIR 601-1, except that Cb and Cr are
        /// normalized to the range 0..MaxSampleValue rather than -0.5 .. 0.5.
        /// The conversion equations to be implemented are therefore
        /// Y  =  0.29900 * R + 0.58700 * G + 0.11400 * B
        /// Cb = -0.16874 * R - 0.33126 * G + 0.50000 * B  + MediumSampleValue
        /// Cr =  0.50000 * R - 0.41869 * G - 0.08131 * B  + MediumSampleValue
        /// (These numbers are derived from TIFF 6.0 section 21, dated 3-June-92.)
        /// To avoid floating-point arithmetic, we represent the fractional constants
        /// as integers scaled up by 2^16 (about 4 digits precision); we have to divide
        /// the products by 2^16, with appropriate rounding, to get the correct answer.
        /// For even more speed, we avoid doing any multiplications in the inner loop
        /// by pre-calculating the constants times R,G,B for all possible values.
        /// For 8-bit JSAMPLEs this is very reasonable (only 256 entries per table);
        /// for 12-bit samples it is still acceptable.  It's not very reasonable for
        /// 16-bit samples, but if you want lossless storage you shouldn't be changing
        /// colorspace anyway.
        /// The MediumSampleValue offsets and the rounding fudge-factor of 0.5 are included
        /// in the tables to save adding them separately in the inner loop.
        /// </summary>
        private void rgb_ycc_convert(byte[][] input_buf, int input_row, byte[][][] output_buf, int output_row, int num_rows)
        {
            int num_cols = m_cinfo.m_image_width;
            for (int row = 0; row < num_rows; row++)
            {
                int columnOffset = 0;
                for (int col = 0; col < num_cols; col++)
                {
                    int r = input_buf[input_row + row][columnOffset + JpegConstants.Offset_RGB_Red];
                    int g = input_buf[input_row + row][columnOffset + JpegConstants.Offset_RGB_Green];
                    int b = input_buf[input_row + row][columnOffset + JpegConstants.Offset_RGB_Blue];
                    columnOffset += JpegConstants.RGB_PixelLength;

                    /* If the inputs are 0..MaxSampleValue, the outputs of these equations
                     * must be too; we do not need an explicit range-limiting operation.
                     * Hence the value being shifted is never negative, and we don't
                     * need the general RIGHT_SHIFT macro.
                     */
                    /* Y */
                    output_buf[0][output_row][col] = (byte)((m_rgb_ycc_tab[r + R_Y_OFF] + m_rgb_ycc_tab[g + G_Y_OFF] + m_rgb_ycc_tab[b + B_Y_OFF]) >> SCALEBITS);
                    /* Cb */
                    output_buf[1][output_row][col] = (byte)((m_rgb_ycc_tab[r + R_CB_OFF] + m_rgb_ycc_tab[g + G_CB_OFF] + m_rgb_ycc_tab[b + B_CB_OFF]) >> SCALEBITS);
                    /* Cr */
                    output_buf[2][output_row][col] = (byte)((m_rgb_ycc_tab[r + R_CR_OFF] + m_rgb_ycc_tab[g + G_CR_OFF] + m_rgb_ycc_tab[b + B_CR_OFF]) >> SCALEBITS);
                }

                output_row++;
            }
        }
        #endregion

        #region RGB to Grayscale
        /// <summary>
        /// Convert some rows of samples to the JPEG colorspace.
        /// This version handles RGB->grayscale conversion, which is the same
        /// as the RGB->Y portion of RGB->YCbCr.
        /// We assume rgb_ycc_start has been called (we only use the Y tables).
        /// </summary>
        private void rgb_gray_convert(byte[][] input_buf, int input_row, byte[][][] output_buf, int output_row, int num_rows)
        {
            int num_cols = m_cinfo.m_image_width;
            for (int row = 0; row < num_rows; row++)
            {
                int columnOffset = 0;
                for (int col = 0; col < num_cols; col++)
                {
                    int r = input_buf[input_row + row][columnOffset + JpegConstants.Offset_RGB_Red];
                    int g = input_buf[input_row + row][columnOffset + JpegConstants.Offset_RGB_Green];
                    int b = input_buf[input_row + row][columnOffset + JpegConstants.Offset_RGB_Blue];
                    columnOffset += JpegConstants.RGB_PixelLength;

                    /* Y */
                    output_buf[0][output_row][col] = (byte)((m_rgb_ycc_tab[r + R_Y_OFF] + m_rgb_ycc_tab[g + G_Y_OFF] + m_rgb_ycc_tab[b + B_Y_OFF]) >> SCALEBITS);
                }

                output_row++;
            }
        }
        #endregion

        #region CMYK to YCCK
        /// <summary>
        /// Convert some rows of samples to the JPEG colorspace.
        /// This version handles Adobe-style CMYK->YCCK conversion,
        /// where we convert R=1-C, G=1-M, and B=1-Y to YCbCr using the same
        /// conversion as above, while passing K (black) unchanged.
        /// We assume rgb_ycc_start has been called.
        /// </summary>
        private void cmyk_ycck_convert(byte[][] input_buf, int input_row, byte[][][] output_buf, int output_row, int num_rows)
        {
            int num_cols = m_cinfo.m_image_width;
            for (int row = 0; row < num_rows; row++)
            {
                int columnOffset = 0;
                for (int col = 0; col < num_cols; col++)
                {
                    int r = JpegConstants.MaxSampleValue - input_buf[input_row + row][columnOffset];
                    int g = JpegConstants.MaxSampleValue - input_buf[input_row + row][columnOffset + 1];
                    int b = JpegConstants.MaxSampleValue - input_buf[input_row + row][columnOffset + 2];

                    /* K passes through as-is */
                    /* don't need GETJSAMPLE here */
                    output_buf[3][output_row][col] = input_buf[input_row + row][columnOffset + 3];
                    columnOffset += 4;

                    /* If the inputs are 0..MaxSampleValue, the outputs of these equations
                     * must be too; we do not need an explicit range-limiting operation.
                     * Hence the value being shifted is never negative, and we don't
                     * need the general RIGHT_SHIFT macro.
                     */
                    /* Y */
                    output_buf[0][output_row][col] = (byte)((m_rgb_ycc_tab[r + R_Y_OFF] + m_rgb_ycc_tab[g + G_Y_OFF] + m_rgb_ycc_tab[b + B_Y_OFF]) >> SCALEBITS);
                    /* Cb */
                    output_buf[1][output_row][col] = (byte)((m_rgb_ycc_tab[r + R_CB_OFF] + m_rgb_ycc_tab[g + G_CB_OFF] + m_rgb_ycc_tab[b + B_CB_OFF]) >> SCALEBITS);
                    /* Cr */
                    output_buf[2][output_row][col] = (byte)((m_rgb_ycc_tab[r + R_CR_OFF] + m_rgb_ycc_tab[g + G_CR_OFF] + m_rgb_ycc_tab[b + B_CR_OFF]) >> SCALEBITS);
                }

                output_row++;
            }
        }
        #endregion

        #region Grayscale Conversion
        /// <summary>
        /// Convert some rows of samples to the JPEG colorspace.
        /// This version handles grayscale output with no conversion.
        /// The source can be either plain grayscale or YCbCr (since Y == gray).
        /// </summary>
        private void grayscale_convert(byte[][] input_buf, int input_row, byte[][][] output_buf, int output_row, int num_rows)
        {
            int num_cols = m_cinfo.m_image_width;
            int instride = m_cinfo.m_input_components;

            for (int row = 0; row < num_rows; row++)
            {
                int columnOffset = 0;
                for (int col = 0; col < num_cols; col++)
                {
                    /* don't need GETJSAMPLE() here */
                    output_buf[0][output_row][col] = input_buf[input_row + row][columnOffset];
                    columnOffset += instride;
                }

                output_row++;
            }
        }
        #endregion

        #region Null Conversion
        /// <summary>
        /// Convert some rows of samples to the JPEG colorspace.
        /// This version handles multi-component colorspaces without conversion.
        /// We assume input_components == num_components.
        /// </summary>
        private void null_convert(byte[][] input_buf, int input_row, byte[][][] output_buf, int output_row, int num_rows)
        {
            int nc = m_cinfo.m_num_components;
            int num_cols = m_cinfo.m_image_width;

            for (int row = 0; row < num_rows; row++)
            {
                /* It seems fastest to make a separate pass for each component. */
                for (int ci = 0; ci < nc; ci++)
                {
                    int columnOffset = 0;
                    for (int col = 0; col < num_cols; col++)
                    {
                        /* don't need GETJSAMPLE() here */
                        output_buf[ci][output_row][col] = input_buf[input_row + row][columnOffset + ci];
                        columnOffset += nc;
                    }
                }

                output_row++;
            }
        }
        #endregion
    }
    #endregion

    #region ColorDeconverter
    /// <summary>
    /// ColorSpace conversion
    /// </summary>
    class ColorDeconverter
    {
        private const int SCALEBITS = 16;  /* speediest right-shift on some machines */
        private const int ONE_HALF = 1 << (SCALEBITS - 1);

        private enum ColorConverter
        {
            grayscale_converter,
            ycc_rgb_converter,
            gray_rgb_converter,
            null_converter,
            ycck_cmyk_converter
        }

        private ColorConverter m_converter;
        private JpegDecompressor m_cinfo;

        private int[] m_perComponentOffsets;

        /* Private state for YCC->RGB conversion */
        private int[] m_Cr_r_tab;      /* => table for Cr to R conversion */
        private int[] m_Cb_b_tab;      /* => table for Cb to B conversion */
        private int[] m_Cr_g_tab;        /* => table for Cr to G conversion */
        private int[] m_Cb_g_tab;        /* => table for Cb to G conversion */

        /// <summary>
        /// Module initialization routine for output colorspace conversion.
        /// </summary>
        public ColorDeconverter(JpegDecompressor cinfo)
        {
            m_cinfo = cinfo;

            /* Make sure num_components agrees with jpeg_color_space */
            switch (cinfo.m_jpeg_color_space)
            {
                case ColorSpace.Grayscale:
                    if (cinfo.m_num_components != 1)
                        throw new Exception("Bogus Jpeg colorspace!");
                    break;

                case ColorSpace.RGB:
                case ColorSpace.YCbCr:
                    if (cinfo.m_num_components != 3)
                        throw new Exception("Bogus Jpeg colorspace!");
                    break;

                case ColorSpace.CMYK:
                case ColorSpace.YCCK:
                    if (cinfo.m_num_components != 4)
                        throw new Exception("Bogus Jpeg colorspace!");
                    break;

                default:
                    /* JCS_UNKNOWN can be anything */
                    if (cinfo.m_num_components < 1)
                        throw new Exception("Bogus Jpeg colorspace!");
                    break;
            }

            /* Set out_color_components and conversion method based on requested space.
            * Also clear the component_needed flags for any unused components,
            * so that earlier pipeline stages can avoid useless computation.
            */

            switch (cinfo.m_out_color_space)
            {
                case ColorSpace.Grayscale:
                    cinfo.m_out_color_components = 1;
                    if (cinfo.m_jpeg_color_space == ColorSpace.Grayscale || cinfo.m_jpeg_color_space == ColorSpace.YCbCr)
                    {
                        m_converter = ColorConverter.grayscale_converter;
                        /* For color->grayscale conversion, only the Y (0) component is needed */
                        for (int ci = 1; ci < cinfo.m_num_components; ci++)
                            cinfo.Comp_info[ci].component_needed = false;
                    }
                    else
                        throw new Exception("Unsupported color conversion request.");
                    break;

                case ColorSpace.RGB:
                    cinfo.m_out_color_components = JpegConstants.RGB_PixelLength;
                    if (cinfo.m_jpeg_color_space == ColorSpace.YCbCr)
                    {
                        m_converter = ColorConverter.ycc_rgb_converter;
                        build_ycc_rgb_table();
                    }
                    else if (cinfo.m_jpeg_color_space == ColorSpace.Grayscale)
                        m_converter = ColorConverter.gray_rgb_converter;
                    else if (cinfo.m_jpeg_color_space == ColorSpace.RGB)
                        m_converter = ColorConverter.null_converter;
                    else
                        throw new Exception("Unsupported color conversion request.");
                    break;

                case ColorSpace.CMYK:
                    cinfo.m_out_color_components = 4;
                    if (cinfo.m_jpeg_color_space == ColorSpace.YCCK)
                    {
                        m_converter = ColorConverter.ycck_cmyk_converter;
                        build_ycc_rgb_table();
                    }
                    else if (cinfo.m_jpeg_color_space == ColorSpace.CMYK)
                        m_converter = ColorConverter.null_converter;
                    else
                        throw new Exception("Unsupported color conversion request.");
                    break;

                default:
                    /* Permit null conversion to same output space */
                    if (cinfo.m_out_color_space == cinfo.m_jpeg_color_space)
                    {
                        cinfo.m_out_color_components = cinfo.m_num_components;
                        m_converter = ColorConverter.null_converter;
                    }
                    else
                    {
                        /* unsupported non-null conversion */
                        throw new Exception("Unsupported color conversion request.");
                    }
                    break;
            }

            if (cinfo.m_quantize_colors)
                cinfo.m_output_components = 1; /* single colormapped output component */
            else
                cinfo.m_output_components = cinfo.m_out_color_components;
        }

        /// <summary>
        /// Convert some rows of samples to the output colorspace.
        /// 
        /// Note that we change from non-interleaved, one-plane-per-component format
        /// to interleaved-pixel format.  The output buffer is therefore three times
        /// as wide as the input buffer.
        /// A starting row offset is provided only for the input buffer.  The caller
        /// can easily adjust the passed output_buf value to accommodate any row
        /// offset required on that side.
        /// </summary>
        public void color_convert(ComponentBuffer[] input_buf, int[] perComponentOffsets, int input_row, byte[][] output_buf, int output_row, int num_rows)
        {
            m_perComponentOffsets = perComponentOffsets;

            switch (m_converter)
            {
                case ColorConverter.grayscale_converter:
                    grayscale_convert(input_buf, input_row, output_buf, output_row, num_rows);
                    break;
                case ColorConverter.ycc_rgb_converter:
                    ycc_rgb_convert(input_buf, input_row, output_buf, output_row, num_rows);
                    break;
                case ColorConverter.gray_rgb_converter:
                    gray_rgb_convert(input_buf, input_row, output_buf, output_row, num_rows);
                    break;
                case ColorConverter.null_converter:
                    null_convert(input_buf, input_row, output_buf, output_row, num_rows);
                    break;
                case ColorConverter.ycck_cmyk_converter:
                    ycck_cmyk_convert(input_buf, input_row, output_buf, output_row, num_rows);
                    break;
                default:
                    throw new Exception("Unsupported color conversion request.");
            }
        }

        private static int FIX(double x)
        {
            return (int)(x * (1L << SCALEBITS) + 0.5);
        }


        #region YCbCr to RGB
        /**************** YCbCr -> RGB conversion: most common case **************/

        /*
         * YCbCr is defined per CCIR 601-1, except that Cb and Cr are
         * normalized to the range 0..MaxSampleValue rather than -0.5 .. 0.5.
         * The conversion equations to be implemented are therefore
         *  R = Y                + 1.40200 * Cr
         *  G = Y - 0.34414 * Cb - 0.71414 * Cr
         *  B = Y + 1.77200 * Cb
         * where Cb and Cr represent the incoming values less MediumSampleValue.
         * (These numbers are derived from TIFF 6.0 section 21, dated 3-June-92.)
         *
         * To avoid floating-point arithmetic, we represent the fractional constants
         * as integers scaled up by 2^16 (about 4 digits precision); we have to divide
         * the products by 2^16, with appropriate rounding, to get the correct answer.
         * Notice that Y, being an integral input, does not contribute any fraction
         * so it need not participate in the rounding.
         *
         * For even more speed, we avoid doing any multiplications in the inner loop
         * by pre-calculating the constants times Cb and Cr for all possible values.
         * For 8-bit JSAMPLEs this is very reasonable (only 256 entries per table);
         * for 12-bit samples it is still acceptable.  It's not very reasonable for
         * 16-bit samples, but if you want lossless storage you shouldn't be changing
         * colorspace anyway.
         * The Cr=>R and Cb=>B values can be rounded to integers in advance; the
         * values for the G calculation are left scaled up, since we must add them
         * together before rounding.
         */

        /// <summary>
        /// Initialize tables for YCC->RGB colorspace conversion.
        /// </summary>
        private void build_ycc_rgb_table()
        {
            m_Cr_r_tab = new int[JpegConstants.MaxSampleValue + 1];
            m_Cb_b_tab = new int[JpegConstants.MaxSampleValue + 1];
            m_Cr_g_tab = new int[JpegConstants.MaxSampleValue + 1];
            m_Cb_g_tab = new int[JpegConstants.MaxSampleValue + 1];

            for (int i = 0, x = -JpegConstants.MediumSampleValue; i <= JpegConstants.MaxSampleValue; i++, x++)
            {
                /* i is the actual input pixel value, in the range 0..MaxSampleValue */
                /* The Cb or Cr value we are thinking of is x = i - MediumSampleValue */
                /* Cr=>R value is nearest int to 1.40200 * x */
                m_Cr_r_tab[i] = JpegUtils.RIGHT_SHIFT(FIX(1.40200) * x + ONE_HALF, SCALEBITS);

                /* Cb=>B value is nearest int to 1.77200 * x */
                m_Cb_b_tab[i] = JpegUtils.RIGHT_SHIFT(FIX(1.77200) * x + ONE_HALF, SCALEBITS);

                /* Cr=>G value is scaled-up -0.71414 * x */
                m_Cr_g_tab[i] = (-FIX(0.71414)) * x;

                /* Cb=>G value is scaled-up -0.34414 * x */
                /* We also add in ONE_HALF so that need not do it in inner loop */
                m_Cb_g_tab[i] = (-FIX(0.34414)) * x + ONE_HALF;
            }
        }

        private void ycc_rgb_convert(ComponentBuffer[] input_buf, int input_row, byte[][] output_buf, int output_row, int num_rows)
        {
            int component0RowOffset = m_perComponentOffsets[0];
            int component1RowOffset = m_perComponentOffsets[1];
            int component2RowOffset = m_perComponentOffsets[2];

            byte[] limit = m_cinfo.m_sample_range_limit;
            int limitOffset = m_cinfo.m_sampleRangeLimitOffset;

            for (int row = 0; row < num_rows; row++)
            {
                int columnOffset = 0;
                for (int col = 0; col < m_cinfo.m_output_width; col++)
                {
                    int y = input_buf[0][input_row + component0RowOffset][col];
                    int cb = input_buf[1][input_row + component1RowOffset][col];
                    int cr = input_buf[2][input_row + component2RowOffset][col];

                    /* Range-limiting is essential due to noise introduced by DCT losses. */
                    output_buf[output_row + row][columnOffset + JpegConstants.Offset_RGB_Red] = limit[limitOffset + y + m_Cr_r_tab[cr]];
                    output_buf[output_row + row][columnOffset + JpegConstants.Offset_RGB_Green] = limit[limitOffset + y + JpegUtils.RIGHT_SHIFT(m_Cb_g_tab[cb] + m_Cr_g_tab[cr], SCALEBITS)];
                    output_buf[output_row + row][columnOffset + JpegConstants.Offset_RGB_Blue] = limit[limitOffset + y + m_Cb_b_tab[cb]];
                    columnOffset += JpegConstants.RGB_PixelLength;
                }

                input_row++;
            }
        }
        #endregion

        #region YCCK to CMYK
        /// <summary>
        /// Adobe-style YCCK->CMYK conversion.
        /// We convert YCbCr to R=1-C, G=1-M, and B=1-Y using the same
        /// conversion as above, while passing K (black) unchanged.
        /// We assume build_ycc_rgb_table has been called.
        /// </summary>
        private void ycck_cmyk_convert(ComponentBuffer[] input_buf, int input_row, byte[][] output_buf, int output_row, int num_rows)
        {
            int component0RowOffset = m_perComponentOffsets[0];
            int component1RowOffset = m_perComponentOffsets[1];
            int component2RowOffset = m_perComponentOffsets[2];
            int component3RowOffset = m_perComponentOffsets[3];

            byte[] limit = m_cinfo.m_sample_range_limit;
            int limitOffset = m_cinfo.m_sampleRangeLimitOffset;

            int num_cols = m_cinfo.m_output_width;
            for (int row = 0; row < num_rows; row++)
            {
                int columnOffset = 0;
                for (int col = 0; col < num_cols; col++)
                {
                    int y = input_buf[0][input_row + component0RowOffset][col];
                    int cb = input_buf[1][input_row + component1RowOffset][col];
                    int cr = input_buf[2][input_row + component2RowOffset][col];

                    /* Range-limiting is essential due to noise introduced by DCT losses. */
                    output_buf[output_row + row][columnOffset] = limit[limitOffset + JpegConstants.MaxSampleValue - (y + m_Cr_r_tab[cr])]; /* red */
                    output_buf[output_row + row][columnOffset + 1] = limit[limitOffset + JpegConstants.MaxSampleValue - (y + JpegUtils.RIGHT_SHIFT(m_Cb_g_tab[cb] + m_Cr_g_tab[cr], SCALEBITS))]; /* green */
                    output_buf[output_row + row][columnOffset + 2] = limit[limitOffset + JpegConstants.MaxSampleValue - (y + m_Cb_b_tab[cb])]; /* blue */

                    /* K passes through unchanged */
                    /* don't need GETJSAMPLE here */
                    output_buf[output_row + row][columnOffset + 3] = input_buf[3][input_row + component3RowOffset][col];
                    columnOffset += 4;
                }

                input_row++;
            }
        }
        #endregion

        #region Grayscale to RGB
        /// <summary>
        /// Convert grayscale to RGB: just duplicate the gray-level three times.
        /// This is provided to support applications that don't want to cope
        /// with grayscale as a separate case.
        /// </summary>
        private void gray_rgb_convert(ComponentBuffer[] input_buf, int input_row, byte[][] output_buf, int output_row, int num_rows)
        {
            int component0RowOffset = m_perComponentOffsets[0];
            int component1RowOffset = m_perComponentOffsets[1];
            int component2RowOffset = m_perComponentOffsets[2];

            int num_cols = m_cinfo.m_output_width;
            for (int row = 0; row < num_rows; row++)
            {
                int columnOffset = 0;
                for (int col = 0; col < num_cols; col++)
                {
                    /* We can dispense with GETJSAMPLE() here */
                    output_buf[output_row + row][columnOffset + JpegConstants.Offset_RGB_Red] = input_buf[0][input_row + component0RowOffset][col];
                    output_buf[output_row + row][columnOffset + JpegConstants.Offset_RGB_Green] = input_buf[0][input_row + component1RowOffset][col];
                    output_buf[output_row + row][columnOffset + JpegConstants.Offset_RGB_Blue] = input_buf[0][input_row + component2RowOffset][col];
                    columnOffset += JpegConstants.RGB_PixelLength;
                }

                input_row++;
            }
        }
        #endregion

        #region Grayscale Conversion
        /// <summary>
        /// Color conversion for grayscale: just copy the data.
        /// This also works for YCbCr -> grayscale conversion, in which
        /// we just copy the Y (luminance) component and ignore chrominance.
        /// </summary>
        private void grayscale_convert(ComponentBuffer[] input_buf, int input_row, byte[][] output_buf, int output_row, int num_rows)
        {
            JpegUtils.jcopy_sample_rows(input_buf[0], input_row + m_perComponentOffsets[0], output_buf, output_row, num_rows, m_cinfo.m_output_width);
        }
        #endregion

        #region Null Conversion
        /// <summary>
        /// Color conversion for no colorspace change: just copy the data,
        /// converting from separate-planes to interleaved representation.
        /// </summary>
        private void null_convert(ComponentBuffer[] input_buf, int input_row, byte[][] output_buf, int output_row, int num_rows)
        {
            for (int row = 0; row < num_rows; row++)
            {
                for (int ci = 0; ci < m_cinfo.m_num_components; ci++)
                {
                    int columnIndex = 0;
                    int componentOffset = 0;
                    int perComponentOffset = m_perComponentOffsets[ci];

                    for (int count = m_cinfo.m_output_width; count > 0; count--)
                    {
                        /* needn't bother with GETJSAMPLE() here */
                        output_buf[output_row + row][ci + componentOffset] = input_buf[ci][input_row + perComponentOffset][columnIndex];
                        componentOffset += m_cinfo.m_num_components;
                        columnIndex++;
                    }
                }

                input_row++;
            }
        }
        #endregion
    }
    #endregion

    #region ColorQuantizer
    /// <summary>
    /// Color quantization or color precision reduction
    /// </summary>
    interface ColorQuantizer
    {
        void start_pass(bool is_pre_scan);

        void color_quantize(byte[][] input_buf, int in_row, byte[][] output_buf, int out_row, int num_rows);

        void finish_pass();
        void new_color_map();
    }
    #endregion

    #region ComponentBuffer
    /// <summary>
    /// Encapsulates buffer of image samples for one color component
    /// When provided with funny indices (see JpegDecompressorMainController for 
    /// explanation of what it is) uses them for non-linear row access.
    /// </summary>
    class ComponentBuffer
    {
        private byte[][] m_buffer;

        // array of funny indices
        private int[] m_funnyIndices;

        // index of "first funny index" (used because some code uses negative 
        // indices when retrieve rows)
        // see for example UpsamplerImpl.h2v2_fancy_upsample
        private int m_funnyOffset;

        public ComponentBuffer()
        {
        }

        public ComponentBuffer(byte[][] buf, int[] funnyIndices, int funnyOffset)
        {
            SetBuffer(buf, funnyIndices, funnyOffset);
        }

        public void SetBuffer(byte[][] buf, int[] funnyIndices, int funnyOffset)
        {
            m_buffer = buf;
            m_funnyIndices = funnyIndices;
            m_funnyOffset = funnyOffset;
        }

        public byte[] this[int i]
        {
            get
            {
                if (m_funnyIndices == null)
                    return m_buffer[i];

                return m_buffer[m_funnyIndices[i + m_funnyOffset]];
            }
        }
    }
    #endregion

    #region CompressionParameters
    /// <summary>
    /// Parameters of compression.
    /// </summary>
    /// <remarks>Being used in <see cref="M:BitMiracle.LibJpeg.JpegImage.WriteJpeg(System.IO.Stream,BitMiracle.LibJpeg.CompressionParameters)"/></remarks>
    public class CompressionParameters
    {
        private int m_quality = 75;
        private int m_smoothingFactor;
        private bool m_simpleProgressive;

        /// <summary>
        /// Initializes a new instance of the <see cref="CompressionParameters"/> class.
        /// </summary>
        public CompressionParameters()
        {
        }

        internal CompressionParameters(CompressionParameters parameters)
        {
            if (parameters == null)
                throw new ArgumentNullException("parameters");

            m_quality = parameters.m_quality;
            m_smoothingFactor = parameters.m_smoothingFactor;
            m_simpleProgressive = parameters.m_simpleProgressive;
        }

        /// <summary>
        /// Determines whether the specified <see cref="System.Object"/> is equal to this instance.
        /// </summary>
        /// <param name="obj">The <see cref="System.Object"/> to compare with this instance.</param>
        /// <returns>
        /// 	<c>true</c> if the specified <see cref="System.Object"/> is equal to this instance; otherwise, <c>false</c>.
        /// </returns>
        public override bool Equals(object obj)
        {
            CompressionParameters parameters = obj as CompressionParameters;
            if (parameters == null)
                return false;

            return (m_quality == parameters.m_quality &&
                    m_smoothingFactor == parameters.m_smoothingFactor &&
                    m_simpleProgressive == parameters.m_simpleProgressive);
        }

        /// <summary>
        /// Returns a hash code for this instance.
        /// </summary>
        /// <returns>
        /// A hash code for this instance, suitable for use in hashing algorithms 
        /// and data structures like a hash table. 
        /// </returns>
        public override int GetHashCode()
        {
            return base.GetHashCode();
        }

        /// <summary>
        /// Gets or sets the quality of JPEG image.
        /// </summary>
        /// <remarks>Default value: 75<br/>
        /// The quality value is expressed on the 0..100 scale.
        /// </remarks>
        /// <value>The quality of JPEG image.</value>
        public int Quality
        {
            get { return m_quality; }
            set { m_quality = value; }
        }

        /// <summary>
        /// Gets or sets the coefficient of image smoothing.
        /// </summary>
        /// <remarks>Default value: 0<br/>
        /// If non-zero, the input image is smoothed; the value should be 1 for
        /// minimal smoothing to 100 for maximum smoothing.
        /// </remarks>
        /// <value>The coefficient of image smoothing.</value>
        public int SmoothingFactor
        {
            get { return m_smoothingFactor; }
            set { m_smoothingFactor = value; }
        }

        /// <summary>
        /// Gets or sets a value indicating whether to write a progressive-JPEG file.
        /// </summary>
        /// <value>
        /// <c>true</c> for writing a progressive-JPEG file; <c>false</c> 
        /// for non-progressive JPEG files.
        /// </value>
        public bool SimpleProgressive
        {
            get { return m_simpleProgressive; }
            set { m_simpleProgressive = value; }
        }
    }
    #endregion

    #region DecompressionParameters
    class DecompressionParameters
    {
        private ColorSpace m_outColorspace = ColorSpace.Unknown;
        private int m_scaleNumerator = 1;
        private int m_scaleDenominator = 1;
        private bool m_bufferedImage;
        private bool m_rawDataOut;
        private DCTMethod m_dctMethod = (DCTMethod)JpegConstants.DefaultDCTMethod;
        private DitherMode m_ditherMode = DitherMode.FloydStein;
        private bool m_doFancyUpsampling = true;
        private bool m_doBlockSmoothing = true;
        private bool m_quantizeColors;
        private bool m_twoPassQuantize = true;
        private int m_desiredNumberOfColors = 256;
        private bool m_enableOnePassQuantizer;
        private bool m_enableExternalQuant;
        private bool m_enableTwoPassQuantizer;
        private int m_traceLevel;

        public int TraceLevel
        {
            get
            {
                return m_traceLevel;
            }
            set
            {
                m_traceLevel = value;
            }
        }

        /* Decompression processing parameters --- these fields must be set before
         * calling jpeg_start_decompress().  Note that jpeg_read_header() initializes
         * them to default values.
         */

        /// <summary>
        /// colorspace for output
        /// </summary>
        public ColorSpace OutColorspace
        {
            get
            {
                return m_outColorspace;
            }
            set
            {
                m_outColorspace = value;
            }
        }

        /// <summary>
        /// fraction by which to scale image
        /// </summary>
        public int ScaleNumerator
        {
            get
            {
                return m_scaleNumerator;
            }
            set
            {
                m_scaleNumerator = value;
            }
        }

        public int ScaleDenominator
        {
            get
            {
                return m_scaleDenominator;
            }
            set
            {
                m_scaleDenominator = value;
            }
        }

        /// <summary>
        /// true=multiple output passes
        /// </summary>
        public bool BufferedImage
        {
            get
            {
                return m_bufferedImage;
            }
            set
            {
                m_bufferedImage = value;
            }
        }

        /// <summary>
        /// true=downsampled data wanted
        /// </summary>
        public bool RawDataOut
        {
            get
            {
                return m_rawDataOut;
            }
            set
            {
                m_rawDataOut = value;
            }
        }

        /// <summary>
        /// IDCT algorithm selector
        /// </summary>
        public DCTMethod DCTMethod
        {
            get
            {
                return m_dctMethod;
            }
            set
            {
                m_dctMethod = value;
            }
        }

        /// <summary>
        /// true=apply fancy upsampling
        /// </summary>
        public bool DoFancyUpsampling
        {
            get
            {
                return m_doFancyUpsampling;
            }
            set
            {
                m_doFancyUpsampling = value;
            }
        }

        /// <summary>
        /// true=apply interblock smoothing
        /// </summary>
        public bool DoBlockSmoothing
        {
            get
            {
                return m_doBlockSmoothing;
            }
            set
            {
                m_doBlockSmoothing = value;
            }
        }

        /// <summary>
        /// true=colormapped output wanted
        /// </summary>
        public bool QuantizeColors
        {
            get
            {
                return m_quantizeColors;
            }
            set
            {
                m_quantizeColors = value;
            }
        }

        /* the following are ignored if not quantize_colors: */

        /// <summary>
        /// type of color dithering to use
        /// </summary>
        public DitherMode DitherMode
        {
            get
            {
                return m_ditherMode;
            }
            set
            {
                m_ditherMode = value;
            }
        }

        /// <summary>
        /// true=use two-pass color quantization
        /// </summary>
        public bool TwoPassQuantize
        {
            get
            {
                return m_twoPassQuantize;
            }
            set
            {
                m_twoPassQuantize = value;
            }
        }

        /// <summary>
        /// max # colors to use in created colormap
        /// </summary>
        public int DesiredNumberOfColors
        {
            get
            {
                return m_desiredNumberOfColors;
            }
            set
            {
                m_desiredNumberOfColors = value;
            }
        }

        /* these are significant only in buffered-image mode: */

        /// <summary>
        /// enable future use of 1-pass quantizer
        /// </summary>
        public bool EnableOnePassQuantizer
        {
            get
            {
                return m_enableOnePassQuantizer;
            }
            set
            {
                m_enableOnePassQuantizer = value;
            }
        }

        /// <summary>
        /// enable future use of external colormap
        /// </summary>
        public bool EnableExternalQuant
        {
            get
            {
                return m_enableExternalQuant;
            }
            set
            {
                m_enableExternalQuant = value;
            }
        }

        /// <summary>
        /// enable future use of 2-pass quantizer
        /// </summary>
        public bool EnableTwoPassQuantizer
        {
            get
            {
                return m_enableTwoPassQuantizer;
            }
            set
            {
                m_enableTwoPassQuantizer = value;
            }
        }
    }
    #endregion

    #region DecompressorToJpegImage
    class DecompressorToJpegImage : IDecompressor
    {
        private JpegImage m_jpegImage;

        internal DecompressorToJpegImage(JpegImage jpegImage)
        {
            m_jpegImage = jpegImage;
        }

        public override Stream Output
        {
            get
            {
                return null;
            }
        }

        public override void SetImageAttributes(LoadedImageAttributes parameters)
        {
            m_jpegImage.Width = parameters.Width;
            m_jpegImage.Height = parameters.Height;
            m_jpegImage.BitsPerComponent = 8;
            m_jpegImage.ComponentsPerSample = (byte)parameters.ComponentsPerSample;
            m_jpegImage.Colorspace = parameters.Colorspace;
        }

        public override void BeginWrite()
        {
        }

        public override void ProcessPixelsRow(byte[] row)
        {
            SampleRow samplesRow = new SampleRow(row, m_jpegImage.Width, m_jpegImage.BitsPerComponent, m_jpegImage.ComponentsPerSample);
            m_jpegImage.addSampleRow(samplesRow);
        }

        public override void EndWrite()
        {
        }
    }
    #endregion

    #region DerivedTable
    /// <summary>
    /// Derived data constructed for each Huffman table
    /// </summary>
    class DerivedTable
    {
        /* Basic tables: (element [0] of each array is unused) */
        public int[] maxcode = new int[18];      /* largest code of length k (-1 if none) */
        /* (maxcode[17] is a sentinel to ensure jpeg_huff_decode terminates) */
        public int[] valoffset = new int[17];        /* huffval[] offset for codes of length k */
        /* valoffset[k] = huffval[] index of 1st symbol of code length k, less
        * the smallest code of length k; so given a code of length k, the
        * corresponding symbol is huffval[code + valoffset[k]]
        */

        /* Link to public Huffman table (needed only in jpeg_huff_decode) */
        public JpegHuffmanTable pub;

        /* Lookahead tables: indexed by the next HuffmanLookaheadDistance bits of
        * the input data stream.  If the next Huffman code is no more
        * than HuffmanLookaheadDistance bits long, we can obtain its length and
        * the corresponding symbol directly from these tables.
        */
        public int[] look_nbits = new int[1 << JpegConstants.HuffmanLookaheadDistance]; /* # bits, or 0 if too long */
        public byte[] look_sym = new byte[1 << JpegConstants.HuffmanLookaheadDistance]; /* symbol, or unused */
    }
    #endregion

    #region DestinationManager
    /// <summary>
    /// Data destination object for compression.
    /// </summary>
    public abstract class DestinationManager
    {
        private byte[] m_buffer;
        private int m_position;
        private int m_free_in_buffer;  /* # of byte spaces remaining in buffer */

        /// <summary>
        /// Initializes this instance.
        /// </summary>
        public abstract void init_destination();

        /// <summary>
        /// Empties output buffer.
        /// </summary>
        /// <returns><c>true</c> if operation succeed; otherwise, <c>false</c></returns>
        public abstract bool empty_output_buffer();

        /// <summary>
        /// Term_destinations this instance.
        /// </summary>
        public abstract void term_destination();

        /// <summary>
        /// Emits a byte.
        /// </summary>
        /// <param name="val">The byte value.</param>
        /// <returns><c>true</c> if operation succeed; otherwise, <c>false</c></returns>
        public virtual bool emit_byte(int val)
        {
            m_buffer[m_position] = (byte)val;
            m_position++;

            if (--m_free_in_buffer == 0)
            {
                if (!empty_output_buffer())
                    return false;
            }

            return true;
        }

        /// <summary>
        /// Initializes the internal buffer.
        /// </summary>
        /// <param name="buffer">The buffer.</param>
        /// <param name="offset">The offset.</param>
        protected void initInternalBuffer(byte[] buffer, int offset)
        {
            m_buffer = buffer;
            m_free_in_buffer = buffer.Length - offset;
            m_position = offset;
        }

        /// <summary>
        /// Gets the number of free bytes in buffer.
        /// </summary>
        /// <value>The number of free bytes in buffer.</value>
        protected int freeInBuffer
        {
            get
            {
                return m_free_in_buffer;
            }
        }
    }
    #endregion

    #region DestinationManagerImpl
    /// <summary>
    /// Expanded data destination object for output to Stream
    /// </summary>
    class DestinationManagerImpl : DestinationManager
    {
        private const int OUTPUT_BUF_SIZE = 4096;   /* choose an efficiently fwrite'able size */

        private JpegCompressor m_cinfo;

        private Stream m_outfile;      /* target stream */
        private byte[] m_buffer;     /* start of buffer */

        public DestinationManagerImpl(JpegCompressor cinfo, Stream alreadyOpenFile)
        {
            m_cinfo = cinfo;
            m_outfile = alreadyOpenFile;
        }

        /// <summary>
        /// Initialize destination --- called by jpeg_start_compress
        /// before any data is actually written.
        /// </summary>
        public override void init_destination()
        {
            /* Allocate the output buffer --- it will be released when done with image */
            m_buffer = new byte[OUTPUT_BUF_SIZE];
            initInternalBuffer(m_buffer, 0);
        }

        /// <summary>
        /// Empty the output buffer --- called whenever buffer fills up.
        /// 
        /// In typical applications, this should write the entire output buffer
        /// (ignoring the current state of next_output_byte and free_in_buffer),
        /// reset the pointer and count to the start of the buffer, and return true
        /// indicating that the buffer has been dumped.
        /// 
        /// In applications that need to be able to suspend compression due to output
        /// overrun, a false return indicates that the buffer cannot be emptied now.
        /// In this situation, the compressor will return to its caller (possibly with
        /// an indication that it has not accepted all the supplied scanlines).  The
        /// application should resume compression after it has made more room in the
        /// output buffer.  Note that there are substantial restrictions on the use of
        /// suspension --- see the documentation.
        /// 
        /// When suspending, the compressor will back up to a convenient restart point
        /// (typically the start of the current MCU). next_output_byte and free_in_buffer
        /// indicate where the restart point will be if the current call returns false.
        /// Data beyond this point will be regenerated after resumption, so do not
        /// write it out when emptying the buffer externally.
        /// </summary>
        public override bool empty_output_buffer()
        {
            writeBuffer(m_buffer.Length);
            initInternalBuffer(m_buffer, 0);
            return true;
        }

        /// <summary>
        /// Terminate destination --- called by jpeg_finish_compress
        /// after all data has been written.  Usually needs to flush buffer.
        /// 
        /// NB: *not* called by jpeg_abort or jpeg_destroy; surrounding
        /// application must deal with any cleanup that should happen even
        /// for error exit.
        /// </summary>
        public override void term_destination()
        {
            int datacount = m_buffer.Length - freeInBuffer;

            /* Write any data remaining in the buffer */
            if (datacount > 0)
                writeBuffer(datacount);

            m_outfile.Flush();
        }

        private void writeBuffer(int dataCount)
        {
            try
            {
                m_outfile.Write(m_buffer, 0, dataCount);
            }
#pragma warning disable 168
            catch (IOException e)
            {
                throw new Exception("Output file write error --- out of disk space?");
            }
            catch (NotSupportedException e)
            {
                throw new Exception("Output file write error --- out of disk space?");
            }
            catch (ObjectDisposedException e)
            {
                throw new Exception("Output file write error --- out of disk space?");
            }
#pragma warning restore 168
        }
    }
    #endregion

    #region Enumerations
    /// <summary>
    /// Operating modes for buffer controllers
    /// </summary>
    enum BufferMode
    {
        /// <summary>
        /// Plain strip-wise operation.
        /// </summary>
        PassThru,
        /* Remaining modes require a full-image buffer to have been created */
        /// <summary>
        /// Run source sub-object only, save output.
        /// </summary>
        SaveSource,
        /// <summary>
        /// Run dest sub-object only, using saved data.
        /// </summary>
        CrankDest,
        /// <summary>
        /// Run both sub-objects, save output
        /// </summary>
        SaveAndPass
    }

    /// <summary>
    /// The unit of density.
    /// </summary>
    /// <seealso cref="JpegCompressor.Density_unit"/>
    /// <seealso cref="JpegDecompressor.Density_unit"/>
    public enum DensityUnit
    {
        /// <summary>
        /// Unknown density
        /// </summary>
        Unknown = 0,
        /// <summary>
        /// Dots/inch
        /// </summary>
        DotsInch = 1,
        /// <summary>
        /// Dots/cm
        /// </summary>
        DotsCm = 2
    }

    /// <summary>
    /// Dithering options for decompression.
    /// </summary>
    /// <seealso cref="JpegDecompressor.Dither_mode"/>
    public enum DitherMode
    {
        /// <summary>
        /// No dithering: fast, very low quality
        /// </summary>
        None,
        /// <summary>
        /// Ordered dither: moderate speed and quality
        /// </summary>
        Ordered,
        /// <summary>
        /// Floyd-Steinberg dither: slow, high quality
        /// </summary>
        FloydStein
    }

    /// <summary>
    /// Describes a result of read operation.
    /// </summary>
    /// <seealso cref="JpegDecompressor.jpeg_consume_input"/>
    public enum ReadResult
    {
        /// <summary>
        /// Suspended due to lack of input data.
        /// Can occur only if a suspending data source is used.
        /// </summary>
        Suspended = 0,
        /// <summary>
        /// Found valid image data-stream.
        /// </summary>
        Header_Ok = 1,
        /// <summary>
        /// Found valid table-specs-only data-stream.
        /// </summary>
        Header_Tables_Only = 2,
        /// <summary>
        /// Reached a SOS marker (the start of a new scan)
        /// </summary>
        Reached_SOS = 3,
        /// <summary>
        /// Reached the EOI marker (end of image)
        /// </summary>
        Reached_EOI = 4,
        /// <summary>
        /// Completed reading one MCU row of compressed data.
        /// </summary>
        Row_Completed = 5,
        /// <summary>
        /// Completed reading last MCU row of current scan.
        /// </summary>
        Scan_Completed = 6
    }

    /// <summary>
    /// Known color spaces. 
    /// </summary>
    public enum ColorSpace
    {
        /// <summary>
        /// Unspecified colorspace
        /// </summary>
        Unknown,
        /// <summary>
        /// Grayscale
        /// </summary>
        Grayscale,
        /// <summary>
        /// RGB
        /// </summary>
        RGB,
        /// <summary>
        /// YCbCr (also known as YUV)
        /// </summary>
        YCbCr,
        /// <summary>
        /// CMYK
        /// </summary>
        CMYK,
        /// <summary>
        /// YCbCrK
        /// </summary>
        YCCK
    }

    /// <summary>
    /// Algorithm used for the DCT step.
    /// </summary>
    /// <remarks>The <c>Float</c> method is very slightly more accurate than the <c>ISLOW</c> method, 
    /// but may give different results on different machines due to varying roundoff behavior. 
    /// The integer methods should give the same results on all machines. On machines with 
    /// sufficiently fast hardware, the floating-point method may also be the fastest. 
    /// The <c>IFAST</c> method is considerably less accurate than the other two; its use is not recommended 
    /// if high quality is a concern.</remarks>
    /// <seealso cref="JpegCompressor.Dct_method"/>
    /// <seealso cref="JpegDecompressor.Dct_method"/>
    public enum DCTMethod
    {
        /// <summary>
        /// Slow but accurate integer algorithm.
        /// </summary>
        IntSlow,
        /// <summary>
        /// Faster, less accurate integer method.
        /// </summary>
        IntFast,
        /// <summary>
        /// Floating-point method.
        /// </summary>
        Float
    }
    #endregion

    #region HuffEntropyDecoder
    /// <summary>
    /// Expanded entropy decoder object for Huffman decoding.
    /// 
    /// The savable_state sub-record contains fields that change within an MCU,
    /// but must not be updated permanently until we complete the MCU.
    /// </summary>
    class HuffEntropyDecoder : JpegEntropyDecoder
    {
        private class savable_state
        {
            public int[] last_dc_val = new int[JpegConstants.MaxComponentsInScan]; /* last DC coef for each component */

            public void Assign(savable_state ss)
            {
                Buffer.BlockCopy(ss.last_dc_val, 0, last_dc_val, 0, last_dc_val.Length * sizeof(int));
            }
        }

        /* These fields are loaded into local variables at start of each MCU.
        * In case of suspension, we exit WITHOUT updating them.
        */
        private SavedBitreadState m_bitstate;    /* Bit buffer at start of MCU */
        private savable_state m_saved = new savable_state();        /* Other state at start of MCU */

        /* These fields are NOT loaded into local working state. */
        private int m_restarts_to_go;    /* MCUs left in this restart interval */

        /* Pointers to derived tables (these workspaces have image lifespan) */
        private DerivedTable[] m_dc_derived_tbls = new DerivedTable[JpegConstants.NumberOfHuffmanTables];
        private DerivedTable[] m_ac_derived_tbls = new DerivedTable[JpegConstants.NumberOfHuffmanTables];

        /* Precalculated info set up by start_pass for use in decode_mcu: */

        /* Pointers to derived tables to be used for each block within an MCU */
        private DerivedTable[] m_dc_cur_tbls = new DerivedTable[JpegConstants.DecompressorMaxBlocksInMCU];
        private DerivedTable[] m_ac_cur_tbls = new DerivedTable[JpegConstants.DecompressorMaxBlocksInMCU];

        /* Whether we care about the DC and AC coefficient values for each block */
        private bool[] m_dc_needed = new bool[JpegConstants.DecompressorMaxBlocksInMCU];
        private bool[] m_ac_needed = new bool[JpegConstants.DecompressorMaxBlocksInMCU];

        public HuffEntropyDecoder(JpegDecompressor cinfo)
        {
            m_cinfo = cinfo;

            /* Mark tables unallocated */
            for (int i = 0; i < JpegConstants.NumberOfHuffmanTables; i++)
                m_dc_derived_tbls[i] = m_ac_derived_tbls[i] = null;
        }

        /// <summary>
        /// Initialize for a Huffman-compressed scan.
        /// </summary>
        public override void start_pass()
        {
            for (int ci = 0; ci < m_cinfo.m_comps_in_scan; ci++)
            {
                JpegComponent componentInfo = m_cinfo.Comp_info[m_cinfo.m_cur_comp_info[ci]];
                int dctbl = componentInfo.Dc_tbl_no;
                int actbl = componentInfo.Ac_tbl_no;

                /* Compute derived values for Huffman tables */
                /* We may do this more than once for a table, but it's not expensive */
                jpeg_make_d_derived_tbl(true, dctbl, ref m_dc_derived_tbls[dctbl]);
                jpeg_make_d_derived_tbl(false, actbl, ref m_ac_derived_tbls[actbl]);

                /* Initialize DC predictions to 0 */
                m_saved.last_dc_val[ci] = 0;
            }

            /* Precalculate decoding info for each block in an MCU of this scan */
            for (int blkn = 0; blkn < m_cinfo.m_blocks_in_MCU; blkn++)
            {
                int ci = m_cinfo.m_MCU_membership[blkn];
                JpegComponent componentInfo = m_cinfo.Comp_info[m_cinfo.m_cur_comp_info[ci]];

                /* Precalculate which table to use for each block */
                m_dc_cur_tbls[blkn] = m_dc_derived_tbls[componentInfo.Dc_tbl_no];
                m_ac_cur_tbls[blkn] = m_ac_derived_tbls[componentInfo.Ac_tbl_no];

                /* Decide whether we really care about the coefficient values */
                if (componentInfo.component_needed)
                {
                    m_dc_needed[blkn] = true;
                    /* we don't need the ACs if producing a 1/8th-size image */
                    m_ac_needed[blkn] = (componentInfo.DCT_scaled_size > 1);
                }
                else
                {
                    m_dc_needed[blkn] = m_ac_needed[blkn] = false;
                }
            }

            /* Initialize bitread state variables */
            m_bitstate.bits_left = 0;
            m_bitstate.get_buffer = 0;
            m_insufficient_data = false;

            /* Initialize restart counter */
            m_restarts_to_go = m_cinfo.m_restart_interval;
        }

        /// <summary>
        /// Decode and return one MCU's worth of Huffman-compressed coefficients.
        /// The coefficients are reordered from zigzag order into natural array order,
        /// but are not dequantized.
        /// 
        /// The i'th block of the MCU is stored into the block pointed to by
        /// MCU_data[i].  WE ASSUME THIS AREA HAS BEEN ZEROED BY THE CALLER.
        /// (Wholesale zeroing is usually a little faster than retail...)
        /// 
        /// Returns false if data source requested suspension.  In that case no
        /// changes have been made to permanent state.  (Exception: some output
        /// coefficients may already have been assigned.  This is harmless for
        /// this module, since we'll just re-assign them on the next call.)
        /// </summary>
        public override bool decode_mcu(JpegBlock[] MCU_data)
        {
            /* Process restart marker if needed; may have to suspend */
            if (m_cinfo.m_restart_interval != 0)
            {
                if (m_restarts_to_go == 0)
                {
                    if (!process_restart())
                        return false;
                }
            }

            /* If we've run out of data, just leave the MCU set to zeroes.
             * This way, we return uniform gray for the remainder of the segment.
             */
            if (!m_insufficient_data)
            {
                /* Load up working state */
                int get_buffer;
                int bits_left;
                WorkingBitreadState br_state = new WorkingBitreadState();
                BITREAD_LOAD_STATE(m_bitstate, out get_buffer, out bits_left, ref br_state);
                savable_state state = new savable_state();
                state.Assign(m_saved);

                /* Outer loop handles each block in the MCU */

                for (int blkn = 0; blkn < m_cinfo.m_blocks_in_MCU; blkn++)
                {
                    /* Decode a single block's worth of coefficients */

                    /* Section F.2.2.1: decode the DC coefficient difference */
                    int s;
                    if (!HUFF_DECODE(out s, ref br_state, m_dc_cur_tbls[blkn], ref get_buffer, ref bits_left))
                        return false;

                    if (s != 0)
                    {
                        if (!CHECK_BIT_BUFFER(ref br_state, s, ref get_buffer, ref bits_left))
                            return false;

                        int r = GET_BITS(s, get_buffer, ref bits_left);
                        s = HUFF_EXTEND(r, s);
                    }

                    if (m_dc_needed[blkn])
                    {
                        /* Convert DC difference to actual value, update last_dc_val */
                        int ci = m_cinfo.m_MCU_membership[blkn];
                        s += state.last_dc_val[ci];
                        state.last_dc_val[ci] = s;

                        /* Output the DC coefficient (assumes jpeg_natural_order[0] = 0) */
                        MCU_data[blkn][0] = (short)s;
                    }

                    if (m_ac_needed[blkn])
                    {
                        /* Section F.2.2.2: decode the AC coefficients */
                        /* Since zeroes are skipped, output area must be cleared beforehand */
                        for (int k = 1; k < JpegConstants.DCTSize2; k++)
                        {
                            if (!HUFF_DECODE(out s, ref br_state, m_ac_cur_tbls[blkn], ref get_buffer, ref bits_left))
                                return false;

                            int r = s >> 4;
                            s &= 15;

                            if (s != 0)
                            {
                                k += r;
                                if (!CHECK_BIT_BUFFER(ref br_state, s, ref get_buffer, ref bits_left))
                                    return false;
                                r = GET_BITS(s, get_buffer, ref bits_left);
                                s = HUFF_EXTEND(r, s);

                                /* Output coefficient in natural (dezigzagged) order.
                                   * Note: the extra entries in jpeg_natural_order[] will save us
                                   * if k >= DCTSize2, which could happen if the data is corrupted.
                                   */
                                MCU_data[blkn][JpegUtils.jpeg_natural_order[k]] = (short)s;
                            }
                            else
                            {
                                if (r != 15)
                                    break;

                                k += 15;
                            }
                        }
                    }
                    else
                    {
                        /* Section F.2.2.2: decode the AC coefficients */
                        /* In this path we just discard the values */
                        for (int k = 1; k < JpegConstants.DCTSize2; k++)
                        {
                            if (!HUFF_DECODE(out s, ref br_state, m_ac_cur_tbls[blkn], ref get_buffer, ref bits_left))
                                return false;

                            int r = s >> 4;
                            s &= 15;

                            if (s != 0)
                            {
                                k += r;
                                if (!CHECK_BIT_BUFFER(ref br_state, s, ref get_buffer, ref bits_left))
                                    return false;

                                DROP_BITS(s, ref bits_left);
                            }
                            else
                            {
                                if (r != 15)
                                    break;

                                k += 15;
                            }
                        }
                    }
                }

                /* Completed MCU, so update state */
                BITREAD_SAVE_STATE(ref m_bitstate, get_buffer, bits_left);
                m_saved.Assign(state);
            }

            /* Account for restart interval (no-op if not using restarts) */
            m_restarts_to_go--;

            return true;

        }

        /// <summary>
        /// Check for a restart marker and resynchronize decoder.
        /// Returns false if must suspend.
        /// </summary>
        private bool process_restart()
        {
            /* Throw away any unused bits remaining in bit buffer; */
            /* include any full bytes in next_marker's count of discarded bytes */
            m_cinfo.m_marker.SkipBytes(m_bitstate.bits_left / 8);
            m_bitstate.bits_left = 0;

            /* Advance past the RSTn marker */
            if (!m_cinfo.m_marker.read_restart_marker())
                return false;

            /* Re-initialize DC predictions to 0 */
            for (int ci = 0; ci < m_cinfo.m_comps_in_scan; ci++)
                m_saved.last_dc_val[ci] = 0;

            /* Reset restart counter */
            m_restarts_to_go = m_cinfo.m_restart_interval;

            /* Reset out-of-data flag, unless read_restart_marker left us smack up
             * against a marker.  In that case we will end up treating the next data
             * segment as empty, and we can avoid producing bogus output pixels by
             * leaving the flag set.
             */
            if (m_cinfo.m_unread_marker == 0)
                m_insufficient_data = false;

            return true;
        }
    }
    #endregion

    #region HuffEntropyEncoder
    /// <summary>
    /// Expanded entropy encoder object for Huffman encoding.
    /// </summary>
    class HuffEntropyEncoder : JpegEntropyEncoder
    {
        /* The savable_state sub-record contains fields that change within an MCU,
        * but must not be updated permanently until we complete the MCU.
        */
        private class savable_state
        {
            public int put_buffer;       /* current bit-accumulation buffer */
            public int put_bits;           /* # of bits now in it */
            public int[] last_dc_val = new int[JpegConstants.MaxComponentsInScan]; /* last DC coef for each component */
        }

        private bool m_gather_statistics;

        private savable_state m_saved = new savable_state();        /* Bit buffer & DC state at start of MCU */

        /* These fields are NOT loaded into local working state. */
        private int m_restarts_to_go;    /* MCUs left in this restart interval */
        private int m_next_restart_num;       /* next restart number to write (0-7) */

        /* Pointers to derived tables (these workspaces have image lifespan) */
        private c_derived_tbl[] m_dc_derived_tbls = new c_derived_tbl[JpegConstants.NumberOfHuffmanTables];
        private c_derived_tbl[] m_ac_derived_tbls = new c_derived_tbl[JpegConstants.NumberOfHuffmanTables];

        /* Statistics tables for optimization */
        private long[][] m_dc_count_ptrs = new long[JpegConstants.NumberOfHuffmanTables][];
        private long[][] m_ac_count_ptrs = new long[JpegConstants.NumberOfHuffmanTables][];

        public HuffEntropyEncoder(JpegCompressor cinfo)
        {
            m_cinfo = cinfo;

            /* Mark tables unallocated */
            for (int i = 0; i < JpegConstants.NumberOfHuffmanTables; i++)
            {
                m_dc_derived_tbls[i] = m_ac_derived_tbls[i] = null;
                m_dc_count_ptrs[i] = m_ac_count_ptrs[i] = null;
            }
        }

        /// <summary>
        /// Initialize for a Huffman-compressed scan.
        /// If gather_statistics is true, we do not output anything during the scan,
        /// just count the Huffman symbols used and generate Huffman code tables.
        /// </summary>
        public override void start_pass(bool gather_statistics)
        {
            m_gather_statistics = gather_statistics;

            for (int ci = 0; ci < m_cinfo.m_comps_in_scan; ci++)
            {
                int dctbl = m_cinfo.Component_info[m_cinfo.m_cur_comp_info[ci]].Dc_tbl_no;
                int actbl = m_cinfo.Component_info[m_cinfo.m_cur_comp_info[ci]].Ac_tbl_no;
                if (m_gather_statistics)
                {
                    /* Check for invalid table indexes */
                    /* (make_c_derived_tbl does this in the other path) */
                    if (dctbl < 0 || dctbl >= JpegConstants.NumberOfHuffmanTables)
                        throw new Exception(String.Format("Huffman table 0x{0:X2} was not defined", dctbl));

                    if (actbl < 0 || actbl >= JpegConstants.NumberOfHuffmanTables)
                        throw new Exception(String.Format("Huffman table 0x{0:X2} was not defined", actbl));

                    /* Allocate and zero the statistics tables */
                    /* Note that jpeg_gen_optimal_table expects 257 entries in each table! */
                    if (m_dc_count_ptrs[dctbl] == null)
                        m_dc_count_ptrs[dctbl] = new long[257];

                    Array.Clear(m_dc_count_ptrs[dctbl], 0, m_dc_count_ptrs[dctbl].Length);

                    if (m_ac_count_ptrs[actbl] == null)
                        m_ac_count_ptrs[actbl] = new long[257];

                    Array.Clear(m_ac_count_ptrs[actbl], 0, m_ac_count_ptrs[actbl].Length);
                }
                else
                {
                    /* Compute derived values for Huffman tables */
                    /* We may do this more than once for a table, but it's not expensive */
                    jpeg_make_c_derived_tbl(true, dctbl, ref m_dc_derived_tbls[dctbl]);
                    jpeg_make_c_derived_tbl(false, actbl, ref m_ac_derived_tbls[actbl]);
                }

                /* Initialize DC predictions to 0 */
                m_saved.last_dc_val[ci] = 0;
            }

            /* Initialize bit buffer to empty */
            m_saved.put_buffer = 0;
            m_saved.put_bits = 0;

            /* Initialize restart stuff */
            m_restarts_to_go = m_cinfo.m_restart_interval;
            m_next_restart_num = 0;
        }

        public override bool encode_mcu(JpegBlock[][] MCU_data)
        {
            if (m_gather_statistics)
                return encode_mcu_gather(MCU_data);

            return encode_mcu_huff(MCU_data);
        }

        public override void finish_pass()
        {
            if (m_gather_statistics)
                finish_pass_gather();
            else
                finish_pass_huff();
        }

        /// <summary>
        /// Encode and output one MCU's worth of Huffman-compressed coefficients.
        /// </summary>
        private bool encode_mcu_huff(JpegBlock[][] MCU_data)
        {
            /* Load up working state */
            savable_state state;
            state = m_saved;

            /* Emit restart marker if needed */
            if (m_cinfo.m_restart_interval != 0)
            {
                if (m_restarts_to_go == 0)
                {
                    if (!emit_restart(state, m_next_restart_num))
                        return false;
                }
            }

            /* Encode the MCU data blocks */
            for (int blkn = 0; blkn < m_cinfo.m_blocks_in_MCU; blkn++)
            {
                int ci = m_cinfo.m_MCU_membership[blkn];
                if (!encode_one_block(state, MCU_data[blkn][0].data, state.last_dc_val[ci],
                    m_dc_derived_tbls[m_cinfo.Component_info[m_cinfo.m_cur_comp_info[ci]].Dc_tbl_no],
                    m_ac_derived_tbls[m_cinfo.Component_info[m_cinfo.m_cur_comp_info[ci]].Ac_tbl_no]))
                {
                    return false;
                }

                /* Update last_dc_val */
                state.last_dc_val[ci] = MCU_data[blkn][0][0];
            }

            /* Completed MCU, so update state */
            m_saved = state;

            /* Update restart-interval state too */
            if (m_cinfo.m_restart_interval != 0)
            {
                if (m_restarts_to_go == 0)
                {
                    m_restarts_to_go = m_cinfo.m_restart_interval;
                    m_next_restart_num++;
                    m_next_restart_num &= 7;
                }

                m_restarts_to_go--;
            }

            return true;
        }

        /// <summary>
        /// Finish up at the end of a Huffman-compressed scan.
        /// </summary>
        private void finish_pass_huff()
        {
            /* Load up working state ... flush_bits needs it */
            savable_state state;
            state = m_saved;

            /* Flush out the last data */
            if (!flush_bits(state))
                throw new Exception("Suspension not allowed here!");

            /* Update state */
            m_saved = state;
        }

        /// <summary>
        /// Trial-encode one MCU's worth of Huffman-compressed coefficients.
        /// No data is actually output, so no suspension return is possible.
        /// </summary>
        private bool encode_mcu_gather(JpegBlock[][] MCU_data)
        {
            /* Take care of restart intervals if needed */
            if (m_cinfo.m_restart_interval != 0)
            {
                if (m_restarts_to_go == 0)
                {
                    /* Re-initialize DC predictions to 0 */
                    for (int ci = 0; ci < m_cinfo.m_comps_in_scan; ci++)
                        m_saved.last_dc_val[ci] = 0;

                    /* Update restart state */
                    m_restarts_to_go = m_cinfo.m_restart_interval;
                }

                m_restarts_to_go--;
            }

            for (int blkn = 0; blkn < m_cinfo.m_blocks_in_MCU; blkn++)
            {
                int ci = m_cinfo.m_MCU_membership[blkn];
                htest_one_block(MCU_data[blkn][0].data, m_saved.last_dc_val[ci],
                    m_dc_count_ptrs[m_cinfo.Component_info[m_cinfo.m_cur_comp_info[ci]].Dc_tbl_no],
                    m_ac_count_ptrs[m_cinfo.Component_info[m_cinfo.m_cur_comp_info[ci]].Ac_tbl_no]);
                m_saved.last_dc_val[ci] = MCU_data[blkn][0][0];
            }

            return true;
        }

        /// <summary>
        /// Finish up a statistics-gathering pass and create the new Huffman tables.
        /// </summary>
        private void finish_pass_gather()
        {
            /* It's important not to apply jpeg_gen_optimal_table more than once
             * per table, because it clobbers the input frequency counts!
             */
            bool[] did_dc = new bool[JpegConstants.NumberOfHuffmanTables];
            bool[] did_ac = new bool[JpegConstants.NumberOfHuffmanTables];

            for (int ci = 0; ci < m_cinfo.m_comps_in_scan; ci++)
            {
                int dctbl = m_cinfo.Component_info[m_cinfo.m_cur_comp_info[ci]].Dc_tbl_no;
                if (!did_dc[dctbl])
                {
                    if (m_cinfo.m_dc_huff_tbl_ptrs[dctbl] == null)
                        m_cinfo.m_dc_huff_tbl_ptrs[dctbl] = new JpegHuffmanTable();

                    jpeg_gen_optimal_table(m_cinfo.m_dc_huff_tbl_ptrs[dctbl], m_dc_count_ptrs[dctbl]);
                    did_dc[dctbl] = true;
                }

                int actbl = m_cinfo.Component_info[m_cinfo.m_cur_comp_info[ci]].Ac_tbl_no;
                if (!did_ac[actbl])
                {
                    if (m_cinfo.m_ac_huff_tbl_ptrs[actbl] == null)
                        m_cinfo.m_ac_huff_tbl_ptrs[actbl] = new JpegHuffmanTable();

                    jpeg_gen_optimal_table(m_cinfo.m_ac_huff_tbl_ptrs[actbl], m_ac_count_ptrs[actbl]);
                    did_ac[actbl] = true;
                }
            }
        }

        /// <summary>
        /// Encode a single block's worth of coefficients
        /// </summary>
        private bool encode_one_block(savable_state state, short[] block, int last_dc_val, c_derived_tbl dctbl, c_derived_tbl actbl)
        {
            /* Encode the DC coefficient difference per section F.1.2.1 */
            int temp = block[0] - last_dc_val;
            int temp2 = temp;
            if (temp < 0)
            {
                temp = -temp;       /* temp is abs value of input */
                /* For a negative input, want temp2 = bitwise complement of abs(input) */
                /* This code assumes we are on a two's complement machine */
                temp2--;
            }

            /* Find the number of bits needed for the magnitude of the coefficient */
            int nbits = 0;
            while (temp != 0)
            {
                nbits++;
                temp >>= 1;
            }

            /* Check for out-of-range coefficient values.
             * Since we're encoding a difference, the range limit is twice as much.
             */
            if (nbits > MAX_HUFFMAN_COEF_BITS + 1)
                throw new Exception("DCT coefficient is out of range!");

            /* Emit the Huffman-coded symbol for the number of bits */
            if (!emit_bits(state, dctbl.ehufco[nbits], dctbl.ehufsi[nbits]))
                return false;

            /* Emit that number of bits of the value, if positive, */
            /* or the complement of its magnitude, if negative. */
            if (nbits != 0)
            {
                /* emit_bits rejects calls with size 0 */
                if (!emit_bits(state, temp2, nbits))
                    return false;
            }

            /* Encode the AC coefficients per section F.1.2.2 */
            int r = 0;          /* r = run length of zeros */
            for (int k = 1; k < JpegConstants.DCTSize2; k++)
            {
                temp = block[JpegUtils.jpeg_natural_order[k]];
                if (temp == 0)
                {
                    r++;
                }
                else
                {
                    /* if run length > 15, must emit special run-length-16 codes (0xF0) */
                    while (r > 15)
                    {
                        if (!emit_bits(state, actbl.ehufco[0xF0], actbl.ehufsi[0xF0]))
                            return false;
                        r -= 16;
                    }

                    temp2 = temp;
                    if (temp < 0)
                    {
                        temp = -temp;       /* temp is abs value of input */
                        /* This code assumes we are on a two's complement machine */
                        temp2--;
                    }

                    /* Find the number of bits needed for the magnitude of the coefficient */
                    nbits = 1;      /* there must be at least one 1 bit */
                    while ((temp >>= 1) != 0)
                        nbits++;

                    /* Check for out-of-range coefficient values */
                    if (nbits > MAX_HUFFMAN_COEF_BITS)
                        throw new Exception("DCT coefficient is out of range!");

                    /* Emit Huffman symbol for run length / number of bits */
                    int i = (r << 4) + nbits;
                    if (!emit_bits(state, actbl.ehufco[i], actbl.ehufsi[i]))
                        return false;

                    /* Emit that number of bits of the value, if positive, */
                    /* or the complement of its magnitude, if negative. */
                    if (!emit_bits(state, temp2, nbits))
                        return false;

                    r = 0;
                }
            }

            /* If the last coef(s) were zero, emit an end-of-block code */
            if (r > 0)
            {
                if (!emit_bits(state, actbl.ehufco[0], actbl.ehufsi[0]))
                    return false;
            }

            return true;
        }

        /// <summary>
        /// Huffman coding optimization.
        /// 
        /// We first scan the supplied data and count the number of uses of each symbol
        /// that is to be Huffman-coded. (This process MUST agree with the code above.)
        /// Then we build a Huffman coding tree for the observed counts.
        /// Symbols which are not needed at all for the particular image are not
        /// assigned any code, which saves space in the DHT marker as well as in
        /// the compressed data.
        /// </summary>
        private void htest_one_block(short[] block, int last_dc_val, long[] dc_counts, long[] ac_counts)
        {
            /* Encode the DC coefficient difference per section F.1.2.1 */
            int temp = block[0] - last_dc_val;
            if (temp < 0)
                temp = -temp;

            /* Find the number of bits needed for the magnitude of the coefficient */
            int nbits = 0;
            while (temp != 0)
            {
                nbits++;
                temp >>= 1;
            }

            /* Check for out-of-range coefficient values.
             * Since we're encoding a difference, the range limit is twice as much.
             */
            if (nbits > MAX_HUFFMAN_COEF_BITS + 1)
                throw new Exception("DCT coefficient is out of range!");

            /* Count the Huffman symbol for the number of bits */
            dc_counts[nbits]++;

            /* Encode the AC coefficients per section F.1.2.2 */
            int r = 0;          /* r = run length of zeros */
            for (int k = 1; k < JpegConstants.DCTSize2; k++)
            {
                temp = block[JpegUtils.jpeg_natural_order[k]];
                if (temp == 0)
                {
                    r++;
                }
                else
                {
                    /* if run length > 15, must emit special run-length-16 codes (0xF0) */
                    while (r > 15)
                    {
                        ac_counts[0xF0]++;
                        r -= 16;
                    }

                    /* Find the number of bits needed for the magnitude of the coefficient */
                    if (temp < 0)
                        temp = -temp;

                    /* Find the number of bits needed for the magnitude of the coefficient */
                    nbits = 1;      /* there must be at least one 1 bit */
                    while ((temp >>= 1) != 0)
                        nbits++;

                    /* Check for out-of-range coefficient values */
                    if (nbits > MAX_HUFFMAN_COEF_BITS)
                        throw new Exception("DCT coefficient is out of range!");

                    /* Count Huffman symbol for run length / number of bits */
                    ac_counts[(r << 4) + nbits]++;

                    r = 0;
                }
            }

            /* If the last coef(s) were zero, emit an end-of-block code */
            if (r > 0)
                ac_counts[0]++;
        }

        private bool emit_byte(int val)
        {
            return m_cinfo.m_dest.emit_byte(val);
        }

        /// <summary>
        /// Only the right 24 bits of put_buffer are used; the valid bits are
        /// left-justified in this part.  At most 16 bits can be passed to emit_bits
        /// in one call, and we never retain more than 7 bits in put_buffer
        /// between calls, so 24 bits are sufficient.
        /// </summary>
        private bool emit_bits(savable_state state, int code, int size)
        {
            // Emit some bits; return true if successful, false if must suspend
            /* This routine is heavily used, so it's worth coding tightly. */
            int put_buffer = code;
            int put_bits = state.put_bits;

            /* if size is 0, caller used an invalid Huffman table entry */
            if (size == 0)
                throw new Exception("Missing Huffman code table entry");

            put_buffer &= (1 << size) - 1; /* mask off any extra bits in code */
            put_bits += size;       /* new number of bits in buffer */
            put_buffer <<= 24 - put_bits; /* align incoming bits */
            put_buffer |= state.put_buffer; /* and merge with old buffer contents */

            while (put_bits >= 8)
            {
                int c = (put_buffer >> 16) & 0xFF;
                if (!emit_byte(c))
                    return false;

                if (c == 0xFF)
                {
                    /* need to stuff a zero byte? */
                    if (!emit_byte(0))
                        return false;
                }

                put_buffer <<= 8;
                put_bits -= 8;
            }

            state.put_buffer = put_buffer; /* update state variables */
            state.put_bits = put_bits;

            return true;
        }

        private bool flush_bits(savable_state state)
        {
            if (!emit_bits(state, 0x7F, 7)) /* fill any partial byte with ones */
                return false;

            state.put_buffer = 0;  /* and reset bit-buffer to empty */
            state.put_bits = 0;
            return true;
        }

        /// <summary>
        /// Emit a restart marker and resynchronize predictions.
        /// </summary>
        private bool emit_restart(savable_state state, int restart_num)
        {
            if (!flush_bits(state))
                return false;

            if (!emit_byte(0xFF))
                return false;

            if (!emit_byte((int)(JpegMarkerType.RST0 + restart_num)))
                return false;

            /* Re-initialize DC predictions to 0 */
            for (int ci = 0; ci < m_cinfo.m_comps_in_scan; ci++)
                state.last_dc_val[ci] = 0;

            /* The restart counter is not updated until we successfully write the MCU. */
            return true;
        }
    }
    #endregion

    #region IDecompressDestination
    /// <summary>
    /// Common interface for processing of decompression.
    /// </summary>
    abstract class IDecompressor
    {
        /// <summary>
        /// Stream with decompressed data
        /// </summary>
        public abstract Stream Output { get; }

        /// <summary>
        /// Implementor of this interface should process image properties received from decompressor.
        /// </summary>
        /// <param name="parameters">Image properties</param>
        public abstract void SetImageAttributes(LoadedImageAttributes parameters);

        /// <summary>
        /// Called before decompression
        /// </summary>
        public abstract void BeginWrite();

        /// <summary>
        /// It called during decompression - pass row of pixels from JPEG
        /// </summary>
        /// <param name="row"></param>
        public abstract void ProcessPixelsRow(byte[] row);

        /// <summary>
        /// Called after decompression
        /// </summary>
        public abstract void EndWrite();
    }

    /// <summary>
    /// Holds parameters of image for decompression (IDecomressDesination)
    /// </summary>
    class LoadedImageAttributes
    {
        private ColorSpace m_colorspace;
        private bool m_quantizeColors;
        private int m_width;
        private int m_height;
        private int m_componentsPerSample;
        private int m_components;
        private int m_actualNumberOfColors;
        private byte[][] m_colormap;
        private DensityUnit m_densityUnit;
        private int m_densityX;
        private int m_densityY;

        /* Decompression processing parameters --- these fields must be set before
         * calling jpeg_start_decompress().  Note that jpeg_read_header() initializes
         * them to default values.
         */

        /// <summary>
        /// colorspace for output
        /// </summary>
        public ColorSpace Colorspace
        {
            get
            {
                return m_colorspace;
            }
            internal set
            {
                m_colorspace = value;
            }
        }

        /// <summary>
        /// true=colormapped output wanted
        /// </summary>
        public bool QuantizeColors
        {
            get
            {
                return m_quantizeColors;
            }
            internal set
            {
                m_quantizeColors = value;
            }
        }

        /* Description of actual output image that will be returned to application.
         * These fields are computed by jpeg_start_decompress().
         * You can also use jpeg_calc_output_dimensions() to determine these values
         * in advance of calling jpeg_start_decompress().
         */

        /// <summary>
        /// scaled image width
        /// </summary>
        public int Width
        {
            get
            {
                return m_width;
            }
            internal set
            {
                m_width = value;
            }
        }

        /// <summary>
        /// scaled image height
        /// </summary>
        public int Height
        {
            get
            {
                return m_height;
            }
            internal set
            {
                m_height = value;
            }
        }

        /// <summary>
        /// # of color components in out_color_space
        /// </summary>
        public int ComponentsPerSample
        {
            get
            {
                return m_componentsPerSample;
            }
            internal set
            {
                m_componentsPerSample = value;
            }
        }

        /// <summary>
        ///  # of color components returned. it is 1 (a colormap index) when 
        /// quantizing colors; otherwise it equals out_color_components.
        /// </summary>
        public int Components
        {
            get
            {
                return m_components;
            }
            internal set
            {
                m_components = value;
            }
        }

        /* When quantizing colors, the output colormap is described by these fields.
         * The application can supply a colormap by setting colormap non-null before
         * calling jpeg_start_decompress; otherwise a colormap is created during
         * jpeg_start_decompress or jpeg_start_output.
         * The map has out_color_components rows and actual_number_of_colors columns.
         */

        /// <summary>
        /// number of entries in use
        /// </summary>
        public int ActualNumberOfColors
        {
            get
            {
                return m_actualNumberOfColors;
            }
            internal set
            {
                m_actualNumberOfColors = value;
            }
        }

        /// <summary>
        /// The color map as a 2-D pixel array
        /// </summary>
        public byte[][] Colormap
        {
            get
            {
                return m_colormap;
            }
            internal set
            {
                m_colormap = value;
            }
        }

        /// <summary>
        /// These fields record data obtained from optional markers 
        /// recognized by the JPEG library.
        ///
        /// JFIF code for pixel size units
        /// </summary>
        public DensityUnit DensityUnit
        {
            get
            {
                return m_densityUnit;
            }
            internal set
            {
                m_densityUnit = value;
            }
        }

        /// <summary>
        /// Horizontal pixel density
        /// </summary>
        public int DensityX
        {
            get
            {
                return m_densityX;
            }
            internal set
            {
                m_densityX = value;
            }
        }

        /// <summary>
        /// Vertical pixel density
        /// </summary>
        public int DensityY
        {
            get
            {
                return m_densityY;
            }
            internal set
            {
                m_densityY = value;
            }
        }
    }
    #endregion

    #region IRawImage
    abstract class IRawImage
    {
        public abstract int Width { get; }
        public abstract int Height { get; }
        public abstract ColorSpace Colorspace { get; }
        public abstract int ComponentsPerPixel { get; }

        public abstract void BeginRead();
        public abstract byte[] GetPixelRow();
        public abstract void EndRead();
    }
    #endregion

    #region Jpeg
    /// <summary>
    /// Internal wrapper for classic jpeg compressor and decompressor
    /// </summary>
    class Jpeg
    {
        private JpegCompressor m_compressor = new JpegCompressor();
        private JpegDecompressor m_decompressor = new JpegDecompressor();

        private CompressionParameters m_compressionParameters = new CompressionParameters();
        private DecompressionParameters m_decompressionParameters = new DecompressionParameters();

        /// <summary>
        /// Advanced users may set specific parameters of compression
        /// </summary>
        public CompressionParameters CompressionParameters
        {
            get
            {
                return m_compressionParameters;
            }
            set
            {
                if (value == null)
                    throw new ArgumentNullException("value");

                m_compressionParameters = value;
            }
        }

        /// <summary>
        /// Advanced users may set specific parameters of decompression
        /// </summary>
        public DecompressionParameters DecompressionParameters
        {
            get
            {
                return m_decompressionParameters;
            }
            set
            {
                if (value == null)
                    throw new ArgumentNullException("value");

                m_decompressionParameters = value;
            }
        }

        /// <summary>
        /// Compresses any image described as ICompressSource to JPEG
        /// </summary>
        /// <param name="source">Contains description of input image</param>
        /// <param name="output">Stream for output of compressed JPEG</param>
        public void Compress(IRawImage source, Stream output)
        {
            if (source == null)
                throw new ArgumentNullException("source");

            if (output == null)
                throw new ArgumentNullException("output");

            m_compressor.Image_width = source.Width;
            m_compressor.Image_height = source.Height;
            m_compressor.In_color_space = (ColorSpace)source.Colorspace;
            m_compressor.Input_components = source.ComponentsPerPixel;
            //m_compressor.Data_precision = source.DataPrecision;

            m_compressor.jpeg_set_defaults();

            //we need to set density parameters after setting of default jpeg parameters
            //m_compressor.Density_unit = source.DensityUnit;
            //m_compressor.X_density = (short)source.DensityX;
            //m_compressor.Y_density = (short)source.DensityY;

            applyParameters(m_compressionParameters);

            // Specify data destination for compression
            m_compressor.jpeg_stdio_dest(output);

            // Start compression
            m_compressor.jpeg_start_compress(true);

            // Process  pixels
            source.BeginRead();
            while (m_compressor.Next_scanline < m_compressor.Image_height)
            {
                byte[] row = source.GetPixelRow();
                if (row == null)
                {
                    throw new InvalidDataException("Row of pixels is null");
                }

                byte[][] rowForDecompressor = new byte[1][];
                rowForDecompressor[0] = row;
                m_compressor.jpeg_write_scanlines(rowForDecompressor, 1);
            }
            source.EndRead();

            // Finish compression and release memory
            m_compressor.jpeg_finish_compress();
        }

        /// <summary>
        /// Decompresses JPEG image to any image described as ICompressDestination
        /// </summary>
        /// <param name="jpeg">Stream with JPEG data</param>
        /// <param name="destination">Stream for output of compressed JPEG</param>
        public void Decompress(Stream jpeg, IDecompressor destination)
        {
            if (jpeg == null)
                throw new ArgumentNullException("jpeg");

            if (destination == null)
                throw new ArgumentNullException("destination");

            beforeDecompress(jpeg);

            // Start decompression
            m_decompressor.jpeg_start_decompress();

            LoadedImageAttributes parameters = getImageParametersFromDecompressor();
            destination.SetImageAttributes(parameters);
            destination.BeginWrite();

            /* Process data */
            while (m_decompressor.Output_scanline < m_decompressor.Output_height)
            {
                byte[][] row = JpegCommonBase.AllocJpegSamples(m_decompressor.Output_width * m_decompressor.Output_components, 1);
                m_decompressor.jpeg_read_scanlines(row, 1);
                destination.ProcessPixelsRow(row[0]);
            }

            destination.EndWrite();

            // Finish decompression and release memory.
            m_decompressor.jpeg_finish_decompress();
        }

        /// <summary>
        /// Tunes decompressor
        /// </summary>
        /// <param name="jpeg">Stream with input compressed JPEG data</param>
        private void beforeDecompress(Stream jpeg)
        {
            m_decompressor.jpeg_stdio_src(jpeg);
            /* Read file header, set default decompression parameters */
            m_decompressor.jpeg_read_header(true);

            applyParameters(m_decompressionParameters);
            m_decompressor.jpeg_calc_output_dimensions();
        }

        private LoadedImageAttributes getImageParametersFromDecompressor()
        {
            LoadedImageAttributes result = new LoadedImageAttributes();
            result.Colorspace = (ColorSpace)m_decompressor.Out_color_space;
            result.QuantizeColors = m_decompressor.Quantize_colors;
            result.Width = m_decompressor.Output_width;
            result.Height = m_decompressor.Output_height;
            result.ComponentsPerSample = m_decompressor.Out_color_components;
            result.Components = m_decompressor.Output_components;
            result.ActualNumberOfColors = m_decompressor.Actual_number_of_colors;
            result.Colormap = m_decompressor.Colormap;
            result.DensityUnit = m_decompressor.Density_unit;
            result.DensityX = m_decompressor.X_density;
            result.DensityY = m_decompressor.Y_density;
            return result;
        }

        public JpegCompressor ClassicCompressor
        {
            get
            {
                return m_compressor;
            }
        }

        public JpegDecompressor ClassicDecompressor
        {
            get
            {
                return m_decompressor;
            }
        }

        /// <summary>
        /// Delegate for application-supplied marker processing methods.
        /// Need not pass marker code since it is stored in cinfo.unread_marker.
        /// </summary>
        public delegate bool MarkerParser(Jpeg decompressor);

        /* Install a special processing method for COM or APPn markers. */
        public void SetMarkerProcessor(int markerCode, MarkerParser routine)
        {
            JpegDecompressor.jpeg_marker_parser_method f = delegate { return routine(this); };
            m_decompressor.jpeg_set_marker_processor(markerCode, f);
        }

        private void applyParameters(DecompressionParameters parameters)
        {
            if (parameters == null)
                throw new ArgumentNullException("'parameters' Cannot be null!");

            if (parameters.OutColorspace != ColorSpace.Unknown)
                m_decompressor.Out_color_space = (ColorSpace)parameters.OutColorspace;

            m_decompressor.Scale_num = parameters.ScaleNumerator;
            m_decompressor.Scale_denom = parameters.ScaleDenominator;
            m_decompressor.Buffered_image = parameters.BufferedImage;
            m_decompressor.Raw_data_out = parameters.RawDataOut;
            m_decompressor.Dct_method = (DCTMethod)parameters.DCTMethod;
            m_decompressor.Dither_mode = (DitherMode)parameters.DitherMode;
            m_decompressor.Do_fancy_upsampling = parameters.DoFancyUpsampling;
            m_decompressor.Do_block_smoothing = parameters.DoBlockSmoothing;
            m_decompressor.Quantize_colors = parameters.QuantizeColors;
            m_decompressor.Two_pass_quantize = parameters.TwoPassQuantize;
            m_decompressor.Desired_number_of_colors = parameters.DesiredNumberOfColors;
            m_decompressor.Enable_1pass_quant = parameters.EnableOnePassQuantizer;
            m_decompressor.Enable_external_quant = parameters.EnableExternalQuant;
            m_decompressor.Enable_2pass_quant = parameters.EnableTwoPassQuantizer;
        }

        private void applyParameters(CompressionParameters parameters)
        {
            if (parameters == null)
                throw new ArgumentNullException("'parameters' Cannot be null!");

            m_compressor.Smoothing_factor = parameters.SmoothingFactor;
            m_compressor.jpeg_set_quality(parameters.Quality, true);
            if (parameters.SimpleProgressive)
                m_compressor.jpeg_simple_progression();
        }
    }
    #endregion

    #region JpegBlock
    /// <summary>
    /// One block of coefficients.
    /// </summary>
    public class JpegBlock
    {
        internal short[] data = new short[JpegConstants.DCTSize2];

        /// <summary>
        /// Gets or sets the element at the specified index.
        /// </summary>
        /// <param name="index">The index of required element.</param>
        /// <value>The required element.</value>
        public short this[int index]
        {
            get
            {
                return data[index];
            }
            set
            {
                data[index] = value;
            }
        }
    }
    #endregion

    #region JpegCompressor
    /// <summary>
    /// JPEG compression routine.
    /// </summary>
    /// <seealso cref="JpegDecompressor"/>
    public class JpegCompressor : JpegCommonBase
    {
        /* These are the sample quantization tables given in JPEG spec section K.1.
         * The spec says that the values given produce "good" quality, and
         * when divided by 2, "very good" quality.
         */
        private static int[] std_luminance_quant_tbl = { 
            16, 11, 10, 16, 24, 40, 51, 61, 12, 12, 14, 19, 26,
            58, 60, 55, 14, 13, 16, 24, 40, 57, 69, 56, 14, 17,
            22, 29, 51, 87, 80, 62, 18, 22, 37, 56, 68, 109,
            103, 77, 24, 35, 55, 64, 81, 104, 113, 92, 49, 64,
            78, 87, 103, 121, 120, 101, 72, 92, 95, 98, 112,
            100, 103, 99 };

        private static int[] std_chrominance_quant_tbl = {
            17, 18, 24, 47, 99, 99, 99, 99, 18, 21, 26, 66,
            99, 99, 99, 99, 24, 26, 56, 99, 99, 99, 99, 99,
            47, 66, 99, 99, 99, 99, 99, 99, 99, 99, 99, 99,
            99, 99, 99, 99, 99, 99, 99, 99, 99, 99, 99, 99,
            99, 99, 99, 99, 99, 99, 99, 99, 99, 99, 99, 99,
            99, 99, 99, 99 };

        // Standard Huffman tables (cf. JPEG standard section K.3)
        // 
        // IMPORTANT: these are only valid for 8-bit data precision!
        private static byte[] bits_dc_luminance = { /* 0-base */ 0, 0, 1, 5, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0 };

        private static byte[] val_dc_luminance = { 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11 };

        private static byte[] bits_dc_chrominance = { /* 0-base */ 0, 0, 3, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0 };

        private static byte[] val_dc_chrominance = { 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11 };

        private static byte[] bits_ac_luminance = { /* 0-base */ 0, 0, 2, 1, 3, 3, 2, 4, 3, 5, 5, 4, 4, 0, 0, 1, 0x7d };

        private static byte[] val_ac_luminance = 
            { 0x01, 0x02, 0x03, 0x00, 0x04, 0x11, 0x05, 0x12, 0x21, 0x31, 0x41, 0x06,
              0x13, 0x51, 0x61, 0x07, 0x22, 0x71, 0x14, 0x32, 0x81, 0x91, 0xa1, 0x08,
              0x23, 0x42, 0xb1, 0xc1, 0x15, 0x52, 0xd1, 0xf0, 0x24, 0x33, 0x62, 0x72,
              0x82, 0x09, 0x0a, 0x16, 0x17, 0x18, 0x19, 0x1a, 0x25, 0x26, 0x27, 0x28,
              0x29, 0x2a, 0x34, 0x35, 0x36, 0x37, 0x38, 0x39, 0x3a, 0x43, 0x44, 0x45,
              0x46, 0x47, 0x48, 0x49, 0x4a, 0x53, 0x54, 0x55, 0x56, 0x57, 0x58, 0x59,
              0x5a, 0x63, 0x64, 0x65, 0x66, 0x67, 0x68, 0x69, 0x6a, 0x73, 0x74, 0x75,
              0x76, 0x77, 0x78, 0x79, 0x7a, 0x83, 0x84, 0x85, 0x86, 0x87, 0x88, 0x89,
              0x8a, 0x92, 0x93, 0x94, 0x95, 0x96, 0x97, 0x98, 0x99, 0x9a, 0xa2, 0xa3,
              0xa4, 0xa5, 0xa6, 0xa7, 0xa8, 0xa9, 0xaa, 0xb2, 0xb3, 0xb4, 0xb5, 0xb6,
              0xb7, 0xb8, 0xb9, 0xba, 0xc2, 0xc3, 0xc4, 0xc5, 0xc6, 0xc7, 0xc8, 0xc9,
              0xca, 0xd2, 0xd3, 0xd4, 0xd5, 0xd6, 0xd7, 0xd8, 0xd9, 0xda, 0xe1, 0xe2,
              0xe3, 0xe4, 0xe5, 0xe6, 0xe7, 0xe8, 0xe9, 0xea, 0xf1, 0xf2, 0xf3, 0xf4,
              0xf5, 0xf6, 0xf7, 0xf8, 0xf9, 0xfa };

        private static byte[] bits_ac_chrominance = { /* 0-base */ 0, 0, 2, 1, 2, 4, 4, 3, 4, 7, 5, 4, 4, 0, 1, 2, 0x77 };

        private static byte[] val_ac_chrominance = 
            { 0x00, 0x01, 0x02, 0x03, 0x11, 0x04, 0x05, 0x21, 0x31, 0x06, 0x12, 0x41,
              0x51, 0x07, 0x61, 0x71, 0x13, 0x22, 0x32, 0x81, 0x08, 0x14, 0x42, 0x91,
              0xa1, 0xb1, 0xc1, 0x09, 0x23, 0x33, 0x52, 0xf0, 0x15, 0x62, 0x72, 0xd1,
              0x0a, 0x16, 0x24, 0x34, 0xe1, 0x25, 0xf1, 0x17, 0x18, 0x19, 0x1a, 0x26,
              0x27, 0x28, 0x29, 0x2a, 0x35, 0x36, 0x37, 0x38, 0x39, 0x3a, 0x43, 0x44,
              0x45, 0x46, 0x47, 0x48, 0x49, 0x4a, 0x53, 0x54, 0x55, 0x56, 0x57, 0x58,
              0x59, 0x5a, 0x63, 0x64, 0x65, 0x66, 0x67, 0x68, 0x69, 0x6a, 0x73, 0x74,
              0x75, 0x76, 0x77, 0x78, 0x79, 0x7a, 0x82, 0x83, 0x84, 0x85, 0x86, 0x87,
              0x88, 0x89, 0x8a, 0x92, 0x93, 0x94, 0x95, 0x96, 0x97, 0x98, 0x99, 0x9a,
              0xa2, 0xa3, 0xa4, 0xa5, 0xa6, 0xa7, 0xa8, 0xa9, 0xaa, 0xb2, 0xb3, 0xb4,
              0xb5, 0xb6, 0xb7, 0xb8, 0xb9, 0xba, 0xc2, 0xc3, 0xc4, 0xc5, 0xc6, 0xc7,
              0xc8, 0xc9, 0xca, 0xd2, 0xd3, 0xd4, 0xd5, 0xd6, 0xd7, 0xd8, 0xd9, 0xda,
              0xe2, 0xe3, 0xe4, 0xe5, 0xe6, 0xe7, 0xe8, 0xe9, 0xea, 0xf2, 0xf3, 0xf4,
              0xf5, 0xf6, 0xf7, 0xf8, 0xf9, 0xfa };

        /* Destination for compressed data */
        internal DestinationManager m_dest;

        internal int m_image_width; /* input image width */
        internal int m_image_height;    /* input image height */
        internal int m_input_components;       /* # of color components in input image */
        internal ColorSpace m_in_color_space;   /* colorspace of input image */

        internal int m_data_precision;     /* bits of precision in image data */
        internal int m_num_components;     /* # of color components in JPEG image */
        internal ColorSpace m_jpeg_color_space; /* colorspace of JPEG image */

        /* comp_info[i] describes component that appears i'th in SOF */
        private JpegComponent[] m_comp_info;

        /* ptrs to coefficient quantization tables, or null if not defined */
        internal JpegQuantizationTable[] m_quant_tbl_ptrs = new JpegQuantizationTable[JpegConstants.NumberOfQuantTables];

        /* ptrs to Huffman coding tables, or null if not defined */
        internal JpegHuffmanTable[] m_dc_huff_tbl_ptrs = new JpegHuffmanTable[JpegConstants.NumberOfHuffmanTables];
        internal JpegHuffmanTable[] m_ac_huff_tbl_ptrs = new JpegHuffmanTable[JpegConstants.NumberOfHuffmanTables];

        /* The default value of scan_info is null, which causes a single-scan
         * sequential JPEG file to be emitted.  To create a multi-scan file,
         * set num_scans and scan_info to point to an array of scan definitions.
         */
        internal int m_num_scans;      /* # of entries in scan_info array */

        /* script for multi-scan file, or null */
        internal JpegScanInfo[] m_scan_info;

        internal bool m_raw_data_in;       /* true=caller supplies downsampled data */
        internal bool m_optimize_coding;   /* true=optimize entropy encoding parms */
        internal bool m_CCIR601_sampling;  /* true=first samples are cosited */
        internal int m_smoothing_factor;       /* 1..100, or 0 for no input smoothing */
        internal DCTMethod m_dct_method;    /* DCT algorithm selector */

        internal int m_restart_interval; /* MCUs per restart, or 0 for no restart */
        internal int m_restart_in_rows;        /* if > 0, MCU rows per restart interval */

        internal bool m_write_JFIF_header; /* should a JFIF marker be written? */
        internal byte m_JFIF_major_version;   /* What to write for the JFIF version number */
        internal byte m_JFIF_minor_version;

        internal DensityUnit m_density_unit;     /* JFIF code for pixel size units */
        internal short m_X_density;       /* Horizontal pixel density */
        internal short m_Y_density;       /* Vertical pixel density */
        internal bool m_write_Adobe_marker;    /* should an Adobe marker be written? */

        internal int m_next_scanline;   /* 0 .. image_height-1  */

        /* Remaining fields are known throughout compressor, but generally
         * should not be touched by a surrounding application.
         */

        /*
         * These fields are computed during compression startup
         */
        internal bool m_progressive_mode;  /* true if scan script uses progressive mode */
        internal int m_max_h_samp_factor;  /* largest h_samp_factor */
        internal int m_max_v_samp_factor;  /* largest v_samp_factor */

        internal int m_total_iMCU_rows; /* # of iMCU rows to be input to coefficient controller */
        /* The coefficient controller receives data in units of MCU rows as defined
         * for fully interleaved scans (whether the JPEG file is interleaved or not).
         * There are v_samp_factor * DCTSize sample rows of each component in an
         * "iMCU" (interleaved MCU) row.
         */

        /*
         * These fields are valid during any one scan.
         * They describe the components and MCUs actually appearing in the scan.
         */
        internal int m_comps_in_scan;      /* # of JPEG components in this scan */
        internal int[] m_cur_comp_info = new int[JpegConstants.MaxComponentsInScan];
        /* *cur_comp_info[i] is index of m_comp_info that describes component that appears i'th in SOS */

        internal int m_MCUs_per_row;    /* # of MCUs across the image */
        internal int m_MCU_rows_in_scan;    /* # of MCU rows in the image */

        internal int m_blocks_in_MCU;      /* # of DCT blocks per MCU */
        internal int[] m_MCU_membership = new int[JpegConstants.CompressorMaxBlocksInMCU];
        /* MCU_membership[i] is index in cur_comp_info of component owning */
        /* i'th block in an MCU */

        /* progressive JPEG parameters for scan */
        internal int m_Ss;
        internal int m_Se;
        internal int m_Ah;
        internal int m_Al;

        /*
         * Links to compression subobjects (methods and private variables of modules)
         */
        internal JpegCompressorMaster m_master;
        internal JpegCompressorMainController m_main;
        internal JpegCompressorPrepController m_prep;
        internal JpegCompressorCoefController m_coef;
        internal JpegMarkerWriter m_marker;
        internal ColorConverter m_cconvert;
        internal JpegDownsampler m_downsample;
        internal JpegFowardDCT m_fdct;
        internal JpegEntropyEncoder m_entropy;
        internal JpegScanInfo[] m_script_space; /* workspace for jpeg_simple_progression */
        internal int m_script_space_size;

        /// <summary>
        /// Initializes a new instance of the <see cref="JpegCompressor"/> class.
        /// </summary>
        public JpegCompressor()
            : base()
        {
            initialize();
        }

        /// <summary>
        /// Retrieves <c>false</c> because this is not decompressor.
        /// </summary>
        /// <value><c>false</c></value>
        public override bool IsDecompressor
        {
            get { return false; }
        }

        /// <summary>
        /// Gets or sets the destination for compressed data
        /// </summary>
        /// <value>The destination for compressed data.</value>
        public LibJpeg.DestinationManager Dest
        {
            get { return m_dest; }
            set { m_dest = value; }
        }

        /* Description of source image --- these fields must be filled in by
         * outer application before starting compression.  in_color_space must
         * be correct before you can even call jpeg_set_defaults().
         */

        /// <summary>
        /// Gets or sets the width of image, in pixels.
        /// </summary>
        /// <value>The width of image.</value>
        /// <seealso href="07136fd7-d482-48de-b88c-1a4b9658c69e.htm" target="_self">Compression details</seealso>
        public int Image_width
        {
            get { return m_image_width; }
            set { m_image_width = value; }
        }

        /// <summary>
        /// Gets or sets the height of image, in pixels.
        /// </summary>
        /// <value>The height of image.</value>
        /// <seealso href="07136fd7-d482-48de-b88c-1a4b9658c69e.htm" target="_self">Compression details</seealso>
        public int Image_height
        {
            get { return m_image_height; }
            set { m_image_height = value; }
        }

        /// <summary>
        /// Gets or sets the number of color channels (components per pixel)
        /// </summary>
        /// <value>The number of color channels.</value>
        /// <seealso href="07136fd7-d482-48de-b88c-1a4b9658c69e.htm" target="_self">Compression details</seealso>
        public int Input_components
        {
            get { return m_input_components; }
            set { m_input_components = value; }
        }

        /// <summary>
        /// Gets or sets the color space of source image.
        /// </summary>
        /// <value>The color space.</value>
        /// <seealso href="07136fd7-d482-48de-b88c-1a4b9658c69e.htm" target="_self">Compression details</seealso>
        /// <seealso href="c90654b9-f3f4-4319-80d1-979c73d84e76.htm" target="_self">Special color spaces</seealso>
        public LibJpeg.ColorSpace In_color_space
        {
            get { return m_in_color_space; }
            set { m_in_color_space = value; }
        }

        /* Compression parameters --- these fields must be set before calling
         * jpeg_start_compress().  We recommend calling jpeg_set_defaults() to
         * initialize everything to reasonable defaults, then changing anything
         * the application specifically wants to change.  That way you won't get
         * burnt when new parameters are added.  Also note that there are several
         * helper routines to simplify changing parameters.
         */

        // bits of precision in image data


        /// <summary>
        /// Gets or sets the number of bits of precision in image data.
        /// </summary>
        /// <remarks>Default value: 8<br/>
        /// The number of bits.
        /// </remarks>
        /// <value>The data precision.</value>
        public int Data_precision
        {
            get { return m_data_precision; }
            set { m_data_precision = value; }
        }

        /// <summary>
        /// Gets or sets the number of color components for JPEG color space.
        /// </summary>
        /// <value>The number of color components for JPEG color space.</value>
        public int Num_components
        {
            get { return m_num_components; }
            set { m_num_components = value; }
        }

        /// <summary>
        /// Gets or sets the JPEG color space.
        /// </summary>
        /// <remarks>We recommend to use <see cref="jpeg_set_colorspace"/> if you want to change this.</remarks>
        /// <value>The JPEG color space.</value>
        public ColorSpace Jpeg_color_space
        {
            get { return m_jpeg_color_space; }
            set { m_jpeg_color_space = value; }
        }

        // true=caller supplies downsampled data


        /// <summary>
        /// Gets or sets a value indicating whether you will be supplying raw data.
        /// </summary>
        /// <remarks>Default value: <c>false</c></remarks>
        /// <value><c>true</c> if you will be supplying raw data; otherwise, <c>false</c>.</value>
        /// <seealso cref="JpegCompressor.jpeg_write_raw_data"/>
        public bool Raw_data_in
        {
            get { return m_raw_data_in; }
            set { m_raw_data_in = value; }
        }

        /// <summary>
        /// Gets or sets a value indicating a way of using Huffman coding tables.
        /// </summary>
        /// <remarks>When this is <c>true</c>, you need not supply Huffman tables at all, and any you do supply will be overwritten.</remarks>
        /// <value><c>true</c> causes the compressor to compute optimal Huffman coding tables 
        /// for the image. This requires an extra pass over the data and therefore costs a good 
        /// deal of space and time. The default is <c>false</c>, which tells the compressor to use the 
        /// supplied or default Huffman tables. In most cases optimal tables save only a few 
        /// percent of file size compared to the default tables.</value>
        /// <seealso href="ce3f6712-3633-4a58-af07-626a4fba9ae4.htm" target="_self">Compression parameter selection</seealso>
        public bool Optimize_coding
        {
            get { return m_optimize_coding; }
            set { m_optimize_coding = value; }
        }

        /// <summary>
        /// Gets or sets a value indicating whether first samples are cosited.
        /// </summary>
        /// <value><c>true</c> if first samples are cosited; otherwise, <c>false</c>.</value>
        public bool CCIR601_sampling
        {
            get { return m_CCIR601_sampling; }
            set { m_CCIR601_sampling = value; }
        }

        /// <summary>
        /// Gets or sets the coefficient of image smoothing.
        /// </summary>
        /// <remarks>Default value: 0<br/>
        /// If non-zero, the input image is smoothed; the value should be 1 for minimal smoothing 
        /// to 100 for maximum smoothing.</remarks>
        /// <value>The coefficient of image smoothing.</value>
        /// <seealso href="ce3f6712-3633-4a58-af07-626a4fba9ae4.htm" target="_self">Compression parameter selection</seealso>
        public int Smoothing_factor
        {
            get { return m_smoothing_factor; }
            set { m_smoothing_factor = value; }
        }

        /// <summary>
        /// Gets or sets the algorithm used for the DCT step.
        /// </summary>
        /// <value>The DCT algorithm.</value>
        /// <seealso href="ce3f6712-3633-4a58-af07-626a4fba9ae4.htm" target="_self">Compression parameter selection</seealso>
        public DCTMethod Dct_method
        {
            get { return m_dct_method; }
            set { m_dct_method = value; }
        }

        /* The restart interval can be specified in absolute MCUs by setting
         * restart_interval, or in MCU rows by setting restart_in_rows
         * (in which case the correct restart_interval will be figured
         * for each scan).
         */

        /// <summary>
        /// Gets or sets the exact interval in MCU blocks.
        /// </summary>
        /// <remarks>Default value: 0<br/>
        /// One restart marker per MCU row is often a good choice. The overhead of restart markers 
        /// is higher in grayscale JPEG files than in color files, and MUCH higher in progressive JPEGs. 
        /// If you use restarts, you may want to use larger intervals in those cases.</remarks>
        /// <value>The restart interval.</value>
        /// <seealso cref="JpegCompressor.Restart_in_rows"/>
        /// <seealso href="ce3f6712-3633-4a58-af07-626a4fba9ae4.htm" target="_self">Compression parameter selection</seealso>
        public int Restart_interval
        {
            get { return m_restart_interval; }
            set { m_restart_interval = value; }
        }

        /// <summary>
        /// Gets or sets the interval in MCU rows.
        /// </summary>
        /// <remarks>Default value: 0<br/>
        /// If Restart_in_rows is not 0, then <see cref="JpegCompressor.Restart_interval"/> is set 
        /// after the image width in MCUs is computed.<br/>
        /// One restart marker per MCU row is often a good choice. 
        /// The overhead of restart markers is higher in grayscale JPEG files than in color files, and MUCH higher in progressive JPEGs. If you use restarts, you may want to use larger intervals in those cases.
        /// </remarks>
        /// <value>The restart interval in MCU rows.</value>
        /// <seealso cref="JpegCompressor.Restart_interval"/>
        /// <seealso href="ce3f6712-3633-4a58-af07-626a4fba9ae4.htm" target="_self">Compression parameter selection</seealso>
        public int Restart_in_rows
        {
            get { return m_restart_in_rows; }
            set { m_restart_in_rows = value; }
        }

        /* Parameters controlling emission of special markers. */

        /// <summary>
        /// Gets or sets a value indicating whether the JFIF APP0 marker is emitted.
        /// </summary>
        /// <remarks><see cref="JpegCompressor.jpeg_set_defaults"/> and 
        /// <see cref="JpegCompressor.jpeg_set_colorspace"/> set this <c>true</c> 
        /// if a JFIF-legal JPEG color space (i.e., YCbCr or grayscale) is selected, otherwise <c>false</c>.</remarks>
        /// <value><c>true</c> if JFIF APP0 marker is emitted; otherwise, <c>false</c>.</value>
        /// <seealso cref="JpegCompressor.JFIF_major_version"/>
        /// <seealso href="ce3f6712-3633-4a58-af07-626a4fba9ae4.htm" target="_self">Compression parameter selection</seealso>
        public bool Write_JFIF_header
        {
            get { return m_write_JFIF_header; }
            set { m_write_JFIF_header = value; }
        }

        /// <summary>
        /// Gets or sets the version number to be written into the JFIF marker.
        /// </summary>
        /// <remarks><see cref="JpegCompressor.jpeg_set_defaults"/> initializes the version to 
        /// 1.01 (major=minor=1). You should set it to 1.02 (major=1, minor=2) if you plan to write any 
        /// JFIF 1.02 extension markers.</remarks>
        /// <value>The version number to be written into the JFIF marker.</value>
        /// <seealso cref="JpegCompressor.JFIF_minor_version"/>
        /// <seealso cref="JpegCompressor.Write_JFIF_header"/>
        public byte JFIF_major_version
        {
            get { return m_JFIF_major_version; }
            set { m_JFIF_major_version = value; }
        }

        /// <summary>
        /// Gets or sets the version number to be written into the JFIF marker.
        /// </summary>
        /// <remarks><see cref="JpegCompressor.jpeg_set_defaults"/> initializes the version to 
        /// 1.01 (major=minor=1). You should set it to 1.02 (major=1, minor=2) if you plan to write any 
        /// JFIF 1.02 extension markers.</remarks>
        /// <value>The version number to be written into the JFIF marker.</value>
        /// <seealso cref="JpegCompressor.JFIF_major_version"/>
        /// <seealso cref="JpegCompressor.Write_JFIF_header"/>
        public byte JFIF_minor_version
        {
            get { return m_JFIF_minor_version; }
            set { m_JFIF_minor_version = value; }
        }

        /* These three values are not used by the JPEG code, merely copied */
        /* into the JFIF APP0 marker.  density_unit can be 0 for unknown, */
        /* 1 for dots/inch, or 2 for dots/cm.  Note that the pixel aspect */
        /* ratio is defined by X_density/Y_density even when density_unit=0. */

        /// <summary>
        /// Gets or sets the resolution information to be written into the JFIF marker; not used otherwise.
        /// </summary>
        /// <remarks>Default value: <see cref="F:BitMiracle.LibJpeg.Classic.DensityUnit.Unknown"/><br/>
        /// The pixel aspect ratio is defined by 
        /// <see cref="JpegCompressor.X_density"/>/<see cref="JpegCompressor.Y_density"/> 
        /// even when Density_unit is <see cref="F:BitMiracle.LibJpeg.Classic.DensityUnit.Unknown">Unknown</see>.</remarks>
        /// <value>The density unit.</value>
        /// <seealso cref="JpegCompressor.X_density"/>
        /// <seealso cref="JpegCompressor.Y_density"/>
        /// <seealso href="ce3f6712-3633-4a58-af07-626a4fba9ae4.htm" target="_self">Compression parameter selection</seealso>
        public DensityUnit Density_unit
        {
            get { return m_density_unit; }
            set { m_density_unit = value; }
        }

        // Horizontal pixel density

        /// <summary>
        /// Gets or sets the horizontal component of pixel ratio.
        /// </summary>
        /// <remarks>Default value: 1</remarks>
        /// <value>The horizontal density.</value>
        /// <seealso cref="JpegCompressor.Density_unit"/>
        /// <seealso cref="JpegCompressor.Y_density"/>
        public short X_density
        {
            get { return m_X_density; }
            set { m_X_density = value; }
        }

        /// <summary>
        /// Gets or sets the vertical component of pixel ratio.
        /// </summary>
        /// <remarks>Default value: 1</remarks>
        /// <value>The vertical density.</value>
        /// <seealso cref="JpegCompressor.Density_unit"/>
        /// <seealso cref="JpegCompressor.X_density"/>
        public short Y_density
        {
            get { return m_Y_density; }
            set { m_Y_density = value; }
        }

        /// <summary>
        /// Gets or sets a value indicating whether to emit Adobe APP14 marker.
        /// </summary>
        /// <remarks><see cref="JpegCompressor.jpeg_set_defaults"/> and <see cref="JpegCompressor.jpeg_set_colorspace"/> 
        /// set this <c>true</c> if JPEG color space RGB, CMYK, or YCCK is selected, otherwise <c>false</c>. 
        /// It is generally a bad idea to set both <see cref="JpegCompressor.Write_JFIF_header"/> and 
        /// <see cref="JpegCompressor.Write_Adobe_marker"/>. 
        /// In fact, you probably shouldn't change the default settings at all - the default behavior ensures that the JPEG file's 
        /// color space can be recognized by the decoder.</remarks>
        /// <value>If <c>true</c> an Adobe APP14 marker is emitted; <c>false</c>, otherwise.</value>
        /// <seealso href="ce3f6712-3633-4a58-af07-626a4fba9ae4.htm" target="_self">Compression parameter selection</seealso>
        public bool Write_Adobe_marker
        {
            get { return m_write_Adobe_marker; }
            set { m_write_Adobe_marker = value; }
        }

        /// <summary>
        /// Gets the largest vertical sample factor.
        /// </summary>
        /// <value>The largest vertical sample factor.</value>
        /// <seealso href="ce3f6712-3633-4a58-af07-626a4fba9ae4.htm" target="_self">Compression parameter selection</seealso>
        public int Max_v_samp_factor
        {
            get { return m_max_v_samp_factor; }
        }

        /// <summary>
        /// Gets the components that appears in SOF.
        /// </summary>
        /// <value>The component info array.</value>
        public JpegComponent[] Component_info
        {
            get { return m_comp_info; }
        }

        /* ptrs to coefficient quantization tables, or null if not defined */
        /* ptrs to coefficient quantization tables, or null if not defined */
        /// <summary>
        /// Gets the coefficient quantization tables.
        /// </summary>
        /// <value>The coefficient quantization tables or null if not defined.</value>
        public JpegQuantizationTable[] Quant_tbl_ptrs
        {
            get { return m_quant_tbl_ptrs; }
        }

        /// <summary>
        /// Gets the Huffman coding tables.
        /// </summary>
        /// <value>The Huffman coding tables or null if not defined.</value>
        public JpegHuffmanTable[] Dc_huff_tbl_ptrs
        {
            get { return m_dc_huff_tbl_ptrs; }
        }

        /// <summary>
        /// Gets the Huffman coding tables.
        /// </summary>
        /// <value>The Huffman coding tables or null if not defined.</value>
        public JpegHuffmanTable[] Ac_huff_tbl_ptrs
        {
            get { return m_ac_huff_tbl_ptrs; }
        }

        /// <summary>
        /// Gets the index of next scanline to be written to <see cref="JpegCompressor.jpeg_write_scanlines"/>.
        /// </summary>
        /// <remarks>Application may use this to control its processing loop, 
        /// e.g., "while (Next_scanline &lt; Image_height)"</remarks>
        /// <value>Range: from 0 to (Image_height - 1)</value>
        /// <seealso cref="JpegCompressor.jpeg_write_scanlines"/>
        public int Next_scanline
        {
            get { return m_next_scanline; }
        }

        /// <summary>
        /// Abort processing of a JPEG compression operation.
        /// </summary>
        public void jpeg_abort_compress()
        {
            // use common routine
            jpeg_abort();
        }

        /// <summary>
        /// Forcibly suppress or un-suppress all quantization and Huffman tables.
        /// </summary>
        /// <remarks>Marks all currently defined tables as already written (if suppress)
        /// or not written (if !suppress). This will control whether they get 
        /// emitted by a subsequent <see cref="JpegCompressor.jpeg_start_compress"/> call.<br/>
        /// 
        /// This routine is exported for use by applications that want to produce
        /// abbreviated JPEG datastreams.</remarks>
        /// <param name="suppress">if set to <c>true</c> then suppress tables; 
        /// otherwise unsuppress.</param>
        public void jpeg_suppress_tables(bool suppress)
        {
            for (int i = 0; i < JpegConstants.NumberOfQuantTables; i++)
            {
                if (m_quant_tbl_ptrs[i] != null)
                    m_quant_tbl_ptrs[i].Sent_table = suppress;
            }

            for (int i = 0; i < JpegConstants.NumberOfHuffmanTables; i++)
            {
                if (m_dc_huff_tbl_ptrs[i] != null)
                    m_dc_huff_tbl_ptrs[i].Sent_table = suppress;

                if (m_ac_huff_tbl_ptrs[i] != null)
                    m_ac_huff_tbl_ptrs[i].Sent_table = suppress;
            }
        }

        /// <summary>
        /// Finishes JPEG compression.
        /// </summary>
        /// <remarks>If a multipass operating mode was selected, this may do a great 
        /// deal of work including most of the actual output.</remarks>
        public void jpeg_finish_compress()
        {
            int iMCU_row;

            if (m_global_state == JpegState.CSTATE_SCANNING || m_global_state == JpegState.CSTATE_RAW_OK)
            {
                /* Terminate first pass */
                if (m_next_scanline < m_image_height)
                    throw new Exception("Application transferred too few scanlines");
                m_master.finish_pass();
            }
            else if (m_global_state != JpegState.CSTATE_WRCOEFS)
                throw new Exception(String.Format("Improper call to JPEG library in state {0}", (int)m_global_state));

            /* Perform any remaining passes */
            while (!m_master.IsLastPass())
            {
                m_master.prepare_for_pass();
                for (iMCU_row = 0; iMCU_row < m_total_iMCU_rows; iMCU_row++)
                {

                    /* We bypass the main controller and invoke coef controller directly;
                    * all work is being done from the coefficient buffer.
                    */
                    if (!m_coef.compress_data(null))
                        throw new Exception("Suspension not allowed here");
                }

                m_master.finish_pass();
            }

            /* Write EOI, do final cleanup */
            m_marker.write_file_trailer();
            m_dest.term_destination();

            /* We can use jpeg_abort to release memory and reset global_state */
            jpeg_abort();
        }


        /// <summary>
        /// Write a special marker.
        /// </summary>
        /// <remarks>This is only recommended for writing COM or APPn markers. 
        /// Must be called after <see cref="JpegCompressor.jpeg_start_compress"/> and before first call to 
        /// <see cref="JpegCompressor.jpeg_write_scanlines"/> or <see cref="JpegCompressor.jpeg_write_raw_data"/>.
        /// </remarks>
        /// <param name="marker">Specify the marker type parameter as <see cref="JpegMarkerType"/>.COM for COM or 
        /// <see cref="JpegMarkerType"/>.APP0 + n for APPn. (Actually, jpeg_write_marker will let you write any marker type, 
        /// but we don't recommend writing any other kinds of marker)</param>
        /// <param name="data">The data associated with the marker.</param>
        /// <seealso href="81c88818-a5d7-4550-9ce5-024a768f7b1e.htm" target="_self">Special markers</seealso>
        /// <seealso cref="JpegMarkerType"/>
        public void jpeg_write_marker(int marker, byte[] data)
        {
            if (m_next_scanline != 0 || (m_global_state != JpegState.CSTATE_SCANNING && m_global_state != JpegState.CSTATE_RAW_OK && m_global_state != JpegState.CSTATE_WRCOEFS))
                throw new Exception(String.Format("Improper call to JPEG library in state {0}", (int)m_global_state));

            m_marker.write_marker_header(marker, data.Length);

            for (int i = 0; i < data.Length; i++)
                m_marker.write_marker_byte(data[i]);
        }

        /// <summary>
        /// Writes special marker's header.
        /// </summary>
        /// <param name="marker">Special marker.</param>
        /// <param name="datalen">Length of data associated with the marker.</param>
        /// <remarks>After calling this method you need to call <see cref="JpegCompressor.jpeg_write_m_byte"/>
        /// exactly the number of times given in the length parameter.<br/>
        /// This method lets you empty the output buffer partway through a marker, which might be important when 
        /// using a suspending data destination module. In any case, if you are using a suspending destination, 
        /// you should flush its buffer after inserting any special markers.</remarks>
        /// <seealso cref="JpegCompressor.jpeg_write_m_byte"/>
        /// <seealso cref="JpegCompressor.jpeg_write_marker"/>
        /// <seealso href="81c88818-a5d7-4550-9ce5-024a768f7b1e.htm" target="_self">Special markers</seealso>
        public void jpeg_write_m_header(int marker, int datalen)
        {
            if (m_next_scanline != 0 || (m_global_state != JpegState.CSTATE_SCANNING && m_global_state != JpegState.CSTATE_RAW_OK && m_global_state != JpegState.CSTATE_WRCOEFS))
                throw new Exception(String.Format("Improper call to JPEG library in state {0}", (int)m_global_state));

            m_marker.write_marker_header(marker, datalen);
        }

        /// <summary>
        /// Writes a byte of special marker's data.
        /// </summary>
        /// <param name="val">The byte of data.</param>
        /// <seealso cref="JpegCompressor.jpeg_write_m_header"/>
        public void jpeg_write_m_byte(byte val)
        {
            m_marker.write_marker_byte(val);
        }

        /// <summary>
        /// Alternate compression function: just write an abbreviated table file.
        /// </summary>
        /// <remarks>Before calling this, all parameters and a data destination must be set up.<br/>
        /// 
        /// To produce a pair of files containing abbreviated tables and abbreviated
        /// image data, one would proceed as follows:<br/>
        /// 
        /// <c>Initialize JPEG object<br/>
        /// Set JPEG parameters<br/>
        /// Set destination to table file<br/>
        /// <see cref="JpegCompressor.jpeg_write_tables">jpeg_write_tables();</see><br/>
        /// Set destination to image file<br/>
        /// <see cref="JpegCompressor.jpeg_start_compress">jpeg_start_compress(false);</see><br/>
        /// Write data...<br/>
        /// <see cref="JpegCompressor.jpeg_finish_compress">jpeg_finish_compress();</see><br/>
        /// </c><br/>
        /// 
        /// jpeg_write_tables has the side effect of marking all tables written
        /// (same as <see cref="JpegCompressor.jpeg_suppress_tables">jpeg_suppress_tables(true)</see>).
        /// Thus a subsequent <see cref="JpegCompressor.jpeg_start_compress">jpeg_start_compress</see> 
        /// will not re-emit the tables unless it is passed <c>write_all_tables=true</c>.
        /// </remarks>
        public void jpeg_write_tables()
        {
            if (m_global_state != JpegState.CSTATE_START)
                throw new Exception(String.Format("Improper call to JPEG library in state {0}", (int)m_global_state));

            /* (Re)initialize destination modules */
            m_dest.init_destination();

            /* Initialize the marker writer ... bit of a crock to do it here. */
            m_marker = new JpegMarkerWriter(this);

            /* Write them tables! */
            m_marker.write_tables_only();

            /* And clean up. */
            m_dest.term_destination();
        }

        /// <summary>
        /// Sets output stream.
        /// </summary>
        /// <param name="outfile">The output stream.</param>
        /// <remarks>The caller must have already opened the stream, and is responsible
        /// for closing it after finishing compression.</remarks>
        /// <seealso href="07136fd7-d482-48de-b88c-1a4b9658c69e.htm" target="_self">Compression details</seealso>
        public void jpeg_stdio_dest(Stream outfile)
        {
            m_dest = new DestinationManagerImpl(this, outfile);
        }

        /// <summary>
        /// Jpeg_set_defaultses this instance.
        /// </summary>
        /// <remarks>Uses only the input image's color space (property <see cref="JpegCompressor.In_color_space"/>, 
        /// which must already be set in <see cref="JpegCompressor"/>). Many applications will only need 
        /// to use this routine and perhaps <see cref="JpegCompressor.jpeg_set_quality"/>.
        /// </remarks>
        /// <seealso href="ce3f6712-3633-4a58-af07-626a4fba9ae4.htm" target="_self">Compression parameter selection</seealso>
        public void jpeg_set_defaults()
        {
            /* Safety check to ensure start_compress not called yet. */
            if (m_global_state != JpegState.CSTATE_START)
                throw new Exception(String.Format("Improper call to JPEG library in state {0}", (int)m_global_state));

            /* Allocate comp_info array large enough for maximum component count.
            * Array is made permanent in case application wants to compress
            * multiple images at same param settings.
            */
            if (m_comp_info == null)
            {
                m_comp_info = JpegComponent.createArrayOfComponents(JpegConstants.MaxComponents);
            }

            /* Initialize everything not dependent on the color space */

            m_data_precision = JpegConstants.BitsInSample;

            /* Set up two quantization tables using default quality of 75 */
            jpeg_set_quality(75, true);

            /* Set up two Huffman tables */
            std_huff_tables();

            /* Default is no multiple-scan output */
            m_scan_info = null;
            m_num_scans = 0;

            /* Expect normal source image, not raw downsampled data */
            m_raw_data_in = false;

            /* By default, don't do extra passes to optimize entropy coding */
            m_optimize_coding = false;

            /* The standard Huffman tables are only valid for 8-bit data precision.
            * If the precision is higher, force optimization on so that usable
            * tables will be computed.  This test can be removed if default tables
            * are supplied that are valid for the desired precision.
            */
            if (m_data_precision > 8)
                m_optimize_coding = true;

            /* By default, use the simpler non-cosited sampling alignment */
            m_CCIR601_sampling = false;

            /* No input smoothing */
            m_smoothing_factor = 0;

            /* DCT algorithm preference */
            m_dct_method = JpegConstants.DefaultDCTMethod;

            /* No restart markers */
            m_restart_interval = 0;
            m_restart_in_rows = 0;

            /* Fill in default JFIF marker parameters.  Note that whether the marker
            * will actually be written is determined by jpeg_set_colorspace.
            *
            * By default, the library emits JFIF version code 1.01.
            * An application that wants to emit JFIF 1.02 extension markers should set
            * JFIF_minor_version to 2.  We could probably get away with just defaulting
            * to 1.02, but there may still be some decoders in use that will complain
            * about that; saying 1.01 should minimize compatibility problems.
            */
            m_JFIF_major_version = 1; /* Default JFIF version = 1.01 */
            m_JFIF_minor_version = 1;
            m_density_unit = DensityUnit.Unknown;    /* Pixel size is unknown by default */
            m_X_density = 1;       /* Pixel aspect ratio is square by default */
            m_Y_density = 1;

            /* Choose JPEG colorspace based on input space, set defaults accordingly */
            jpeg_default_colorspace();
        }

        // Compression parameter setup aids

        /// <summary>
        /// Set the JPEG colorspace (property <see cref="JpegCompressor.Jpeg_color_space"/>,
        /// and choose colorspace-dependent parameters appropriately.
        /// </summary>
        /// <param name="colorspace">The required colorspace.</param>
        /// <remarks>See <see href="c90654b9-f3f4-4319-80d1-979c73d84e76.htm" target="_self">Special color spaces</see>, 
        /// below, before using this. A large number of parameters, including all per-component parameters, 
        /// are set by this routine; if you want to twiddle individual parameters you should call 
        /// <c>jpeg_set_colorspace</c> before rather than after.</remarks>
        /// <seealso href="ce3f6712-3633-4a58-af07-626a4fba9ae4.htm" target="_self">Compression parameter selection</seealso>
        /// <seealso href="c90654b9-f3f4-4319-80d1-979c73d84e76.htm" target="_self">Special color spaces</seealso>
        public void jpeg_set_colorspace(ColorSpace colorspace)
        {
            int ci;

            /* Safety check to ensure start_compress not called yet. */
            if (m_global_state != JpegState.CSTATE_START)
                throw new Exception(String.Format("Improper call to JPEG library in state {0}", (int)m_global_state));

            /* For all colorspaces, we use Q and Huff tables 0 for luminance components,
            * tables 1 for chrominance components.
            */

            m_jpeg_color_space = colorspace;

            m_write_JFIF_header = false; /* No marker for non-JFIF colorspaces */
            m_write_Adobe_marker = false; /* write no Adobe marker by default */

            switch (colorspace)
            {
                case ColorSpace.Grayscale:
                    m_write_JFIF_header = true; /* Write a JFIF marker */
                    m_num_components = 1;
                    /* JFIF specifies component ID 1 */
                    jpeg_set_colorspace_SET_COMP(0, 1, 1, 1, 0, 0, 0);
                    break;
                case ColorSpace.RGB:
                    m_write_Adobe_marker = true; /* write Adobe marker to flag RGB */
                    m_num_components = 3;
                    jpeg_set_colorspace_SET_COMP(0, 0x52 /* 'R' */, 1, 1, 0, 0, 0);
                    jpeg_set_colorspace_SET_COMP(1, 0x47 /* 'G' */, 1, 1, 0, 0, 0);
                    jpeg_set_colorspace_SET_COMP(2, 0x42 /* 'B' */, 1, 1, 0, 0, 0);
                    break;
                case ColorSpace.YCbCr:
                    m_write_JFIF_header = true; /* Write a JFIF marker */
                    m_num_components = 3;
                    /* JFIF specifies component IDs 1,2,3 */
                    /* We default to 2x2 subsamples of chrominance */
                    jpeg_set_colorspace_SET_COMP(0, 1, 2, 2, 0, 0, 0);
                    jpeg_set_colorspace_SET_COMP(1, 2, 1, 1, 1, 1, 1);
                    jpeg_set_colorspace_SET_COMP(2, 3, 1, 1, 1, 1, 1);
                    break;
                case ColorSpace.CMYK:
                    m_write_Adobe_marker = true; /* write Adobe marker to flag CMYK */
                    m_num_components = 4;
                    jpeg_set_colorspace_SET_COMP(0, 0x43 /* 'C' */, 1, 1, 0, 0, 0);
                    jpeg_set_colorspace_SET_COMP(1, 0x4D /* 'M' */, 1, 1, 0, 0, 0);
                    jpeg_set_colorspace_SET_COMP(2, 0x59 /* 'Y' */, 1, 1, 0, 0, 0);
                    jpeg_set_colorspace_SET_COMP(3, 0x4B /* 'K' */, 1, 1, 0, 0, 0);
                    break;
                case ColorSpace.YCCK:
                    m_write_Adobe_marker = true; /* write Adobe marker to flag YCCK */
                    m_num_components = 4;
                    jpeg_set_colorspace_SET_COMP(0, 1, 2, 2, 0, 0, 0);
                    jpeg_set_colorspace_SET_COMP(1, 2, 1, 1, 1, 1, 1);
                    jpeg_set_colorspace_SET_COMP(2, 3, 1, 1, 1, 1, 1);
                    jpeg_set_colorspace_SET_COMP(3, 4, 2, 2, 0, 0, 0);
                    break;
                case ColorSpace.Unknown:
                    m_num_components = m_input_components;
                    if (m_num_components < 1 || m_num_components > JpegConstants.MaxComponents)
                        throw new Exception(String.Format("Too many color components: {0}, max {1}", m_num_components, JpegConstants.MaxComponents));
                    for (ci = 0; ci < m_num_components; ci++)
                    {
                        jpeg_set_colorspace_SET_COMP(ci, ci, 1, 1, 0, 0, 0);
                    }
                    break;
                default:
                    throw new Exception("Bad Jpeg ColorSpace.");
            }
        }

        /// <summary>
        /// Select an appropriate JPEG colorspace based on <see cref="JpegCompressor.In_color_space"/>,
        /// and calls <see cref="JpegCompressor.jpeg_set_colorspace"/>
        /// </summary>
        /// <remarks>This is actually a subroutine of <see cref="jpeg_set_defaults"/>. 
        /// It's broken out in case you want to change just the colorspace-dependent JPEG parameters.</remarks>
        /// <seealso href="ce3f6712-3633-4a58-af07-626a4fba9ae4.htm" target="_self">Compression parameter selection</seealso>
        public void jpeg_default_colorspace()
        {
            switch (m_in_color_space)
            {
                case ColorSpace.Grayscale:
                    jpeg_set_colorspace(ColorSpace.Grayscale);
                    break;
                case ColorSpace.RGB:
                    jpeg_set_colorspace(ColorSpace.YCbCr);
                    break;
                case ColorSpace.YCbCr:
                    jpeg_set_colorspace(ColorSpace.YCbCr);
                    break;
                case ColorSpace.CMYK:
                    jpeg_set_colorspace(ColorSpace.CMYK); /* By default, no translation */
                    break;
                case ColorSpace.YCCK:
                    jpeg_set_colorspace(ColorSpace.YCCK);
                    break;
                case ColorSpace.Unknown:
                    jpeg_set_colorspace(ColorSpace.Unknown);
                    break;
                default:
                    throw new Exception("Bad input colorspace!");
            }
        }

        /// <summary>
        /// Constructs JPEG quantization tables appropriate for the indicated quality setting.
        /// </summary>
        /// <param name="quality">The quality value is expressed on the 0..100 scale recommended by IJG.</param>
        /// <param name="force_baseline">If <c>true</c>, then the quantization table entries are constrained 
        /// to the range 1..255 for full JPEG baseline compatibility. In the current implementation, 
        /// this only makes a difference for quality settings below 25, and it effectively prevents 
        /// very small/low quality files from being generated. The IJG decoder is capable of reading 
        /// the non-baseline files generated at low quality settings when <c>force_baseline</c> is <c>false</c>,
        /// but other decoders may not be.</param>
        /// <remarks>Note that the exact mapping from quality values to tables may change in future IJG releases 
        /// as more is learned about DCT quantization.</remarks>
        /// <seealso href="ce3f6712-3633-4a58-af07-626a4fba9ae4.htm" target="_self">Compression parameter selection</seealso>
        public void jpeg_set_quality(int quality, bool force_baseline)
        {
            /* Convert user 0-100 rating to percentage scaling */
            quality = jpeg_quality_scaling(quality);

            /* Set up standard quality tables */
            jpeg_set_linear_quality(quality, force_baseline);
        }

        /// <summary>
        /// Same as <see cref="jpeg_set_quality"/> except that the generated tables are the 
        /// sample tables given in the JPEG specification section K.1, multiplied by 
        /// the specified scale factor.
        /// </summary>
        /// <param name="scale_factor">The scale_factor.</param>
        /// <param name="force_baseline">If <c>true</c>, then the quantization table entries are 
        /// constrained to the range 1..255 for full JPEG baseline compatibility. In the current 
        /// implementation, this only makes a difference for quality settings below 25, and it 
        /// effectively prevents very small/low quality files from being generated. The IJG decoder 
        /// is capable of reading the non-baseline files generated at low quality settings when 
        /// <c>force_baseline</c> is <c>false</c>, but other decoders may not be.</param>
        /// <remarks>Note that larger scale factors give lower quality. This entry point is 
        /// useful for conforming to the Adobe PostScript DCT conventions, but we do not 
        /// recommend linear scaling as a user-visible quality scale otherwise.
        /// </remarks>
        /// <seealso href="ce3f6712-3633-4a58-af07-626a4fba9ae4.htm" target="_self">Compression parameter selection</seealso>
        public void jpeg_set_linear_quality(int scale_factor, bool force_baseline)
        {
            /* Set up two quantization tables using the specified scaling */
            jpeg_add_quant_table(0, std_luminance_quant_tbl, scale_factor, force_baseline);
            jpeg_add_quant_table(1, std_chrominance_quant_tbl, scale_factor, force_baseline);
        }

        /// <summary>
        /// Allows an arbitrary quantization table to be created.
        /// </summary>
        /// <param name="which_tbl">Indicates which table slot to fill.</param>
        /// <param name="basic_table">An array of 64 unsigned integers given in normal array order.
        /// These values are multiplied by <c>scale_factor/100</c> and then clamped to the range 1..65535 
        /// (or to 1..255 if <c>force_baseline</c> is <c>true</c>).<br/>
        /// The basic table should be given in JPEG zigzag order.
        /// </param>
        /// <param name="scale_factor">Multiplier for values in <c>basic_table</c>.</param>
        /// <param name="force_baseline">Defines range of values in <c>basic_table</c>. 
        /// If <c>true</c> - 1..255, otherwise - 1..65535.</param>
        /// <seealso href="ce3f6712-3633-4a58-af07-626a4fba9ae4.htm" target="_self">Compression parameter selection</seealso>
        public void jpeg_add_quant_table(int which_tbl, int[] basic_table, int scale_factor, bool force_baseline)
        {
            /* Safety check to ensure start_compress not called yet. */
            if (m_global_state != JpegState.CSTATE_START)
                throw new Exception(String.Format("Improper call to JPEG library in state {0}", (int)m_global_state));

            if (which_tbl < 0 || which_tbl >= JpegConstants.NumberOfQuantTables)
                throw new Exception(String.Format("Bogus DQT index {0}", which_tbl));

            if (m_quant_tbl_ptrs[which_tbl] == null)
                m_quant_tbl_ptrs[which_tbl] = new JpegQuantizationTable();

            for (int i = 0; i < JpegConstants.DCTSize2; i++)
            {
                int temp = (basic_table[i] * scale_factor + 50) / 100;

                /* limit the values to the valid range */
                if (temp <= 0)
                    temp = 1;

                /* max quantizer needed for 12 bits */
                if (temp > 32767)
                    temp = 32767;

                /* limit to baseline range if requested */
                if (force_baseline && temp > 255)
                    temp = 255;

                m_quant_tbl_ptrs[which_tbl].quantval[i] = (short)temp;
            }

            /* Initialize sent_table false so table will be written to JPEG file. */
            m_quant_tbl_ptrs[which_tbl].Sent_table = false;
        }

        /// <summary>
        /// Converts a value on the IJG-recommended quality scale to a linear scaling percentage.
        /// </summary>
        /// <param name="quality">The IJG-recommended quality scale. Should be 0 (terrible) to 100 (very good).</param>
        /// <returns>The linear scaling percentage.</returns>
        /// <seealso href="ce3f6712-3633-4a58-af07-626a4fba9ae4.htm" target="_self">Compression parameter selection</seealso>
        public static int jpeg_quality_scaling(int quality)
        {
            /* Safety limit on quality factor.  Convert 0 to 1 to avoid zero divide. */
            if (quality <= 0)
                quality = 1;

            if (quality > 100)
                quality = 100;

            /* The basic table is used as-is (scaling 100) for a quality of 50.
            * Qualities 50..100 are converted to scaling percentage 200 - 2*Q;
            * note that at Q=100 the scaling is 0, which will cause jpeg_add_quant_table
            * to make all the table entries 1 (hence, minimum quantization loss).
            * Qualities 1..50 are converted to scaling percentage 5000/Q.
            */
            if (quality < 50)
                quality = 5000 / quality;
            else
                quality = 200 - quality * 2;

            return quality;
        }

        /// <summary>
        /// Generates a default scan script for writing a progressive-JPEG file.
        /// </summary>
        /// <remarks>This is the recommended method of creating a progressive file, unless you want 
        /// to make a custom scan sequence. You must ensure that the JPEG color space is 
        /// set correctly before calling this routine.</remarks>
        /// <seealso href="ce3f6712-3633-4a58-af07-626a4fba9ae4.htm" target="_self">Compression parameter selection</seealso>
        public void jpeg_simple_progression()
        {
            /* Safety check to ensure start_compress not called yet. */
            if (m_global_state != JpegState.CSTATE_START)
                throw new Exception(String.Format("Improper call to JPEG library in state {0}", (int)m_global_state));

            /* Figure space needed for script.  Calculation must match code below! */
            int nscans;
            if (m_num_components == 3 && m_jpeg_color_space == ColorSpace.YCbCr)
            {
                /* Custom script for YCbCr color images. */
                nscans = 10;
            }
            else
            {
                /* All-purpose script for other color spaces. */
                if (m_num_components > JpegConstants.MaxComponentsInScan)
                {
                    /* 2 DC + 4 AC scans per component */
                    nscans = 6 * m_num_components;
                }
                else
                {
                    /* 2 DC scans; 4 AC scans per component */
                    nscans = 2 + 4 * m_num_components;
                }
            }

            /* Allocate space for script.
            * We need to put it in the permanent pool in case the application performs
            * multiple compressions without changing the settings.  To avoid a memory
            * leak if jpeg_simple_progression is called repeatedly for the same JPEG
            * object, we try to re-use previously allocated space, and we allocate
            * enough space to handle YCbCr even if initially asked for grayscale.
            */
            if (m_script_space == null || m_script_space_size < nscans)
            {
                m_script_space_size = Math.Max(nscans, 10);
                m_script_space = new JpegScanInfo[m_script_space_size];
                for (int i = 0; i < m_script_space_size; i++)
                    m_script_space[i] = new JpegScanInfo();
            }

            m_scan_info = m_script_space;
            m_num_scans = nscans;

            int scanIndex = 0;
            if (m_num_components == 3 && m_jpeg_color_space == ColorSpace.YCbCr)
            {
                /* Custom script for YCbCr color images. */
                /* Initial DC scan */
                fill_dc_scans(ref scanIndex, m_num_components, 0, 1);

                /* Initial AC scan: get some luma data out in a hurry */
                fill_a_scan(ref scanIndex, 0, 1, 5, 0, 2);

                /* Chroma data is too small to be worth expending many scans on */
                fill_a_scan(ref scanIndex, 2, 1, 63, 0, 1);
                fill_a_scan(ref scanIndex, 1, 1, 63, 0, 1);

                /* Complete spectral selection for luma AC */
                fill_a_scan(ref scanIndex, 0, 6, 63, 0, 2);

                /* Refine next bit of luma AC */
                fill_a_scan(ref scanIndex, 0, 1, 63, 2, 1);

                /* Finish DC successive approximation */
                fill_dc_scans(ref scanIndex, m_num_components, 1, 0);

                /* Finish AC successive approximation */
                fill_a_scan(ref scanIndex, 2, 1, 63, 1, 0);
                fill_a_scan(ref scanIndex, 1, 1, 63, 1, 0);

                /* Luma bottom bit comes last since it's usually largest scan */
                fill_a_scan(ref scanIndex, 0, 1, 63, 1, 0);
            }
            else
            {
                /* All-purpose script for other color spaces. */
                /* Successive approximation first pass */
                fill_dc_scans(ref scanIndex, m_num_components, 0, 1);
                fill_scans(ref scanIndex, m_num_components, 1, 5, 0, 2);
                fill_scans(ref scanIndex, m_num_components, 6, 63, 0, 2);

                /* Successive approximation second pass */
                fill_scans(ref scanIndex, m_num_components, 1, 63, 2, 1);

                /* Successive approximation final pass */
                fill_dc_scans(ref scanIndex, m_num_components, 1, 0);
                fill_scans(ref scanIndex, m_num_components, 1, 63, 1, 0);
            }
        }

        // Main entry points for compression

        /// <summary>
        /// Starts JPEG compression.
        /// </summary>
        /// <param name="write_all_tables">Write or not write all quantization and Huffman tables.</param>
        /// <remarks>Before calling this, all parameters and a data destination must be set up.</remarks>
        /// <seealso cref="JpegCompressor.jpeg_suppress_tables"/>
        /// <seealso cref="JpegCompressor.jpeg_write_tables"/>
        /// <seealso href="07136fd7-d482-48de-b88c-1a4b9658c69e.htm" target="_self">Compression details</seealso>
        public void jpeg_start_compress(bool write_all_tables)
        {
            if (m_global_state != JpegState.CSTATE_START)
                throw new Exception(String.Format("Improper call to JPEG library in state {0}", (int)m_global_state));

            if (write_all_tables)
                jpeg_suppress_tables(false); /* mark all tables to be written */

            /* (Re)initialize destination modules */
            m_dest.init_destination();

            /* Perform master selection of active modules */
            jinit_compress_master();

            /* Set up for the first pass */
            m_master.prepare_for_pass();

            /* Ready for application to drive first pass through jpeg_write_scanlines
            * or jpeg_write_raw_data.
            */
            m_next_scanline = 0;
            m_global_state = (m_raw_data_in ? JpegState.CSTATE_RAW_OK : JpegState.CSTATE_SCANNING);
        }

        /// <summary>
        /// Write some scanlines of data to the JPEG compressor.
        /// </summary>
        /// <param name="scanlines">The array of scanlines.</param>
        /// <param name="num_lines">The number of scanlines for writing.</param>
        /// <returns>The return value will be the number of lines actually written.<br/>
        /// This should be less than the supplied <c>num_lines</c> only in case that 
        /// the data destination module has requested suspension of the compressor, 
        /// or if more than image_height scanlines are passed in.
        /// </returns>
        /// <remarks>We warn about excess calls to <c>jpeg_write_scanlines()</c> since this likely 
        /// signals an application programmer error. However, excess scanlines passed in the last 
        /// valid call are "silently" ignored, so that the application need not adjust <c>num_lines</c>
        /// for end-of-image when using a multiple-scanline buffer.</remarks>
        /// <seealso href="07136fd7-d482-48de-b88c-1a4b9658c69e.htm" target="_self">Compression details</seealso>
        public int jpeg_write_scanlines(byte[][] scanlines, int num_lines)
        {
            if (m_global_state != JpegState.CSTATE_SCANNING)
                throw new Exception(String.Format("Improper call to JPEG library in state {0}", (int)m_global_state));

            /* Give master control module another chance if this is first call to
            * jpeg_write_scanlines.  This lets output of the frame/scan headers be
            * delayed so that application can write COM, etc, markers between
            * jpeg_start_compress and jpeg_write_scanlines.
            */
            if (m_master.MustCallPassStartup())
                m_master.pass_startup();

            /* Ignore any extra scanlines at bottom of image. */
            int rows_left = m_image_height - m_next_scanline;
            if (num_lines > rows_left)
                num_lines = rows_left;

            int row_ctr = 0;
            m_main.process_data(scanlines, ref row_ctr, num_lines);
            m_next_scanline += row_ctr;
            return row_ctr;
        }

        /// <summary>
        /// Alternate entry point to write raw data.
        /// </summary>
        /// <param name="data">The raw data.</param>
        /// <param name="num_lines">The number of scanlines for writing.</param>
        /// <returns>The number of lines actually written.</returns>
        /// <remarks>Processes exactly one iMCU row per call, unless suspended.
        /// Replaces <see cref="jpeg_write_scanlines"/> when writing raw downsampled data.</remarks>
        public int jpeg_write_raw_data(byte[][][] data, int num_lines)
        {
            if (m_global_state != JpegState.CSTATE_RAW_OK)
                throw new Exception(String.Format("Improper call to JPEG library in state {0}", (int)m_global_state));

            if (m_next_scanline >= m_image_height)
            {
                return 0;
            }

            /* Give master control module another chance if this is first call to
            * jpeg_write_raw_data.  This lets output of the frame/scan headers be
            * delayed so that application can write COM, etc, markers between
            * jpeg_start_compress and jpeg_write_raw_data.
            */
            if (m_master.MustCallPassStartup())
                m_master.pass_startup();

            /* Verify that at least one iMCU row has been passed. */
            int lines_per_iMCU_row = m_max_v_samp_factor * JpegConstants.DCTSize;
            if (num_lines < lines_per_iMCU_row)
                throw new Exception("Buffer passed to JPEG library is too small");

            /* Directly compress the row. */
            if (!m_coef.compress_data(data))
            {
                /* If compressor did not consume the whole row, suspend processing. */
                return 0;
            }

            /* OK, we processed one iMCU row. */
            m_next_scanline += lines_per_iMCU_row;
            return lines_per_iMCU_row;
        }

        /// <summary>
        /// Compression initialization for writing raw-coefficient data. Useful for lossless transcoding.
        /// </summary>
        /// <param name="coef_arrays">The virtual arrays need not be filled or even realized at the time 
        /// <c>jpeg_write_coefficients</c> is called; indeed, the virtual arrays typically will be realized 
        /// during this routine and filled afterwards.
        /// </param>
        /// <remarks>Before calling this, all parameters and a data destination must be set up.
        /// Call <see cref="jpeg_finish_compress"/> to actually write the data.
        /// </remarks>
        public void jpeg_write_coefficients(JpegVirtualArray<JpegBlock>[] coef_arrays)
        {
            if (m_global_state != JpegState.CSTATE_START)
                throw new Exception(String.Format("Improper call to JPEG library in state {0}", (int)m_global_state));

            /* Mark all tables to be written */
            jpeg_suppress_tables(false);

            /* (Re)initialize destination modules */
            m_dest.init_destination();

            /* Perform master selection of active modules */
            transencode_master_selection(coef_arrays);

            /* Wait for jpeg_finish_compress() call */
            m_next_scanline = 0;   /* so jpeg_write_marker works */
            m_global_state = JpegState.CSTATE_WRCOEFS;
        }

        // Compression module initialization routines

        /// <summary>
        /// Initialization of a JPEG compression object
        /// </summary>
        private void initialize()
        {
            m_dest = null;
            m_comp_info = null;

            for (int i = 0; i < JpegConstants.NumberOfQuantTables; i++)
                m_quant_tbl_ptrs[i] = null;

            for (int i = 0; i < JpegConstants.NumberOfHuffmanTables; i++)
            {
                m_dc_huff_tbl_ptrs[i] = null;
                m_ac_huff_tbl_ptrs[i] = null;
            }

            m_script_space = null;

            /* OK, I'm ready */
            m_global_state = JpegState.CSTATE_START;
        }

        /// <summary>
        /// Master selection of compression modules.
        /// This is done once at the start of processing an image.  We determine
        /// which modules will be used and give them appropriate initialization calls.
        /// This routine is in charge of selecting the modules to be executed and
        /// making an initialization call to each one.
        /// </summary>
        private void jinit_compress_master()
        {
            /* Initialize master control (includes parameter checking/processing) */
            jinit_c_master_control(false /* full compression */);

            /* Preprocessing */
            if (!m_raw_data_in)
            {
                m_cconvert = new ColorConverter(this);
                m_downsample = new JpegDownsampler(this);
                m_prep = new JpegCompressorPrepController(this);
            }

            /* Forward DCT */
            m_fdct = new JpegFowardDCT(this);

            /* Entropy encoding: only Huffman coding supported. */
            if (m_progressive_mode)
                m_entropy = new ProgressiveHuffmanEncoder(this);
            else
                m_entropy = new HuffEntropyEncoder(this);

            /* Need a full-image coefficient buffer in any multi-pass mode. */
            m_coef = new CoefControllerImpl(this, (bool)(m_num_scans > 1 || m_optimize_coding));
            jinit_c_main_controller(false /* never need full buffer here */);
            m_marker = new JpegMarkerWriter(this);

            /* Write the datastream header (SOI) immediately.
            * Frame and scan headers are postponed till later.
            * This lets application insert special markers after the SOI.
            */
            m_marker.write_file_header();
        }

        /// <summary>
        /// Initialize master compression control.
        /// </summary>
        private void jinit_c_master_control(bool transcode_only)
        {
            /* Validate parameters, determine derived values */
            initial_setup();

            if (m_scan_info != null)
            {
                validate_script();
            }
            else
            {
                m_progressive_mode = false;
                m_num_scans = 1;
            }

            if (m_progressive_mode)    /*  TEMPORARY HACK ??? */
                m_optimize_coding = true; /* assume default tables no good for progressive mode */

            m_master = new JpegCompressorMaster(this, transcode_only);
        }

        /// <summary>
        /// Initialize main buffer controller.
        /// </summary>
        private void jinit_c_main_controller(bool need_full_buffer)
        {
            /* We don't need to create a buffer in raw-data mode. */
            if (m_raw_data_in)
                return;

            /* Create the buffer.  It holds downsampled data, so each component
            * may be of a different size.
            */
            if (need_full_buffer)
                throw new Exception("Bogus buffer control mode");
            else
                m_main = new JpegCompressorMainController(this);
        }

        /// <summary>
        /// Master selection of compression modules for transcoding.
        /// </summary>
        private void transencode_master_selection(JpegVirtualArray<JpegBlock>[] coef_arrays)
        {
            /* Although we don't actually use input_components for transcoding, 
             * jcmaster.c's initial_setup will complain if input_components is 0.
             */
            m_input_components = 1;

            /* Initialize master control (includes parameter checking/processing) */
            jinit_c_master_control(true /* transcode only */);

            /* Entropy encoding: only Huffman coding supported. */
            if (m_progressive_mode)
                m_entropy = new ProgressiveHuffmanEncoder(this);
            else
                m_entropy = new HuffEntropyEncoder(this);

            /* We need a special coefficient buffer controller. */
            m_coef = new TransCoefControllerImpl(this, coef_arrays);
            m_marker = new JpegMarkerWriter(this);

            /* Write the datastream header (SOI, JFIF) immediately.
            * Frame and scan headers are postponed till later.
            * This lets application insert special markers after the SOI.
            */
            m_marker.write_file_header();
        }

        /// <summary>
        /// Do computations that are needed before master selection phase
        /// </summary>
        private void initial_setup()
        {
            /* Sanity check on image dimensions */
            if (m_image_height <= 0 || m_image_width <= 0 || m_num_components <= 0 || m_input_components <= 0)
                throw new Exception("Empty JPEG image (DNL not supported)");

            /* Make sure image isn't bigger than I can handle */
            if (m_image_height > JpegConstants.JpegMaxDimention || m_image_width > JpegConstants.JpegMaxDimention)
                throw new Exception(String.Format("Maximum supported image dimension is {0} pixels", (int)JpegConstants.JpegMaxDimention));

            /* Width of an input scanline must be representable as int. */
            long samplesperrow = m_image_width * m_input_components;
            int jd_samplesperrow = (int)samplesperrow;
            if ((long)jd_samplesperrow != samplesperrow)
                throw new Exception("Image too wide for this implementation");

            /* For now, precision must match compiled-in value... */
            if (m_data_precision != JpegConstants.BitsInSample)
                throw new Exception(String.Format("Unsupported JPEG data precision {0}", m_data_precision));

            /* Check that number of components won't exceed internal array sizes */
            if (m_num_components > JpegConstants.MaxComponents)
                throw new Exception(String.Format("Too many color components: {0}, max {1}", m_num_components, JpegConstants.MaxComponents));

            /* Compute maximum sampling factors; check factor validity */
            m_max_h_samp_factor = 1;
            m_max_v_samp_factor = 1;
            for (int ci = 0; ci < m_num_components; ci++)
            {
                if (m_comp_info[ci].H_samp_factor <= 0 || m_comp_info[ci].H_samp_factor > JpegConstants.MaxSamplingFactor ||
                    m_comp_info[ci].V_samp_factor <= 0 || m_comp_info[ci].V_samp_factor > JpegConstants.MaxSamplingFactor)
                {
                    throw new Exception("Bogus sampling factors");
                }

                m_max_h_samp_factor = Math.Max(m_max_h_samp_factor, m_comp_info[ci].H_samp_factor);
                m_max_v_samp_factor = Math.Max(m_max_v_samp_factor, m_comp_info[ci].V_samp_factor);
            }

            /* Compute dimensions of components */
            for (int ci = 0; ci < m_num_components; ci++)
            {
                /* Fill in the correct component_index value; don't rely on application */
                m_comp_info[ci].Component_index = ci;

                /* For compression, we never do DCT scaling. */
                m_comp_info[ci].DCT_scaled_size = JpegConstants.DCTSize;

                /* Size in DCT blocks */
                m_comp_info[ci].Width_in_blocks = JpegUtils.jdiv_round_up(
                    m_image_width * m_comp_info[ci].H_samp_factor, m_max_h_samp_factor * JpegConstants.DCTSize);
                m_comp_info[ci].height_in_blocks = JpegUtils.jdiv_round_up(
                    m_image_height * m_comp_info[ci].V_samp_factor, m_max_v_samp_factor * JpegConstants.DCTSize);

                /* Size in samples */
                m_comp_info[ci].downsampled_width = JpegUtils.jdiv_round_up(
                    m_image_width * m_comp_info[ci].H_samp_factor, m_max_h_samp_factor);
                m_comp_info[ci].downsampled_height = JpegUtils.jdiv_round_up(
                    m_image_height * m_comp_info[ci].V_samp_factor, m_max_v_samp_factor);

                /* Mark component needed (this flag isn't actually used for compression) */
                m_comp_info[ci].component_needed = true;
            }

            /* Compute number of fully interleaved MCU rows (number of times that
            * main controller will call coefficient controller).
            */
            m_total_iMCU_rows = JpegUtils.jdiv_round_up(m_image_height, m_max_v_samp_factor * JpegConstants.DCTSize);
        }

        /// <summary>
        /// Verify that the scan script in scan_info[] is valid; 
        /// also determine whether it uses progressive JPEG, and set progressive_mode.
        /// </summary>
        private void validate_script()
        {
            if (m_num_scans <= 0)
                throw new Exception(String.Format("Invalid scan script at entry {0}", 0));

            /* For sequential JPEG, all scans must have Ss=0, Se=DCTSize2-1;
            * for progressive JPEG, no scan can have this.
            */
            int[][] last_bitpos = new int[JpegConstants.MaxComponents][];
            for (int i = 0; i < JpegConstants.MaxComponents; i++)
                last_bitpos[i] = new int[JpegConstants.DCTSize2];

            bool[] component_sent = new bool[JpegConstants.MaxComponents];

            /* -1 until that coefficient has been seen; then last Al for it */
            if (m_scan_info[0].Ss != 0 || m_scan_info[0].Se != JpegConstants.DCTSize2 - 1)
            {
                m_progressive_mode = true;
                for (int ci = 0; ci < m_num_components; ci++)
                {
                    for (int coefi = 0; coefi < JpegConstants.DCTSize2; coefi++)
                        last_bitpos[ci][coefi] = -1;
                }
            }
            else
            {
                m_progressive_mode = false;
                for (int ci = 0; ci < m_num_components; ci++)
                    component_sent[ci] = false;
            }

            for (int scanno = 1; scanno <= m_num_scans; scanno++)
            {
                JpegScanInfo scanInfo = m_scan_info[scanno - 1];

                /* Validate component indexes */
                int ncomps = scanInfo.comps_in_scan;
                if (ncomps <= 0 || ncomps > JpegConstants.MaxComponentsInScan)
                    throw new Exception(String.Format("Too many color components: {0}, max {1}", ncomps, JpegConstants.MaxComponentsInScan));

                for (int ci = 0; ci < ncomps; ci++)
                {
                    int thisi = scanInfo.component_index[ci];
                    if (thisi < 0 || thisi >= m_num_components)
                        throw new Exception(String.Format("Invalid scan script at entry {0}", scanno));

                    /* Components must appear in SOF order within each scan */
                    if (ci > 0 && thisi <= scanInfo.component_index[ci - 1])
                        throw new Exception(String.Format("Invalid scan script at entry {0}", scanno));
                }

                /* Validate progression parameters */
                int Ss = scanInfo.Ss;
                int Se = scanInfo.Se;
                int Ah = scanInfo.Ah;
                int Al = scanInfo.Al;
                if (m_progressive_mode)
                {
                    /* The JPEG spec simply gives the ranges 0..13 for Ah and Al, but that
                    * seems wrong: the upper bound ought to depend on data precision.
                    * Perhaps they really meant 0..N+1 for N-bit precision.
                    * Here we allow 0..10 for 8-bit data; Al larger than 10 results in
                    * out-of-range reconstructed DC values during the first DC scan,
                    * which might cause problems for some decoders.
                    */
                    const int MAX_AH_AL = 10;
                    if (Ss < 0 || Ss >= JpegConstants.DCTSize2 || Se < Ss || Se >= JpegConstants.DCTSize2 ||
                        Ah < 0 || Ah > MAX_AH_AL || Al < 0 || Al > MAX_AH_AL)
                    {
                        throw new Exception(String.Format("Invalid progressive parameters at scan script entry {0}", scanno));
                    }

                    if (Ss == 0)
                    {
                        if (Se != 0)        /* DC and AC together not OK */
                            throw new Exception(String.Format("Invalid progressive parameters at scan script entry {0}", scanno));
                    }
                    else
                    {
                        if (ncomps != 1)    /* AC scans must be for only one component */
                            throw new Exception(String.Format("Invalid progressive parameters at scan script entry {0}", scanno));
                    }

                    for (int ci = 0; ci < ncomps; ci++)
                    {
                        int lastBitComponentIndex = scanInfo.component_index[ci];
                        if (Ss != 0 && last_bitpos[lastBitComponentIndex][0] < 0) /* AC without prior DC scan */
                            throw new Exception(String.Format("Invalid progressive parameters at scan script entry {0}", scanno));

                        for (int coefi = Ss; coefi <= Se; coefi++)
                        {
                            if (last_bitpos[lastBitComponentIndex][coefi] < 0)
                            {
                                /* first scan of this coefficient */
                                if (Ah != 0)
                                    throw new Exception(String.Format("Invalid progressive parameters at scan script entry {0}", scanno));
                            }
                            else
                            {
                                /* not first scan */
                                if (Ah != last_bitpos[lastBitComponentIndex][coefi] || Al != Ah - 1)
                                    throw new Exception(String.Format("Invalid progressive parameters at scan script entry {0}", scanno));
                            }

                            last_bitpos[lastBitComponentIndex][coefi] = Al;
                        }
                    }
                }
                else
                {
                    /* For sequential JPEG, all progression parameters must be these: */
                    if (Ss != 0 || Se != JpegConstants.DCTSize2 - 1 || Ah != 0 || Al != 0)
                        throw new Exception(String.Format("Invalid progressive parameters at scan script entry {0}", scanno));

                    /* Make sure components are not sent twice */
                    for (int ci = 0; ci < ncomps; ci++)
                    {
                        int thisi = scanInfo.component_index[ci];
                        if (component_sent[thisi])
                            throw new Exception(String.Format("Invalid scan script at entry {0}", scanno));

                        component_sent[thisi] = true;
                    }
                }
            }

            /* Now verify that everything got sent. */
            if (m_progressive_mode)
            {
                /* For progressive mode, we only check that at least some DC data
                * got sent for each component; the spec does not require that all bits
                * of all coefficients be transmitted.  Would it be wiser to enforce
                * transmission of all coefficient bits??
                */
                for (int ci = 0; ci < m_num_components; ci++)
                {
                    if (last_bitpos[ci][0] < 0)
                        throw new Exception("Scan script does not transmit all data");
                }
            }
            else
            {
                for (int ci = 0; ci < m_num_components; ci++)
                {
                    if (!component_sent[ci])
                        throw new Exception("Scan script does not transmit all data");
                }
            }
        }

        // Huffman table setup routines

        /// <summary>
        /// Set up the standard Huffman tables (cf. JPEG standard section K.3)
        /// 
        /// IMPORTANT: these are only valid for 8-bit data precision!
        /// </summary>
        private void std_huff_tables()
        {
            add_huff_table(ref m_dc_huff_tbl_ptrs[0], bits_dc_luminance, val_dc_luminance);
            add_huff_table(ref m_ac_huff_tbl_ptrs[0], bits_ac_luminance, val_ac_luminance);
            add_huff_table(ref m_dc_huff_tbl_ptrs[1], bits_dc_chrominance, val_dc_chrominance);
            add_huff_table(ref m_ac_huff_tbl_ptrs[1], bits_ac_chrominance, val_ac_chrominance);
        }

        /// <summary>
        /// Define a Huffman table
        /// </summary>
        private void add_huff_table(ref JpegHuffmanTable htblptr, byte[] bits, byte[] val)
        {
            if (htblptr == null)
                htblptr = new JpegHuffmanTable();

            /* Copy the number-of-symbols-of-each-code-length counts */
            Buffer.BlockCopy(bits, 0, htblptr.Bits, 0, htblptr.Bits.Length);

            /* Validate the counts.  We do this here mainly so we can copy the right
            * number of symbols from the val[] array, without risking marching off
            * the end of memory. HuffEntropyEncoder will do a more thorough test later.
            */
            int nsymbols = 0;
            for (int len = 1; len <= 16; len++)
                nsymbols += bits[len];

            if (nsymbols < 1 || nsymbols > 256)
                throw new Exception("Bogus Huffman table definition");

            Buffer.BlockCopy(val, 0, htblptr.Huffval, 0, nsymbols);

            /* Initialize sent_table false so table will be written to JPEG file. */
            htblptr.Sent_table = false;
        }

        /// <summary>
        /// Support routine: generate one scan for specified component
        /// </summary>
        private void fill_a_scan(ref int scanIndex, int ci, int Ss, int Se, int Ah, int Al)
        {
            m_script_space[scanIndex].comps_in_scan = 1;
            m_script_space[scanIndex].component_index[0] = ci;
            m_script_space[scanIndex].Ss = Ss;
            m_script_space[scanIndex].Se = Se;
            m_script_space[scanIndex].Ah = Ah;
            m_script_space[scanIndex].Al = Al;
            scanIndex++;
        }

        /// <summary>
        /// Support routine: generate interleaved DC scan if possible, else N scans
        /// </summary>
        private void fill_dc_scans(ref int scanIndex, int ncomps, int Ah, int Al)
        {
            if (ncomps <= JpegConstants.MaxComponentsInScan)
            {
                /* Single interleaved DC scan */
                m_script_space[scanIndex].comps_in_scan = ncomps;
                for (int ci = 0; ci < ncomps; ci++)
                    m_script_space[scanIndex].component_index[ci] = ci;

                m_script_space[scanIndex].Ss = 0;
                m_script_space[scanIndex].Se = 0;
                m_script_space[scanIndex].Ah = Ah;
                m_script_space[scanIndex].Al = Al;
                scanIndex++;
            }
            else
            {
                /* Non-interleaved DC scan for each component */
                fill_scans(ref scanIndex, ncomps, 0, 0, Ah, Al);
            }
        }

        /// <summary>
        /// Support routine: generate one scan for each component
        /// </summary>
        private void fill_scans(ref int scanIndex, int ncomps, int Ss, int Se, int Ah, int Al)
        {
            for (int ci = 0; ci < ncomps; ci++)
            {
                m_script_space[scanIndex].comps_in_scan = 1;
                m_script_space[scanIndex].component_index[0] = ci;
                m_script_space[scanIndex].Ss = Ss;
                m_script_space[scanIndex].Se = Se;
                m_script_space[scanIndex].Ah = Ah;
                m_script_space[scanIndex].Al = Al;
                scanIndex++;
            }
        }

        private void jpeg_set_colorspace_SET_COMP(int index, int id, int hsamp, int vsamp, int quant, int dctbl, int actbl)
        {
            m_comp_info[index].Component_id = id;
            m_comp_info[index].H_samp_factor = hsamp;
            m_comp_info[index].V_samp_factor = vsamp;
            m_comp_info[index].Quant_tbl_no = quant;
            m_comp_info[index].Dc_tbl_no = dctbl;
            m_comp_info[index].Ac_tbl_no = actbl;
        }
    }
    #endregion

    #region JpegCompressorCoefController
    /// <summary>
    /// Coefficient buffer control
    /// </summary>
    interface JpegCompressorCoefController
    {
        void start_pass(BufferMode pass_mode);
        bool compress_data(byte[][][] input_buf);
    }
    #endregion

    #region JpegCompressorMainController
    /// <summary>
    /// Main buffer control (downsampled-data buffer)
    /// </summary>
    class JpegCompressorMainController
    {
        private JpegCompressor m_cinfo;

        private int m_cur_iMCU_row;    /* number of current iMCU row */
        private int m_rowgroup_ctr;    /* counts row groups received in iMCU row */
        private bool m_suspended;     /* remember if we suspended output */

        /* If using just a strip buffer, this points to the entire set of buffers
        * (we allocate one for each component).  In the full-image case, this
        * points to the currently accessible strips of the virtual arrays.
        */
        private byte[][][] m_buffer = new byte[JpegConstants.MaxComponents][][];

        public JpegCompressorMainController(JpegCompressor cinfo)
        {
            m_cinfo = cinfo;

            /* Allocate a strip buffer for each component */
            for (int ci = 0; ci < cinfo.m_num_components; ci++)
            {
                m_buffer[ci] = JpegCommonBase.AllocJpegSamples(
                    cinfo.Component_info[ci].Width_in_blocks * JpegConstants.DCTSize,
                    cinfo.Component_info[ci].V_samp_factor * JpegConstants.DCTSize);
            }
        }

        // Initialize for a processing pass.
        public void start_pass(BufferMode pass_mode)
        {
            /* Do nothing in raw-data mode. */
            if (m_cinfo.m_raw_data_in)
                return;

            m_cur_iMCU_row = 0; /* initialize counters */
            m_rowgroup_ctr = 0;
            m_suspended = false;

            if (pass_mode != BufferMode.PassThru)
                throw new Exception("Bogus buffer control mode!");
        }

        /// <summary>
        /// Process some data.
        /// This routine handles the simple pass-through mode,
        /// where we have only a strip buffer.
        /// </summary>
        public void process_data(byte[][] input_buf, ref int in_row_ctr, int in_rows_avail)
        {
            while (m_cur_iMCU_row < m_cinfo.m_total_iMCU_rows)
            {
                /* Read input data if we haven't filled the main buffer yet */
                if (m_rowgroup_ctr < JpegConstants.DCTSize)
                    m_cinfo.m_prep.pre_process_data(input_buf, ref in_row_ctr, in_rows_avail, m_buffer, ref m_rowgroup_ctr, JpegConstants.DCTSize);

                /* If we don't have a full iMCU row buffered, return to application for
                 * more data.  Note that preprocessor will always pad to fill the iMCU row
                 * at the bottom of the image.
                 */
                if (m_rowgroup_ctr != JpegConstants.DCTSize)
                    return;

                /* Send the completed row to the compressor */
                if (!m_cinfo.m_coef.compress_data(m_buffer))
                {
                    /* If compressor did not consume the whole row, then we must need to
                     * suspend processing and return to the application.  In this situation
                     * we pretend we didn't yet consume the last input row; otherwise, if
                     * it happened to be the last row of the image, the application would
                     * think we were done.
                     */
                    if (!m_suspended)
                    {
                        in_row_ctr--;
                        m_suspended = true;
                    }

                    return;
                }

                /* We did finish the row.  Undo our little suspension hack if a previous
                 * call suspended; then mark the main buffer empty.
                 */
                if (m_suspended)
                {
                    in_row_ctr++;
                    m_suspended = false;
                }

                m_rowgroup_ctr = 0;
                m_cur_iMCU_row++;
            }
        }
    }
    #endregion

    #region JpegCompressorMaster
    /// <summary>
    /// Master control module
    /// </summary>
    class JpegCompressorMaster
    {
        private enum c_pass_type
        {
            main_pass,      /* input data, also do first output step */
            huff_opt_pass,  /* Huffman code optimization pass */
            output_pass     /* data output pass */
        }

        private JpegCompressor m_cinfo;

        private bool m_call_pass_startup; /* True if pass_startup must be called */
        private bool m_is_last_pass;      /* True during last pass */

        private c_pass_type m_pass_type;  /* the type of the current pass */

        private int m_pass_number;        /* # of passes completed */
        private int m_total_passes;       /* total # of passes needed */

        private int m_scan_number;        /* current index in scan_info[] */

        public JpegCompressorMaster(JpegCompressor cinfo, bool transcode_only)
        {
            m_cinfo = cinfo;

            if (transcode_only)
            {
                /* no main pass in transcoding */
                if (cinfo.m_optimize_coding)
                    m_pass_type = c_pass_type.huff_opt_pass;
                else
                    m_pass_type = c_pass_type.output_pass;
            }
            else
            {
                /* for normal compression, first pass is always this type: */
                m_pass_type = c_pass_type.main_pass;
            }

            if (cinfo.m_optimize_coding)
                m_total_passes = cinfo.m_num_scans * 2;
            else
                m_total_passes = cinfo.m_num_scans;
        }

        /// <summary>
        /// Per-pass setup.
        /// 
        /// This is called at the beginning of each pass.  We determine which 
        /// modules will be active during this pass and give them appropriate 
        /// start_pass calls. 
        /// We also set is_last_pass to indicate whether any more passes will 
        /// be required.
        /// </summary>
        public void prepare_for_pass()
        {
            switch (m_pass_type)
            {
                case c_pass_type.main_pass:
                    prepare_for_main_pass();
                    break;
                case c_pass_type.huff_opt_pass:
                    if (!prepare_for_huff_opt_pass())
                        break;
                    prepare_for_output_pass();
                    break;
                case c_pass_type.output_pass:
                    prepare_for_output_pass();
                    break;
            }

            m_is_last_pass = (m_pass_number == m_total_passes - 1);
        }

        /// <summary>
        /// Special start-of-pass hook.
        /// 
        /// This is called by jpeg_write_scanlines if call_pass_startup is true.
        /// In single-pass processing, we need this hook because we don't want to
        /// write frame/scan headers during jpeg_start_compress; we want to let the
        /// application write COM markers etc. between jpeg_start_compress and the
        /// jpeg_write_scanlines loop.
        /// In multi-pass processing, this routine is not used.
        /// </summary>
        public void pass_startup()
        {
            m_cinfo.m_master.m_call_pass_startup = false; /* reset flag so call only once */

            m_cinfo.m_marker.write_frame_header();
            m_cinfo.m_marker.write_scan_header();
        }

        /// <summary>
        /// Finish up at end of pass.
        /// </summary>
        public void finish_pass()
        {
            /* The entropy coder always needs an end-of-pass call,
            * either to analyze statistics or to flush its output buffer.
            */
            m_cinfo.m_entropy.finish_pass();

            /* Update state for next pass */
            switch (m_pass_type)
            {
                case c_pass_type.main_pass:
                    /* next pass is either output of scan 0 (after optimization)
                    * or output of scan 1 (if no optimization).
                    */
                    m_pass_type = c_pass_type.output_pass;
                    if (!m_cinfo.m_optimize_coding)
                        m_scan_number++;
                    break;
                case c_pass_type.huff_opt_pass:
                    /* next pass is always output of current scan */
                    m_pass_type = c_pass_type.output_pass;
                    break;
                case c_pass_type.output_pass:
                    /* next pass is either optimization or output of next scan */
                    if (m_cinfo.m_optimize_coding)
                        m_pass_type = c_pass_type.huff_opt_pass;
                    m_scan_number++;
                    break;
            }

            m_pass_number++;
        }

        public bool IsLastPass()
        {
            return m_is_last_pass;
        }

        public bool MustCallPassStartup()
        {
            return m_call_pass_startup;
        }

        private void prepare_for_main_pass()
        {
            /* Initial pass: will collect input data, and do either Huffman
            * optimization or data output for the first scan.
            */
            select_scan_parameters();
            per_scan_setup();

            if (!m_cinfo.m_raw_data_in)
            {
                m_cinfo.m_cconvert.start_pass();
                m_cinfo.m_prep.start_pass(BufferMode.PassThru);
            }

            m_cinfo.m_fdct.start_pass();
            m_cinfo.m_entropy.start_pass(m_cinfo.m_optimize_coding);
            m_cinfo.m_coef.start_pass((m_total_passes > 1 ? BufferMode.SaveAndPass : BufferMode.PassThru));
            m_cinfo.m_main.start_pass(BufferMode.PassThru);

            if (m_cinfo.m_optimize_coding)
            {
                /* No immediate data output; postpone writing frame/scan headers */
                m_call_pass_startup = false;
            }
            else
            {
                /* Will write frame/scan headers at first jpeg_write_scanlines call */
                m_call_pass_startup = true;
            }
        }

        private bool prepare_for_huff_opt_pass()
        {
            /* Do Huffman optimization for a scan after the first one. */
            select_scan_parameters();
            per_scan_setup();

            if (m_cinfo.m_Ss != 0 || m_cinfo.m_Ah == 0)
            {
                m_cinfo.m_entropy.start_pass(true);
                m_cinfo.m_coef.start_pass(BufferMode.CrankDest);
                m_call_pass_startup = false;
                return false;
            }

            /* Special case: Huffman DC refinement scans need no Huffman table
            * and therefore we can skip the optimization pass for them.
            */
            m_pass_type = c_pass_type.output_pass;
            m_pass_number++;
            return true;
        }

        private void prepare_for_output_pass()
        {
            /* Do a data-output pass. */
            /* We need not repeat per-scan setup if prior optimization pass did it. */
            if (!m_cinfo.m_optimize_coding)
            {
                select_scan_parameters();
                per_scan_setup();
            }

            m_cinfo.m_entropy.start_pass(false);
            m_cinfo.m_coef.start_pass(BufferMode.CrankDest);

            /* We emit frame/scan headers now */
            if (m_scan_number == 0)
                m_cinfo.m_marker.write_frame_header();

            m_cinfo.m_marker.write_scan_header();
            m_call_pass_startup = false;
        }

        // Set up the scan parameters for the current scan
        private void select_scan_parameters()
        {
            if (m_cinfo.m_scan_info != null)
            {
                /* Prepare for current scan --- the script is already validated */
                JpegScanInfo scanInfo = m_cinfo.m_scan_info[m_scan_number];

                m_cinfo.m_comps_in_scan = scanInfo.comps_in_scan;
                for (int ci = 0; ci < scanInfo.comps_in_scan; ci++)
                    m_cinfo.m_cur_comp_info[ci] = scanInfo.component_index[ci];

                m_cinfo.m_Ss = scanInfo.Ss;
                m_cinfo.m_Se = scanInfo.Se;
                m_cinfo.m_Ah = scanInfo.Ah;
                m_cinfo.m_Al = scanInfo.Al;
            }
            else
            {
                /* Prepare for single sequential-JPEG scan containing all components */
                if (m_cinfo.m_num_components > JpegConstants.MaxComponentsInScan)
                    throw new Exception(String.Format("Too many color components: {0}, max {1}", m_cinfo.m_num_components, JpegConstants.MaxComponentsInScan));

                m_cinfo.m_comps_in_scan = m_cinfo.m_num_components;
                for (int ci = 0; ci < m_cinfo.m_num_components; ci++)
                    m_cinfo.m_cur_comp_info[ci] = ci;

                m_cinfo.m_Ss = 0;
                m_cinfo.m_Se = JpegConstants.DCTSize2 - 1;
                m_cinfo.m_Ah = 0;
                m_cinfo.m_Al = 0;
            }
        }

        /// <summary>
        /// Do computations that are needed before processing a JPEG scan
        /// cinfo.comps_in_scan and cinfo.cur_comp_info[] are already set
        /// </summary>
        private void per_scan_setup()
        {
            if (m_cinfo.m_comps_in_scan == 1)
            {
                /* Non-interleaved (single-component) scan */
                int compIndex = m_cinfo.m_cur_comp_info[0];

                /* Overall image size in MCUs */
                m_cinfo.m_MCUs_per_row = m_cinfo.Component_info[compIndex].Width_in_blocks;
                m_cinfo.m_MCU_rows_in_scan = m_cinfo.Component_info[compIndex].height_in_blocks;

                /* For non-interleaved scan, always one block per MCU */
                m_cinfo.Component_info[compIndex].MCU_width = 1;
                m_cinfo.Component_info[compIndex].MCU_height = 1;
                m_cinfo.Component_info[compIndex].MCU_blocks = 1;
                m_cinfo.Component_info[compIndex].MCU_sample_width = JpegConstants.DCTSize;
                m_cinfo.Component_info[compIndex].last_col_width = 1;

                /* For non-interleaved scans, it is convenient to define last_row_height
                * as the number of block rows present in the last iMCU row.
                */
                int tmp = m_cinfo.Component_info[compIndex].height_in_blocks % m_cinfo.Component_info[compIndex].V_samp_factor;
                if (tmp == 0)
                    tmp = m_cinfo.Component_info[compIndex].V_samp_factor;
                m_cinfo.Component_info[compIndex].last_row_height = tmp;

                /* Prepare array describing MCU composition */
                m_cinfo.m_blocks_in_MCU = 1;
                m_cinfo.m_MCU_membership[0] = 0;
            }
            else
            {
                /* Interleaved (multi-component) scan */
                if (m_cinfo.m_comps_in_scan <= 0 || m_cinfo.m_comps_in_scan > JpegConstants.MaxComponentsInScan)
                    throw new Exception(String.Format("Too many color components: {0}, max {1}", m_cinfo.m_comps_in_scan, JpegConstants.MaxComponentsInScan));

                /* Overall image size in MCUs */
                m_cinfo.m_MCUs_per_row = JpegUtils.jdiv_round_up(
                    m_cinfo.m_image_width, m_cinfo.m_max_h_samp_factor * JpegConstants.DCTSize);

                m_cinfo.m_MCU_rows_in_scan = JpegUtils.jdiv_round_up(m_cinfo.m_image_height,
                    m_cinfo.m_max_v_samp_factor * JpegConstants.DCTSize);

                m_cinfo.m_blocks_in_MCU = 0;

                for (int ci = 0; ci < m_cinfo.m_comps_in_scan; ci++)
                {
                    int compIndex = m_cinfo.m_cur_comp_info[ci];

                    /* Sampling factors give # of blocks of component in each MCU */
                    m_cinfo.Component_info[compIndex].MCU_width = m_cinfo.Component_info[compIndex].H_samp_factor;
                    m_cinfo.Component_info[compIndex].MCU_height = m_cinfo.Component_info[compIndex].V_samp_factor;
                    m_cinfo.Component_info[compIndex].MCU_blocks = m_cinfo.Component_info[compIndex].MCU_width * m_cinfo.Component_info[compIndex].MCU_height;
                    m_cinfo.Component_info[compIndex].MCU_sample_width = m_cinfo.Component_info[compIndex].MCU_width * JpegConstants.DCTSize;

                    /* Figure number of non-dummy blocks in last MCU column & row */
                    int tmp = m_cinfo.Component_info[compIndex].Width_in_blocks % m_cinfo.Component_info[compIndex].MCU_width;
                    if (tmp == 0)
                        tmp = m_cinfo.Component_info[compIndex].MCU_width;
                    m_cinfo.Component_info[compIndex].last_col_width = tmp;

                    tmp = m_cinfo.Component_info[compIndex].height_in_blocks % m_cinfo.Component_info[compIndex].MCU_height;
                    if (tmp == 0)
                        tmp = m_cinfo.Component_info[compIndex].MCU_height;
                    m_cinfo.Component_info[compIndex].last_row_height = tmp;

                    /* Prepare array describing MCU composition */
                    int mcublks = m_cinfo.Component_info[compIndex].MCU_blocks;
                    if (m_cinfo.m_blocks_in_MCU + mcublks > JpegConstants.CompressorMaxBlocksInMCU)
                        throw new Exception("Sampling factors too large for interleaved scan");

                    while (mcublks-- > 0)
                        m_cinfo.m_MCU_membership[m_cinfo.m_blocks_in_MCU++] = ci;
                }
            }

            /* Convert restart specified in rows to actual MCU count. */
            /* Note that count must fit in 16 bits, so we provide limiting. */
            if (m_cinfo.m_restart_in_rows > 0)
            {
                int nominal = m_cinfo.m_restart_in_rows * m_cinfo.m_MCUs_per_row;
                m_cinfo.m_restart_interval = Math.Min(nominal, 65535);
            }
        }
    }
    #endregion

    #region JpegCompressorPrepController
    /// <summary>
    /// Compression preprocessing (downsampling input buffer control).
    /// 
    /// For the simple (no-context-row) case, we just need to buffer one
    /// row group's worth of pixels for the downsampling step.  At the bottom of
    /// the image, we pad to a full row group by replicating the last pixel row.
    /// The downsampler's last output row is then replicated if needed to pad
    /// out to a full iMCU row.
    /// 
    /// When providing context rows, we must buffer three row groups' worth of
    /// pixels.  Three row groups are physically allocated, but the row pointer
    /// arrays are made five row groups high, with the extra pointers above and
    /// below "wrapping around" to point to the last and first real row groups.
    /// This allows the downsampler to access the proper context rows.
    /// At the top and bottom of the image, we create dummy context rows by
    /// copying the first or last real pixel row.  This copying could be avoided
    /// by pointer hacking as is done in jdmainct.c, but it doesn't seem worth the
    /// trouble on the compression side.
    /// </summary>
    class JpegCompressorPrepController
    {
        private JpegCompressor m_cinfo;

        /* Downsampling input buffer.  This buffer holds color-converted data
        * until we have enough to do a downsample step.
        */
        private byte[][][] m_color_buf = new byte[JpegConstants.MaxComponents][][];
        private int m_colorBufRowsOffset;

        private int m_rows_to_go;  /* counts rows remaining in source image */
        private int m_next_buf_row;       /* index of next row to store in color_buf */

        private int m_this_row_group;     /* starting row index of group to process */
        private int m_next_buf_stop;      /* downsample when we reach this index */

        public JpegCompressorPrepController(JpegCompressor cinfo)
        {
            m_cinfo = cinfo;

            /* Allocate the color conversion buffer.
            * We make the buffer wide enough to allow the downsampler to edge-expand
            * horizontally within the buffer, if it so chooses.
            */
            if (cinfo.m_downsample.NeedContextRows())
            {
                /* Set up to provide context rows */
                create_context_buffer();
            }
            else
            {
                /* No context, just make it tall enough for one row group */
                for (int ci = 0; ci < cinfo.m_num_components; ci++)
                {
                    m_colorBufRowsOffset = 0;
                    m_color_buf[ci] = JpegCompressor.AllocJpegSamples(
                        (cinfo.Component_info[ci].Width_in_blocks * JpegConstants.DCTSize * cinfo.m_max_h_samp_factor) / cinfo.Component_info[ci].H_samp_factor,
                        cinfo.m_max_v_samp_factor);
                }
            }
        }

        /// <summary>
        /// Initialize for a processing pass.
        /// </summary>
        public void start_pass(BufferMode pass_mode)
        {
            if (pass_mode != BufferMode.PassThru)
                throw new Exception("Bogus buffer control mode!");

            /* Initialize total-height counter for detecting bottom of image */
            m_rows_to_go = m_cinfo.m_image_height;

            /* Mark the conversion buffer empty */
            m_next_buf_row = 0;

            /* Preset additional state variables for context mode.
             * These aren't used in non-context mode, so we needn't test which mode.
             */
            m_this_row_group = 0;

            /* Set next_buf_stop to stop after two row groups have been read in. */
            m_next_buf_stop = 2 * m_cinfo.m_max_v_samp_factor;
        }

        public void pre_process_data(byte[][] input_buf, ref int in_row_ctr, int in_rows_avail, byte[][][] output_buf, ref int out_row_group_ctr, int out_row_groups_avail)
        {
            if (m_cinfo.m_downsample.NeedContextRows())
                pre_process_context(input_buf, ref in_row_ctr, in_rows_avail, output_buf, ref out_row_group_ctr, out_row_groups_avail);
            else
                pre_process_WithoutContext(input_buf, ref in_row_ctr, in_rows_avail, output_buf, ref out_row_group_ctr, out_row_groups_avail);
        }

        /// <summary>
        /// Create the wrapped-around downsampling input buffer needed for context mode.
        /// </summary>
        private void create_context_buffer()
        {
            int rgroup_height = m_cinfo.m_max_v_samp_factor;
            for (int ci = 0; ci < m_cinfo.m_num_components; ci++)
            {
                int samplesPerRow = (m_cinfo.Component_info[ci].Width_in_blocks * JpegConstants.DCTSize * m_cinfo.m_max_h_samp_factor) / m_cinfo.Component_info[ci].H_samp_factor;

                byte[][] fake_buffer = new byte[5 * rgroup_height][];
                for (int i = 1; i < 4 * rgroup_height; i++)
                    fake_buffer[i] = new byte[samplesPerRow];

                /* Allocate the actual buffer space (3 row groups) for this component.
                 * We make the buffer wide enough to allow the downsampler to edge-expand
                 * horizontally within the buffer, if it so chooses.
                 */
                byte[][] true_buffer = JpegCommonBase.AllocJpegSamples(samplesPerRow, 3 * rgroup_height);

                /* Copy true buffer row pointers into the middle of the fake row array */
                for (int i = 0; i < 3 * rgroup_height; i++)
                    fake_buffer[rgroup_height + i] = true_buffer[i];

                /* Fill in the above and below wraparound pointers */
                for (int i = 0; i < rgroup_height; i++)
                {
                    fake_buffer[i] = true_buffer[2 * rgroup_height + i];
                    fake_buffer[4 * rgroup_height + i] = true_buffer[i];
                }

                m_color_buf[ci] = fake_buffer;
                m_colorBufRowsOffset = rgroup_height;
            }
        }

        /// <summary>
        /// Process some data in the simple no-context case.
        /// 
        /// Preprocessor output data is counted in "row groups".  A row group
        /// is defined to be v_samp_factor sample rows of each component.
        /// Downsampling will produce this much data from each max_v_samp_factor
        /// input rows.
        /// </summary>
        private void pre_process_WithoutContext(byte[][] input_buf, ref int in_row_ctr, int in_rows_avail, byte[][][] output_buf, ref int out_row_group_ctr, int out_row_groups_avail)
        {
            while (in_row_ctr < in_rows_avail && out_row_group_ctr < out_row_groups_avail)
            {
                /* Do color conversion to fill the conversion buffer. */
                int inrows = in_rows_avail - in_row_ctr;
                int numrows = m_cinfo.m_max_v_samp_factor - m_next_buf_row;
                numrows = Math.Min(numrows, inrows);
                m_cinfo.m_cconvert.color_convert(input_buf, in_row_ctr, m_color_buf, m_colorBufRowsOffset + m_next_buf_row, numrows);
                in_row_ctr += numrows;
                m_next_buf_row += numrows;
                m_rows_to_go -= numrows;

                /* If at bottom of image, pad to fill the conversion buffer. */
                if (m_rows_to_go == 0 && m_next_buf_row < m_cinfo.m_max_v_samp_factor)
                {
                    for (int ci = 0; ci < m_cinfo.m_num_components; ci++)
                        expand_bottom_edge(m_color_buf[ci], m_colorBufRowsOffset, m_cinfo.m_image_width, m_next_buf_row, m_cinfo.m_max_v_samp_factor);

                    m_next_buf_row = m_cinfo.m_max_v_samp_factor;
                }

                /* If we've filled the conversion buffer, empty it. */
                if (m_next_buf_row == m_cinfo.m_max_v_samp_factor)
                {
                    m_cinfo.m_downsample.downsample(m_color_buf, m_colorBufRowsOffset, output_buf, out_row_group_ctr);
                    m_next_buf_row = 0;
                    out_row_group_ctr++;
                }

                /* If at bottom of image, pad the output to a full iMCU height.
                 * Note we assume the caller is providing a one-iMCU-height output buffer!
                 */
                if (m_rows_to_go == 0 && out_row_group_ctr < out_row_groups_avail)
                {
                    for (int ci = 0; ci < m_cinfo.m_num_components; ci++)
                    {
                        JpegComponent componentInfo = m_cinfo.Component_info[ci];
                        expand_bottom_edge(output_buf[ci], 0, componentInfo.Width_in_blocks * JpegConstants.DCTSize,
                            out_row_group_ctr * componentInfo.V_samp_factor,
                            out_row_groups_avail * componentInfo.V_samp_factor);
                    }

                    out_row_group_ctr = out_row_groups_avail;
                    break;          /* can exit outer loop without test */
                }
            }
        }

        /// <summary>
        /// Process some data in the context case.
        /// </summary>
        private void pre_process_context(byte[][] input_buf, ref int in_row_ctr, int in_rows_avail, byte[][][] output_buf, ref int out_row_group_ctr, int out_row_groups_avail)
        {
            while (out_row_group_ctr < out_row_groups_avail)
            {
                if (in_row_ctr < in_rows_avail)
                {
                    /* Do color conversion to fill the conversion buffer. */
                    int inrows = in_rows_avail - in_row_ctr;
                    int numrows = m_next_buf_stop - m_next_buf_row;
                    numrows = Math.Min(numrows, inrows);
                    m_cinfo.m_cconvert.color_convert(input_buf, in_row_ctr, m_color_buf, m_colorBufRowsOffset + m_next_buf_row, numrows);

                    /* Pad at top of image, if first time through */
                    if (m_rows_to_go == m_cinfo.m_image_height)
                    {
                        for (int ci = 0; ci < m_cinfo.m_num_components; ci++)
                        {
                            for (int row = 1; row <= m_cinfo.m_max_v_samp_factor; row++)
                                JpegUtils.jcopy_sample_rows(m_color_buf[ci], m_colorBufRowsOffset, m_color_buf[ci], m_colorBufRowsOffset - row, 1, m_cinfo.m_image_width);
                        }
                    }

                    in_row_ctr += numrows;
                    m_next_buf_row += numrows;
                    m_rows_to_go -= numrows;
                }
                else
                {
                    /* Return for more data, unless we are at the bottom of the image. */
                    if (m_rows_to_go != 0)
                        break;

                    /* When at bottom of image, pad to fill the conversion buffer. */
                    if (m_next_buf_row < m_next_buf_stop)
                    {
                        for (int ci = 0; ci < m_cinfo.m_num_components; ci++)
                            expand_bottom_edge(m_color_buf[ci], m_colorBufRowsOffset, m_cinfo.m_image_width, m_next_buf_row, m_next_buf_stop);

                        m_next_buf_row = m_next_buf_stop;
                    }
                }

                /* If we've gotten enough data, downsample a row group. */
                if (m_next_buf_row == m_next_buf_stop)
                {
                    m_cinfo.m_downsample.downsample(m_color_buf, m_colorBufRowsOffset + m_this_row_group, output_buf, out_row_group_ctr);
                    out_row_group_ctr++;

                    /* Advance pointers with wraparound as necessary. */
                    m_this_row_group += m_cinfo.m_max_v_samp_factor;
                    int buf_height = m_cinfo.m_max_v_samp_factor * 3;

                    if (m_this_row_group >= buf_height)
                        m_this_row_group = 0;

                    if (m_next_buf_row >= buf_height)
                        m_next_buf_row = 0;

                    m_next_buf_stop = m_next_buf_row + m_cinfo.m_max_v_samp_factor;
                }
            }
        }

        /// <summary>
        /// Expand an image vertically from height input_rows to height output_rows,
        /// by duplicating the bottom row.
        /// </summary>
        private static void expand_bottom_edge(byte[][] image_data, int rowsOffset, int num_cols, int input_rows, int output_rows)
        {
            for (int row = input_rows; row < output_rows; row++)
                JpegUtils.jcopy_sample_rows(image_data, rowsOffset + input_rows - 1, image_data, row, 1, num_cols);
        }
    }
    #endregion

    #region JpegDecompressor
    /// <summary>
    /// JPEG decompression routine.
    /// </summary>
    /// <seealso cref="JpegCompressor"/>
    public class JpegDecompressor : JpegCommonBase
    {
        /// <summary>
        /// The delegate for application-supplied marker processing methods.
        /// </summary>
        /// <param name="cinfo">Decompressor.</param>
        /// <returns>Return <c>true</c> to indicate success. <c>false</c> should be returned only 
        /// if you are using a suspending data source and it tells you to suspend.
        /// </returns>
        /// <remarks>Although the marker code is not explicitly passed, the routine can find it 
        /// in the <see cref="JpegDecompressor.Unread_marker"/>. At the time of call, 
        /// the marker proper has been read from the data source module. The processor routine 
        /// is responsible for reading the marker length word and the remaining parameter bytes, if any.
        /// </remarks>
        public delegate bool jpeg_marker_parser_method(JpegDecompressor cinfo);

        /* Source of compressed data */
        internal Jpeg_Source m_src;

        internal int m_image_width; /* nominal image width (from SOF marker) */
        internal int m_image_height;    /* nominal image height */
        internal int m_num_components;     /* # of color components in JPEG image */
        internal ColorSpace m_jpeg_color_space; /* colorspace of JPEG image */

        internal ColorSpace m_out_color_space; /* colorspace for output */
        internal int m_scale_num;
        internal int m_scale_denom; /* fraction by which to scale image */
        internal bool m_buffered_image;    /* true=multiple output passes */
        internal bool m_raw_data_out;      /* true=downsampled data wanted */
        internal DCTMethod m_dct_method;    /* IDCT algorithm selector */
        internal bool m_do_fancy_upsampling;   /* true=apply fancy up-sampling */
        internal bool m_do_block_smoothing;    /* true=apply inter-block smoothing */
        internal bool m_quantize_colors;   /* true=colormapped output wanted */
        internal DitherMode m_dither_mode;  /* type of color dithering to use */
        internal bool m_two_pass_quantize; /* true=use two-pass color quantization */
        internal int m_desired_number_of_colors;   /* max # colors to use in created colormap */
        internal bool m_enable_1pass_quant;    /* enable future use of 1-pass quantizer */
        internal bool m_enable_external_quant;/* enable future use of external colormap */
        internal bool m_enable_2pass_quant;    /* enable future use of 2-pass quantizer */

        internal int m_output_width;    /* scaled image width */
        internal int m_output_height;   /* scaled image height */
        internal int m_out_color_components;   /* # of color components in out_color_space */
        /* # of color components returned
         * output_components is 1 (a colormap index) when quantizing colors;
         * otherwise it equals out_color_components.
         */
        internal int m_output_components;

        internal int m_rec_outbuf_height;  /* min recommended height of scanline buffer */

        internal int m_actual_number_of_colors;    /* number of entries in use */
        internal byte[][] m_colormap;     /* The color map as a 2-D pixel array */

        internal int m_output_scanline; /* 0 .. output_height-1  */

        internal int m_input_scan_number;  /* Number of SOS markers seen so far */
        internal int m_input_iMCU_row;  /* Number of iMCU rows completed */

        internal int m_output_scan_number; /* Nominal scan number being displayed */
        internal int m_output_iMCU_row; /* Number of iMCU rows read */

        internal int[][] m_coef_bits; /* -1 or current Al value for each coef */

        /* Internal JPEG parameters --- the application usually need not look at
         * these fields.  Note that the decompressor output side may not use
         * any parameters that can change between scans.
         */

        /* Quantization and Huffman tables are carried forward across input
         * data-streams when processing abbreviated JPEG data-streams.
         */

        internal JpegQuantizationTable[] m_quant_tbl_ptrs = new JpegQuantizationTable[JpegConstants.NumberOfQuantTables];
        /* ptrs to coefficient quantization tables, or null if not defined */

        internal JpegHuffmanTable[] m_dc_huff_tbl_ptrs = new JpegHuffmanTable[JpegConstants.NumberOfHuffmanTables];
        internal JpegHuffmanTable[] m_ac_huff_tbl_ptrs = new JpegHuffmanTable[JpegConstants.NumberOfHuffmanTables];
        /* ptrs to Huffman coding tables, or null if not defined */

        /* These parameters are never carried across data-streams, since they
         * are given in SOF/SOS markers or defined to be reset by SOI.
         */

        internal int m_data_precision;     /* bits of precision in image data */

        /* m_comp_info[i] describes component that appears i'th in SOF */
        private JpegComponent[] m_comp_info;

        internal bool m_progressive_mode;  /* true if SOFn specifies progressive mode */

        internal int m_restart_interval; /* MCUs per restart interval, or 0 for no restart */

        /* These fields record data obtained from optional markers recognized by
         * the JPEG library.
         */
        internal bool m_saw_JFIF_marker;   /* true iff a JFIF APP0 marker was found */
        /* Data copied from JFIF marker; only valid if saw_JFIF_marker is true: */
        internal byte m_JFIF_major_version;   /* JFIF version number */
        internal byte m_JFIF_minor_version;

        internal DensityUnit m_density_unit;     /* JFIF code for pixel size units */
        internal short m_X_density;       /* Horizontal pixel density */
        internal short m_Y_density;       /* Vertical pixel density */

        internal bool m_saw_Adobe_marker;  /* true iff an Adobe APP14 marker was found */
        internal byte m_Adobe_transform;  /* Color transform code from Adobe marker */

        internal bool m_CCIR601_sampling;  /* true=first samples are co-sited */

        internal List<JpegMarker> m_marker_list; /* Head of list of saved markers */

        /* Remaining fields are known throughout decompressor, but generally
         * should not be touched by a surrounding application.
         */

        /*
         * These fields are computed during decompression startup
         */
        internal int m_max_h_samp_factor;  /* largest h_samp_factor */
        internal int m_max_v_samp_factor;  /* largest v_samp_factor */

        internal int m_min_DCT_scaled_size;    /* smallest DCT_scaled_size of any component */

        internal int m_total_iMCU_rows; /* # of iMCU rows in image */
        /* The coefficient controller's input and output progress is measured in
         * units of "iMCU" (interleaved MCU) rows.  These are the same as MCU rows
         * in fully interleaved JPEG scans, but are used whether the scan is
         * interleaved or not.  We define an iMCU row as v_samp_factor DCT block
         * rows of each component.  Therefore, the IDCT output contains
         * v_samp_factor*DCT_scaled_size sample rows of a component per iMCU row.
         */

        internal byte[] m_sample_range_limit; /* table for fast range-limiting */
        internal int m_sampleRangeLimitOffset;

        /*
         * These fields are valid during any one scan.
         * They describe the components and MCUs actually appearing in the scan.
         * Note that the decompressor output side must not use these fields.
         */
        internal int m_comps_in_scan;      /* # of JPEG components in this scan */
        internal int[] m_cur_comp_info = new int[JpegConstants.MaxComponentsInScan];
        /* *cur_comp_info[i] describes component that appears i'th in SOS */

        internal int m_MCUs_per_row;    /* # of MCUs across the image */
        internal int m_MCU_rows_in_scan;    /* # of MCU rows in the image */

        internal int m_blocks_in_MCU;      /* # of DCT blocks per MCU */
        internal int[] m_MCU_membership = new int[JpegConstants.DecompressorMaxBlocksInMCU];
        /* MCU_membership[i] is index in cur_comp_info of component owning */
        /* i'th block in an MCU */

        /* progressive JPEG parameters for scan */
        internal int m_Ss;
        internal int m_Se;
        internal int m_Ah;
        internal int m_Al;

        /* This field is shared between entropy decoder and marker parser.
         * It is either zero or the code of a JPEG marker that has been
         * read from the data source, but has not yet been processed.
         */
        internal int m_unread_marker;

        /*
         * Links to decompression sub-objects (methods, private variables of modules)
         */
        internal JpegDecompressorMaster m_master;
        internal JpegDecompressorMainController m_main;
        internal JpegDecompressorCoefController m_coef;
        internal JpegDecompressorPostController m_post;
        internal JpegInputController m_inputctl;
        internal JpegMarkerReader m_marker;
        internal JpegEntropyDecoder m_entropy;
        internal JpegInverseDCT m_idct;
        internal JpegUpsampler m_upsample;
        internal ColorDeconverter m_cconvert;
        internal ColorQuantizer m_cquantize;

        /// <summary>
        /// Initializes a new instance of the <see cref="JpegDecompressor"/> class.
        /// </summary>
        /// <seealso cref="JpegCompressor"/>
        public JpegDecompressor()
            : base()
        {
            initialize();
        }

        /// <summary>
        /// Retrieves <c>true</c> because this is a decompressor.
        /// </summary>
        /// <value><c>true</c></value>
        public override bool IsDecompressor
        {
            get { return true; }
        }

        /// <summary>
        /// Gets or sets the source for decompression.
        /// </summary>
        /// <value>The source for decompression.</value>
        public LibJpeg.Jpeg_Source Src
        {
            get { return m_src; }
            set { m_src = value; }
        }

        /* Basic description of image --- filled in by jpeg_read_header(). */
        /* Application may inspect these values to decide how to process image. */

        /// <summary>
        /// Gets the width of image, set by <see cref="JpegDecompressor.jpeg_read_header"/>
        /// </summary>
        /// <value>The width of image.</value>
        /// <seealso href="0955150c-4ee7-4b0f-a716-4bda2e85652b.htm" target="_self">Decompression parameter selection</seealso>
        public int Image_width
        {
            get { return m_image_width; }
        }

        /// <summary>
        /// Gets the height of image, set by <see cref="JpegDecompressor.jpeg_read_header"/>
        /// </summary>
        /// <value>The height of image.</value>
        /// <seealso href="0955150c-4ee7-4b0f-a716-4bda2e85652b.htm" target="_self">Decompression parameter selection</seealso>
        public int Image_height
        {
            get { return m_image_height; }
        }

        /// <summary>
        /// Gets the number of color components in JPEG image.
        /// </summary>
        /// <value>The number of color components.</value>
        /// <seealso href="0955150c-4ee7-4b0f-a716-4bda2e85652b.htm" target="_self">Decompression parameter selection</seealso>
        public int Num_components
        {
            get { return m_num_components; }
        }

        /// <summary>
        /// Gets or sets the colorspace of JPEG image.
        /// </summary>
        /// <value>The colorspace of JPEG image.</value>
        /// <seealso href="0955150c-4ee7-4b0f-a716-4bda2e85652b.htm" target="_self">Decompression parameter selection</seealso>
        public LibJpeg.ColorSpace Jpeg_color_space
        {
            get { return m_jpeg_color_space; }
            set { m_jpeg_color_space = value; }
        }

        /// <summary>
        /// Gets the list of loaded special markers.
        /// </summary>
        /// <remarks>All the special markers in the file appear in this list, in order of 
        /// their occurrence in the file (but omitting any markers of types you didn't ask for)
        /// </remarks>
        /// <value>The list of loaded special markers.</value>
        /// <seealso href="81c88818-a5d7-4550-9ce5-024a768f7b1e.htm" target="_self">Special markers</seealso>
        public ReadOnlyCollection<JpegMarker> Marker_list
        {
            get
            {
                return m_marker_list.AsReadOnly();
            }
        }

        /* Decompression processing parameters --- these fields must be set before
         * calling jpeg_start_decompress().  Note that jpeg_read_header() initializes
         * them to default values.
         */

        /// <summary>
        /// Gets or sets the output color space.
        /// </summary>
        /// <value>The output color space.</value>
        /// <seealso href="0955150c-4ee7-4b0f-a716-4bda2e85652b.htm" target="_self">Decompression parameter selection</seealso>
        public LibJpeg.ColorSpace Out_color_space
        {
            get { return m_out_color_space; }
            set { m_out_color_space = value; }
        }

        /// <summary>
        /// Gets or sets the numerator of the fraction of image scaling.
        /// </summary>
        /// <value>Scale the image by the fraction Scale_num/<see cref="JpegDecompressor.Scale_denom">Scale_denom</see>. 
        /// Default is 1/1, or no scaling. Currently, the only supported scaling ratios are 1/1, 1/2, 1/4, and 1/8.
        /// (The library design allows for arbitrary scaling ratios but this is not likely to be implemented any time soon.)
        /// </value>
        /// <remarks>Smaller scaling ratios permit significantly faster decoding since fewer pixels 
        /// need to be processed and a simpler <see cref="DCTMethod">DCT method</see> can be used.</remarks>
        /// <seealso cref="JpegDecompressor.Scale_denom"/>
        /// <seealso href="0955150c-4ee7-4b0f-a716-4bda2e85652b.htm" target="_self">Decompression parameter selection</seealso>
        public int Scale_num
        {
            get { return m_scale_num; }
            set { m_scale_num = value; }
        }

        /// <summary>
        /// Gets or sets the denominator of the fraction of image scaling.
        /// </summary>
        /// <value>Scale the image by the fraction <see cref="JpegDecompressor.Scale_num">Scale_num</see>/Scale_denom. 
        /// Default is 1/1, or no scaling. Currently, the only supported scaling ratios are 1/1, 1/2, 1/4, and 1/8.
        /// (The library design allows for arbitrary scaling ratios but this is not likely to be implemented any time soon.)
        /// </value>
        /// <remarks>Smaller scaling ratios permit significantly faster decoding since fewer pixels 
        /// need to be processed and a simpler <see cref="DCTMethod">DCT method</see> can be used.</remarks>
        /// <seealso cref="JpegDecompressor.Scale_num"/>
        /// <seealso href="0955150c-4ee7-4b0f-a716-4bda2e85652b.htm" target="_self">Decompression parameter selection</seealso>
        public int Scale_denom
        {
            get { return m_scale_denom; }
            set { m_scale_denom = value; }
        }

        /// <summary>
        /// Gets or sets a value indicating whether to use buffered-image mode.
        /// </summary>
        /// <value><c>true</c> if buffered-image mode is turned on; otherwise, <c>false</c>.</value>
        /// <seealso href="6dba59c5-d32e-4dfc-87fe-f9eff7004146.htm" target="_self">Buffered-image mode</seealso>
        public bool Buffered_image
        {
            get { return m_buffered_image; }
            set { m_buffered_image = value; }
        }

        /// <summary>
        /// Enable or disable raw data output.
        /// </summary>
        /// <value><c>true</c> if raw data output is enabled; otherwise, <c>false</c>.</value>
        /// <remarks>Default value: <c>false</c><br/>
        /// Set this to true before <see cref="JpegDecompressor.jpeg_start_decompress"/> 
        /// if you need to obtain raw data output.
        /// </remarks>
        /// <seealso cref="jpeg_read_raw_data"/>
        public bool Raw_data_out
        {
            get { return m_raw_data_out; }
            set { m_raw_data_out = value; }
        }

        /// <summary>
        /// Gets or sets the algorithm used for the DCT step.
        /// </summary>
        /// <value>The algorithm used for the DCT step.</value>
        /// <seealso href="0955150c-4ee7-4b0f-a716-4bda2e85652b.htm" target="_self">Decompression parameter selection</seealso>
        public LibJpeg.DCTMethod Dct_method
        {
            get { return m_dct_method; }
            set { m_dct_method = value; }
        }

        /// <summary>
        /// Enable or disable up-sampling of chroma components.
        /// </summary>
        /// <value>If <c>true</c>, do careful up-sampling of chroma components. 
        /// If <c>false</c>, a faster but sloppier method is used. 
        /// The visual impact of the sloppier method is often very small.
        /// </value>
        /// <remarks>Default value: <c>true</c></remarks>
        /// <seealso href="0955150c-4ee7-4b0f-a716-4bda2e85652b.htm" target="_self">Decompression parameter selection</seealso>
        public bool Do_fancy_upsampling
        {
            get { return m_do_fancy_upsampling; }
            set { m_do_fancy_upsampling = value; }
        }

        /// <summary>
        /// Apply inter-block smoothing in early stages of decoding progressive JPEG files.
        /// </summary>
        /// <value>If <c>true</c>, inter-block smoothing is applied in early stages of decoding progressive JPEG files; 
        /// if <c>false</c>, not. Early progression stages look "fuzzy" with smoothing, "blocky" without.</value>
        /// <remarks>Default value: <c>true</c><br/>
        /// In any case, block smoothing ceases to be applied after the first few AC coefficients are 
        /// known to full accuracy, so it is relevant only when using 
        /// <see href="6dba59c5-d32e-4dfc-87fe-f9eff7004146.htm" target="_self">buffered-image mode</see> for progressive images.
        /// </remarks>
        /// <seealso href="0955150c-4ee7-4b0f-a716-4bda2e85652b.htm" target="_self">Decompression parameter selection</seealso>
        public bool Do_block_smoothing
        {
            get { return m_do_block_smoothing; }
            set { m_do_block_smoothing = value; }
        }

        /// <summary>
        /// Colors quantization.
        /// </summary>
        /// <value>If set <c>true</c>, colormapped output will be delivered.<br/>
        /// Default value: <c>false</c>, meaning that full-color output will be delivered.
        /// </value>
        /// <seealso href="0955150c-4ee7-4b0f-a716-4bda2e85652b.htm" target="_self">Decompression parameter selection</seealso>
        public bool Quantize_colors
        {
            get { return m_quantize_colors; }
            set { m_quantize_colors = value; }
        }

        /* the following are ignored if not quantize_colors: */

        /// <summary>
        /// Selects color dithering method.
        /// </summary>
        /// <value>Default value: <see cref="DitherMode.FloydStein"/>.</value>
        /// <remarks>Ignored if <see cref="JpegDecompressor.Quantize_colors"/> is <c>false</c>.<br/>
        /// At present, ordered dither is implemented only in the single-pass, standard-colormap case. 
        /// If you ask for ordered dither when <see cref="JpegDecompressor.Two_pass_quantize"/> is <c>true</c>
        /// or when you supply an external color map, you'll get F-S dithering.
        /// </remarks>
        /// <seealso cref="JpegDecompressor.Quantize_colors"/>
        /// <seealso href="0955150c-4ee7-4b0f-a716-4bda2e85652b.htm" target="_self">Decompression parameter selection</seealso>
        public LibJpeg.DitherMode Dither_mode
        {
            get { return m_dither_mode; }
            set { m_dither_mode = value; }
        }

        /// <summary>
        /// Gets or sets a value indicating whether to use two-pass color quantization.
        /// </summary>
        /// <value>If <c>true</c>, an extra pass over the image is made to select a custom color map for the image.
        /// This usually looks a lot better than the one-size-fits-all colormap that is used otherwise.
        /// Ignored when the application supplies its own color map.<br/>
        /// 
        /// Default value: <c>true</c>
        /// </value>
        /// <remarks>Ignored if <see cref="JpegDecompressor.Quantize_colors"/> is <c>false</c>.<br/>
        /// </remarks>
        /// <seealso cref="JpegDecompressor.Quantize_colors"/>
        /// <seealso href="0955150c-4ee7-4b0f-a716-4bda2e85652b.htm" target="_self">Decompression parameter selection</seealso>
        public bool Two_pass_quantize
        {
            get { return m_two_pass_quantize; }
            set { m_two_pass_quantize = value; }
        }

        /// <summary>
        /// Maximum number of colors to use in generating a library-supplied color map.
        /// </summary>
        /// <value>Default value: 256.</value>
        /// <remarks>Ignored if <see cref="JpegDecompressor.Quantize_colors"/> is <c>false</c>.<br/>
        /// The actual number of colors is returned in a <see cref="JpegDecompressor.Actual_number_of_colors"/>.
        /// </remarks>
        /// <seealso cref="JpegDecompressor.Quantize_colors"/>
        /// <seealso href="0955150c-4ee7-4b0f-a716-4bda2e85652b.htm" target="_self">Decompression parameter selection</seealso>
        public int Desired_number_of_colors
        {
            get { return m_desired_number_of_colors; }
            set { m_desired_number_of_colors = value; }
        }

        /* these are significant only in buffered-image mode: */

        /// <summary>
        /// Enable future use of 1-pass quantizer.
        /// </summary>
        /// <value>Default value: <c>false</c></value>
        /// <remarks>Significant only in buffered-image mode.</remarks>
        /// <seealso href="6dba59c5-d32e-4dfc-87fe-f9eff7004146.htm" target="_self">Buffered-image mode</seealso>
        public bool Enable_1pass_quant
        {
            get { return m_enable_1pass_quant; }
            set { m_enable_1pass_quant = value; }
        }

        /// <summary>
        /// Enable future use of external colormap.
        /// </summary>
        /// <value>Default value: <c>false</c></value>
        /// <remarks>Significant only in buffered-image mode.</remarks>
        /// <seealso href="6dba59c5-d32e-4dfc-87fe-f9eff7004146.htm" target="_self">Buffered-image mode</seealso>
        public bool Enable_external_quant
        {
            get { return m_enable_external_quant; }
            set { m_enable_external_quant = value; }
        }

        /// <summary>
        /// Enable future use of 2-pass quantizer.
        /// </summary>
        /// <value>Default value: <c>false</c></value>
        /// <remarks>Significant only in buffered-image mode.</remarks>
        /// <seealso href="6dba59c5-d32e-4dfc-87fe-f9eff7004146.htm" target="_self">Buffered-image mode</seealso>
        public bool Enable_2pass_quant
        {
            get { return m_enable_2pass_quant; }
            set { m_enable_2pass_quant = value; }
        }

        /* Description of actual output image that will be returned to application.
         * These fields are computed by jpeg_start_decompress().
         * You can also use jpeg_calc_output_dimensions() to determine these values
         * in advance of calling jpeg_start_decompress().
         */

        /// <summary>
        /// Gets the actual width of output image.
        /// </summary>
        /// <value>The width of output image.</value>
        /// <remarks>Computed by <see cref="JpegDecompressor.jpeg_start_decompress"/>.
        /// You can also use <see cref="JpegDecompressor.jpeg_calc_output_dimensions"/> to determine this value
        /// in advance of calling <see cref="JpegDecompressor.jpeg_start_decompress"/>.</remarks>
        /// <seealso cref="JpegDecompressor.Output_height"/>
        public int Output_width
        {
            get { return m_output_width; }
        }

        /// <summary>
        /// Gets the actual height of output image.
        /// </summary>
        /// <value>The height of output image.</value>
        /// <remarks>Computed by <see cref="JpegDecompressor.jpeg_start_decompress"/>.
        /// You can also use <see cref="JpegDecompressor.jpeg_calc_output_dimensions"/> to determine this value
        /// in advance of calling <see cref="JpegDecompressor.jpeg_start_decompress"/>.</remarks>
        /// <seealso cref="JpegDecompressor.Output_width"/>
        public int Output_height
        {
            get { return m_output_height; }
        }

        /// <summary>
        /// Gets the number of color components in <see cref="JpegDecompressor.Out_color_space"/>.
        /// </summary>
        /// <remarks>Computed by <see cref="JpegDecompressor.jpeg_start_decompress"/>.
        /// You can also use <see cref="JpegDecompressor.jpeg_calc_output_dimensions"/> to determine this value
        /// in advance of calling <see cref="JpegDecompressor.jpeg_start_decompress"/>.</remarks>
        /// <value>The number of color components.</value>
        /// <seealso cref="JpegDecompressor.Out_color_space"/>
        /// <seealso href="0955150c-4ee7-4b0f-a716-4bda2e85652b.htm" target="_self">Decompression parameter selection</seealso>
        public int Out_color_components
        {
            get { return m_out_color_components; }
        }

        /// <summary>
        /// Gets the number of color components returned.
        /// </summary>
        /// <remarks>Computed by <see cref="JpegDecompressor.jpeg_start_decompress"/>.
        /// You can also use <see cref="JpegDecompressor.jpeg_calc_output_dimensions"/> to determine this value
        /// in advance of calling <see cref="JpegDecompressor.jpeg_start_decompress"/>.</remarks>
        /// <value>When <see cref="JpegDecompressor.Quantize_colors">quantizing colors</see>, 
        /// <c>Output_components</c> is 1, indicating a single color map index per pixel. 
        /// Otherwise it equals to <see cref="JpegDecompressor.Out_color_components"/>.
        /// </value>
        /// <seealso cref="JpegDecompressor.Out_color_space"/>
        /// <seealso href="0955150c-4ee7-4b0f-a716-4bda2e85652b.htm" target="_self">Decompression parameter selection</seealso>
        public int Output_components
        {
            get { return m_output_components; }
        }

        /// <summary>
        /// Gets the recommended height of scanline buffer.
        /// </summary>
        /// <value>In high-quality modes, <c>Rec_outbuf_height</c> is always 1, but some faster, 
        /// lower-quality modes set it to larger values (typically 2 to 4).</value>
        /// <remarks>Computed by <see cref="JpegDecompressor.jpeg_start_decompress"/>.
        /// You can also use <see cref="JpegDecompressor.jpeg_calc_output_dimensions"/> to determine this value
        /// in advance of calling <see cref="JpegDecompressor.jpeg_start_decompress"/>.<br/>
        /// 
        /// <c>Rec_outbuf_height</c> is the recommended minimum height (in scanlines) 
        /// of the buffer passed to <see cref="JpegDecompressor.jpeg_read_scanlines"/>.
        /// If the buffer is smaller, the library will still work, but time will be wasted due 
        /// to unnecessary data copying. If you are going to ask for a high-speed processing mode, 
        /// you may as well go to the trouble of honoring <c>Rec_outbuf_height</c> so as to avoid data copying.
        /// (An output buffer larger than <c>Rec_outbuf_height</c> lines is OK, but won't provide 
        /// any material speed improvement over that height.)
        /// </remarks>
        /// <seealso href="0955150c-4ee7-4b0f-a716-4bda2e85652b.htm" target="_self">Decompression parameter selection</seealso>
        public int Rec_outbuf_height
        {
            get { return m_rec_outbuf_height; }
        }

        /* When quantizing colors, the output colormap is described by these fields.
         * The application can supply a colormap by setting colormap non-null before
         * calling jpeg_start_decompress; otherwise a colormap is created during
         * jpeg_start_decompress or jpeg_start_output.
         * The map has out_color_components rows and actual_number_of_colors columns.
         */

        /// <summary>
        /// The number of colors in the color map.
        /// </summary>
        /// <value>The number of colors in the color map.</value>
        /// <seealso cref="JpegDecompressor.Colormap"/>
        /// <seealso href="0955150c-4ee7-4b0f-a716-4bda2e85652b.htm" target="_self">Decompression parameter selection</seealso>
        public int Actual_number_of_colors
        {
            get { return m_actual_number_of_colors; }
            set { m_actual_number_of_colors = value; }
        }

        /// <summary>
        /// The color map, represented as a 2-D pixel array of <see cref="JpegDecompressor.Out_color_components"/> rows 
        /// and <see cref="JpegDecompressor.Actual_number_of_colors"/> columns.
        /// </summary>
        /// <value>Colormap is set to <c>null</c> by <see cref="JpegDecompressor.jpeg_read_header"/>.
        /// The application can supply a color map by setting <c>Colormap</c> non-null and setting 
        /// <see cref="JpegDecompressor.Actual_number_of_colors"/> to the map size.
        /// </value>
        /// <remarks>Ignored if not quantizing.<br/>
        /// Implementation restriction: at present, an externally supplied <c>Colormap</c>
        /// is only accepted for 3-component output color spaces.
        /// </remarks>
        /// <seealso cref="JpegDecompressor.Actual_number_of_colors"/>
        /// <seealso cref="JpegDecompressor.Quantize_colors"/>
        /// <seealso href="0955150c-4ee7-4b0f-a716-4bda2e85652b.htm" target="_self">Decompression parameter selection</seealso>
        public byte[][] Colormap
        {
            get { return m_colormap; }
            set { m_colormap = value; }
        }

        /* State variables: these variables indicate the progress of decompression.
         * The application may examine these but must not modify them.
         */

        /* Row index of next scanline to be read from jpeg_read_scanlines().
         * Application may use this to control its processing loop, e.g.,
         * "while (output_scanline < output_height)".
         */

        /// <summary>
        /// Gets the number of scanlines returned so far.
        /// </summary>
        /// <value>The output_scanline.</value>
        /// <remarks>Usually you can just use this variable as the loop counter, 
        /// so that the loop test looks like 
        /// <c>while (cinfo.Output_scanline &lt; cinfo.Output_height)</c></remarks>
        /// <seealso href="9d052723-a7f9-42de-8747-0bd9896f8157.htm" target="_self">Decompression details</seealso>
        public int Output_scanline
        {
            get { return m_output_scanline; }
        }

        /* Current input scan number and number of iMCU rows completed in scan.
         * These indicate the progress of the decompressor input side.
         */

        /// <summary>
        /// Gets the number of SOS markers seen so far.
        /// </summary>
        /// <value>The number of SOS markers seen so far.</value>
        /// <remarks>Indicates the progress of the decompressor input side.</remarks>
        public int Input_scan_number
        {
            get { return m_input_scan_number; }
        }

        /// <summary>
        /// Gets the number of iMCU rows completed.
        /// </summary>
        /// <value>The number of iMCU rows completed.</value>
        /// <remarks>Indicates the progress of the decompressor input side.</remarks>
        public int Input_iMCU_row
        {
            get { return m_input_iMCU_row; }
        }

        /* The "output scan number" is the notional scan being displayed by the
         * output side.  The decompressor will not allow output scan/row number
         * to get ahead of input scan/row, but it can fall arbitrarily far behind.
         */

        /// <summary>
        /// Gets the nominal scan number being displayed.
        /// </summary>
        /// <value>The nominal scan number being displayed.</value>
        public int Output_scan_number
        {
            get { return m_output_scan_number; }
        }

        /// <summary>
        /// Gets the number of iMCU rows read.
        /// </summary>
        /// <value>The number of iMCU rows read.</value>
        public int Output_iMCU_row
        {
            get { return m_output_iMCU_row; }
        }

        /* Current progression status.  coef_bits[c][i] indicates the precision
         * with which component c's DCT coefficient i (in zigzag order) is known.
         * It is -1 when no data has yet been received, otherwise it is the point
         * transform (shift) value for the most recent scan of the coefficient
         * (thus, 0 at completion of the progression).
         * This is null when reading a non-progressive file.
         */

        /// <summary>
        /// Gets the current progression status..
        /// </summary>
        /// <value><c>Coef_bits[c][i]</c> indicates the precision with 
        /// which component c's DCT coefficient i (in zigzag order) is known. 
        /// It is <c>-1</c> when no data has yet been received, otherwise 
        /// it is the point transform (shift) value for the most recent scan of the coefficient 
        /// (thus, 0 at completion of the progression). This is null when reading a non-progressive file.
        /// </value>
        /// <seealso href="bda5b19b-88e0-44bf-97de-cd30fc61bb65.htm" target="_self">Progressive JPEG support</seealso>
        public int[][] Coef_bits
        {
            get { return m_coef_bits; }
        }

        // These fields record data obtained from optional markers 
        // recognized by the JPEG library.

        /// <summary>
        /// Gets the resolution information from JFIF marker.
        /// </summary>
        /// <value>The information from JFIF marker.</value>
        /// <seealso cref="JpegDecompressor.X_density"/>
        /// <seealso cref="JpegDecompressor.Y_density"/>
        /// <seealso href="0955150c-4ee7-4b0f-a716-4bda2e85652b.htm" target="_self">Decompression parameter selection</seealso>
        public DensityUnit Density_unit
        {
            get { return m_density_unit; }
        }

        /// <summary>
        /// Gets the horizontal component of pixel ratio.
        /// </summary>
        /// <value>The horizontal component of pixel ratio.</value>
        /// <seealso cref="JpegDecompressor.Y_density"/>
        /// <seealso cref="JpegDecompressor.Density_unit"/>
        public short X_density
        {
            get { return m_X_density; }
        }

        /// <summary>
        /// Gets the vertical component of pixel ratio.
        /// </summary>
        /// <value>The vertical component of pixel ratio.</value>
        /// <seealso cref="JpegDecompressor.X_density"/>
        /// <seealso cref="JpegDecompressor.Density_unit"/>
        public short Y_density
        {
            get { return m_Y_density; }
        }

        /// <summary>
        /// Gets the data precision.
        /// </summary>
        /// <value>The data precision.</value>
        public int Data_precision
        {
            get { return m_data_precision; }
            //set { m_data_precision = value; }
        }

        /// <summary>
        /// Gets the largest vertical sample factor.
        /// </summary>
        /// <value>The largest vertical sample factor.</value>
        public int Max_v_samp_factor
        {
            get { return m_max_v_samp_factor; }
            //set { m_max_v_samp_factor = value; }
        }

        /// <summary>
        /// Gets the last read and unprocessed JPEG marker.
        /// </summary>
        /// <value>It is either zero or the code of a JPEG marker that has been
        /// read from the data source, but has not yet been processed.
        /// </value>
        /// <seealso cref="JpegDecompressor.jpeg_set_marker_processor"/>
        /// <seealso href="81c88818-a5d7-4550-9ce5-024a768f7b1e.htm" target="_self">Special markers</seealso>
        public int Unread_marker
        {
            get { return m_unread_marker; }
        }

        /// <summary>
        /// Comp_info[i] describes component that appears i'th in SOF
        /// </summary>
        /// <value>The components in SOF.</value>
        /// <seealso cref="JpegComponent"/>
        public JpegComponent[] Comp_info
        {
            get { return m_comp_info; }
            internal set { m_comp_info = value; }
        }

        /// <summary>
        /// Sets input stream.
        /// </summary>
        /// <param name="infile">The input stream.</param>
        /// <remarks>
        /// The caller must have already opened the stream, and is responsible
        /// for closing it after finishing decompression.
        /// </remarks>
        /// <seealso href="9d052723-a7f9-42de-8747-0bd9896f8157.htm" target="_self">Decompression details</seealso>
        public void jpeg_stdio_src(Stream infile)
        {
            /* The source object and input buffer are made permanent so that a series
            * of JPEG images can be read from the same file by calling jpeg_stdio_src
            * only before the first one.  (If we discarded the buffer at the end of
            * one image, we'd likely lose the start of the next one.)
            * This makes it unsafe to use this manager and a different source
            * manager serially with the same JPEG object.  Caveat programmer.
            */
            if (m_src == null)
            {
                /* first time for this JPEG object? */
                m_src = new SourceManagerImpl(this);
            }

            SourceManagerImpl m = m_src as SourceManagerImpl;
            if (m != null)
                m.Attach(infile);
        }

        /// <summary>
        /// Decompression startup: this will read the source datastream header markers, up to the beginning of the compressed data proper.
        /// </summary>
        /// <param name="require_image">Read a description of <b>Return Value</b>.</param>
        /// <returns>
        /// If you pass <c>require_image=true</c> (normal case), you need not check for a
        /// <see cref="ReadResult.Header_Tables_Only"/> return code; an abbreviated file will cause
        /// an error exit. <see cref="ReadResult.Suspended"/> is only possible if you use a data source
        /// module that can give a suspension return.<br/><br/>
        /// 
        /// This method will read as far as the first SOS marker (ie, actual start of compressed data),
        /// and will save all tables and parameters in the JPEG object. It will also initialize the
        /// decompression parameters to default values, and finally return <see cref="ReadResult.Header_Ok"/>.
        /// On return, the application may adjust the decompression parameters and then call
        /// <see cref="JpegDecompressor.jpeg_start_decompress"/>. (Or, if the application only wanted to
        /// determine the image parameters, the data need not be decompressed. In that case, call
        /// <see cref="JpegCommonBase.jpeg_abort"/> to release any temporary space.)<br/><br/>
        /// 
        /// If an abbreviated (tables only) datastream is presented, the routine will return
        /// <see cref="ReadResult.Header_Tables_Only"/> upon reaching EOI. The application may then re-use
        /// the JPEG object to read the abbreviated image datastream(s). It is unnecessary (but OK) to call
        /// <see cref="JpegCommonBase.jpeg_abort">jpeg_abort</see> in this case.
        /// The <see cref="ReadResult.Suspended"/> return code only occurs if the data source module
        /// requests suspension of the decompressor. In this case the application should load more source
        /// data and then re-call <c>jpeg_read_header</c> to resume processing.<br/><br/>
        /// 
        /// If a non-suspending data source is used and <c>require_image</c> is <c>true</c>,
        /// then the return code need not be inspected since only <see cref="ReadResult.Header_Ok"/> is possible.
        /// </returns>
        /// <remarks>Need only initialize JPEG object and supply a data source before calling.<br/>
        /// On return, the image dimensions and other info have been stored in the JPEG object.
        /// The application may wish to consult this information before selecting decompression parameters.<br/>
        /// This routine is now just a front end to <see cref="jpeg_consume_input"/>, with some extra error checking.
        /// </remarks>
        /// <seealso href="9d052723-a7f9-42de-8747-0bd9896f8157.htm" target="_self">Decompression details</seealso>
        /// <seealso href="0955150c-4ee7-4b0f-a716-4bda2e85652b.htm" target="_self">Decompression parameter selection</seealso>
        public ReadResult jpeg_read_header(bool require_image)
        {
            if (m_global_state != JpegState.DSTATE_START && m_global_state != JpegState.DSTATE_INHEADER)
                throw new Exception(String.Format("Improper call to JPEG library in state {0}", (int)m_global_state));

            ReadResult retcode = jpeg_consume_input();

            switch (retcode)
            {
                case ReadResult.Reached_SOS:
                    return ReadResult.Header_Ok;
                case ReadResult.Reached_EOI:
                    if (require_image)      /* Complain if application wanted an image */
                        throw new Exception("JPEG datastream contains no image");
                    /* Reset to start state; it would be safer to require the application to
                    * call jpeg_abort, but we can't change it now for compatibility reasons.
                    * A side effect is to free any temporary memory (there shouldn't be any).
                    */
                    jpeg_abort(); /* sets state = DSTATE_START */
                    return ReadResult.Header_Tables_Only;

                case ReadResult.Suspended:
                    /* no work */
                    break;
            }

            return ReadResult.Suspended;
        }

        //////////////////////////////////////////////////////////////////////////
        // Main entry points for decompression

        /// <summary>
        /// Decompression initialization.
        /// </summary>
        /// <returns>Returns <c>false</c> if suspended. The return value need be inspected 
        /// only if a suspending data source is used.
        /// </returns>
        /// <remarks><see cref="JpegDecompressor.jpeg_read_header">jpeg_read_header</see> must be completed before calling this.<br/>
        /// 
        /// If a multipass operating mode was selected, this will do all but the last pass, and thus may take a great deal of time.
        /// </remarks>
        /// <seealso cref="JpegDecompressor.jpeg_finish_decompress"/>
        /// <seealso href="9d052723-a7f9-42de-8747-0bd9896f8157.htm" target="_self">Decompression details</seealso>
        public bool jpeg_start_decompress()
        {
            if (m_global_state == JpegState.DSTATE_READY)
            {
                /* First call: initialize master control, select active modules */
                m_master = new JpegDecompressorMaster(this);
                if (m_buffered_image)
                {
                    /* No more work here; expecting jpeg_start_output next */
                    m_global_state = JpegState.DSTATE_BUFIMAGE;
                    return true;
                }
                m_global_state = JpegState.DSTATE_PRELOAD;
            }

            if (m_global_state == JpegState.DSTATE_PRELOAD)
            {
                /* If file has multiple scans, absorb them all into the coef buffer */
                if (m_inputctl.HasMultipleScans())
                {
                    for (; ; )
                    {
                        ReadResult retcode;

                        /* Absorb some more input */
                        retcode = m_inputctl.consume_input();
                        if (retcode == ReadResult.Suspended)
                            return false;

                        if (retcode == ReadResult.Reached_EOI)
                            break;
                    }
                }

                m_output_scan_number = m_input_scan_number;
            }
            else if (m_global_state != JpegState.DSTATE_PRESCAN)
                throw new Exception(String.Format("Improper call to JPEG library in state {0}", (int)m_global_state));

            /* Perform any dummy output passes, and set up for the final pass */
            return output_pass_setup();
        }

        /// <summary>
        /// Read some scanlines of data from the JPEG decompressor.
        /// </summary>
        /// <param name="scanlines">Buffer for filling.</param>
        /// <param name="max_lines">Required number of lines.</param>
        /// <returns>The return value will be the number of lines actually read. 
        /// This may be less than the number requested in several cases, including 
        /// bottom of image, data source suspension, and operating modes that emit multiple scanlines at a time.
        /// </returns>
        /// <remarks>We warn about excess calls to <c>jpeg_read_scanlines</c> since this likely signals an 
        /// application programmer error. However, an oversize buffer <c>(max_lines > scanlines remaining)</c> 
        /// is not an error.
        /// </remarks>
        /// <seealso href="9d052723-a7f9-42de-8747-0bd9896f8157.htm" target="_self">Decompression details</seealso>
        public int jpeg_read_scanlines(byte[][] scanlines, int max_lines)
        {
            if (m_global_state != JpegState.DSTATE_SCANNING)
                throw new Exception(String.Format("Improper call to JPEG library in state {0}", (int)m_global_state));

            if (m_output_scanline >= m_output_height)
            {
                return 0;
            }

            /* Process some data */
            int row_ctr = 0;
            m_main.process_data(scanlines, ref row_ctr, max_lines);
            m_output_scanline += row_ctr;
            return row_ctr;
        }

        /// <summary>
        /// Finish JPEG decompression.
        /// </summary>
        /// <returns>Returns <c>false</c> if suspended. The return value need be inspected 
        /// only if a suspending data source is used.
        /// </returns>
        /// <remarks>This will normally just verify the file trailer and release temp storage.</remarks>
        /// <seealso cref="JpegDecompressor.jpeg_start_decompress"/>
        /// <seealso href="9d052723-a7f9-42de-8747-0bd9896f8157.htm" target="_self">Decompression details</seealso>
        public bool jpeg_finish_decompress()
        {
            if ((m_global_state == JpegState.DSTATE_SCANNING || m_global_state == JpegState.DSTATE_RAW_OK) && !m_buffered_image)
            {
                /* Terminate final pass of non-buffered mode */
                if (m_output_scanline < m_output_height)
                    throw new Exception("Application transferred too few scanlines");

                m_master.finish_output_pass();
                m_global_state = JpegState.DSTATE_STOPPING;
            }
            else if (m_global_state == JpegState.DSTATE_BUFIMAGE)
            {
                /* Finishing after a buffered-image operation */
                m_global_state = JpegState.DSTATE_STOPPING;
            }
            else if (m_global_state != JpegState.DSTATE_STOPPING)
            {
                /* STOPPING = repeat call after a suspension, anything else is error */
                throw new Exception(String.Format("Improper call to JPEG library in state {0}", (int)m_global_state));
            }

            /* Read until EOI */
            while (!m_inputctl.EOIReached())
            {
                if (m_inputctl.consume_input() == ReadResult.Suspended)
                {
                    /* Suspend, come back later */
                    return false;
                }
            }

            /* Do final cleanup */
            m_src.term_source();

            /* We can use jpeg_abort to release memory and reset global_state */
            jpeg_abort();
            return true;
        }

        /// <summary>
        /// Alternate entry point to read raw data.
        /// </summary>
        /// <param name="data">The raw data.</param>
        /// <param name="max_lines">The number of scanlines for reading.</param>
        /// <returns>The number of lines actually read.</returns>
        /// <remarks>Replaces <see cref="JpegDecompressor.jpeg_read_scanlines">jpeg_read_scanlines</see> 
        /// when reading raw downsampled data. Processes exactly one iMCU row per call, unless suspended.
        /// </remarks>
        public int jpeg_read_raw_data(byte[][][] data, int max_lines)
        {
            if (m_global_state != JpegState.DSTATE_RAW_OK)
                throw new Exception(String.Format("Improper call to JPEG library in state {0}", (int)m_global_state));

            if (m_output_scanline >= m_output_height)
            {
                return 0;
            }

            /* Verify that at least one iMCU row can be returned. */
            int lines_per_iMCU_row = m_max_v_samp_factor * m_min_DCT_scaled_size;
            if (max_lines < lines_per_iMCU_row)
                throw new Exception("Buffer passed to JPEG library is too small");

            int componentCount = data.Length; // maybe we should use max_lines here
            ComponentBuffer[] cb = new ComponentBuffer[componentCount];
            for (int i = 0; i < componentCount; i++)
            {
                cb[i] = new ComponentBuffer();
                cb[i].SetBuffer(data[i], null, 0);
            }

            /* Decompress directly into user's buffer. */
            if (m_coef.decompress_data(cb) == ReadResult.Suspended)
            {
                /* suspension forced, can do nothing more */
                return 0;
            }

            /* OK, we processed one iMCU row. */
            m_output_scanline += lines_per_iMCU_row;
            return lines_per_iMCU_row;
        }

        //////////////////////////////////////////////////////////////////////////
        // Additional entry points for buffered-image mode.

        /// <summary>
        /// Is there more than one scan?
        /// </summary>
        /// <returns><c>true</c> if image has more than one scan; otherwise, <c>false</c></returns>
        /// <remarks>If you are concerned about maximum performance on baseline JPEG files,
        /// you should use <see href="6dba59c5-d32e-4dfc-87fe-f9eff7004146.htm" target="_self">buffered-image mode</see> only
        /// when the incoming file actually has multiple scans. This can be tested by calling this method.
        /// </remarks>
        public bool jpeg_has_multiple_scans()
        {
            /* Only valid after jpeg_read_header completes */
            if (m_global_state < JpegState.DSTATE_READY || m_global_state > JpegState.DSTATE_STOPPING)
                throw new Exception(String.Format("Improper call to JPEG library in state {0}", (int)m_global_state));

            return m_inputctl.HasMultipleScans();
        }

        /// <summary>
        /// Initialize for an output pass in <see href="6dba59c5-d32e-4dfc-87fe-f9eff7004146.htm" target="_self">buffered-image mode</see>.
        /// </summary>
        /// <param name="scan_number">Indicates which scan of the input file is to be displayed; 
        /// the scans are numbered starting at 1 for this purpose.</param>
        /// <returns><c>true</c> if done; <c>false</c> if suspended</returns>
        /// <seealso cref="JpegDecompressor.jpeg_finish_output"/>
        /// <seealso href="6dba59c5-d32e-4dfc-87fe-f9eff7004146.htm" target="_self">Buffered-image mode</seealso>
        public bool jpeg_start_output(int scan_number)
        {
            if (m_global_state != JpegState.DSTATE_BUFIMAGE && m_global_state != JpegState.DSTATE_PRESCAN)
                throw new Exception(String.Format("Improper call to JPEG library in state {0}", (int)m_global_state));

            /* Limit scan number to valid range */
            if (scan_number <= 0)
                scan_number = 1;

            if (m_inputctl.EOIReached() && scan_number > m_input_scan_number)
                scan_number = m_input_scan_number;

            m_output_scan_number = scan_number;
            /* Perform any dummy output passes, and set up for the real pass */
            return output_pass_setup();
        }

        /// <summary>
        /// Finish up after an output pass in <see href="6dba59c5-d32e-4dfc-87fe-f9eff7004146.htm" target="_self">buffered-image mode</see>.
        /// </summary>
        /// <returns>Returns <c>false</c> if suspended. The return value need be inspected only if a suspending data source is used.</returns>
        /// <seealso cref="JpegDecompressor.jpeg_start_output"/>
        /// <seealso href="6dba59c5-d32e-4dfc-87fe-f9eff7004146.htm" target="_self">Buffered-image mode</seealso>
        public bool jpeg_finish_output()
        {
            if ((m_global_state == JpegState.DSTATE_SCANNING || m_global_state == JpegState.DSTATE_RAW_OK) && m_buffered_image)
            {
                /* Terminate this pass. */
                /* We do not require the whole pass to have been completed. */
                m_master.finish_output_pass();
                m_global_state = JpegState.DSTATE_BUFPOST;
            }
            else if (m_global_state != JpegState.DSTATE_BUFPOST)
            {
                /* BUFPOST = repeat call after a suspension, anything else is error */
                throw new Exception(String.Format("Improper call to JPEG library in state {0}", (int)m_global_state));
            }

            /* Read markers looking for SOS or EOI */
            while (m_input_scan_number <= m_output_scan_number && !m_inputctl.EOIReached())
            {
                if (m_inputctl.consume_input() == ReadResult.Suspended)
                {
                    /* Suspend, come back later */
                    return false;
                }
            }

            m_global_state = JpegState.DSTATE_BUFIMAGE;
            return true;
        }

        /// <summary>
        /// Indicates if we have finished reading the input file.
        /// </summary>
        /// <returns><c>true</c> if we have finished reading the input file.</returns>
        /// <seealso href="6dba59c5-d32e-4dfc-87fe-f9eff7004146.htm" target="_self">Buffered-image mode</seealso>
        public bool jpeg_input_complete()
        {
            /* Check for valid jpeg object */
            if (m_global_state < JpegState.DSTATE_START || m_global_state > JpegState.DSTATE_STOPPING)
                throw new Exception(String.Format("Improper call to JPEG library in state {0}", (int)m_global_state));

            return m_inputctl.EOIReached();
        }

        /// <summary>
        /// Consume data in advance of what the decompressor requires.
        /// </summary>
        /// <returns>The result of data consumption.</returns>
        /// <remarks>This routine can be called at any time after initializing the JPEG object.
        /// It reads some additional data and returns when one of the indicated significant events
        /// occurs. If called after the EOI marker is reached, it will immediately return
        /// <see cref="ReadResult.Reached_EOI"/> without attempting to read more data.</remarks>
        public ReadResult jpeg_consume_input()
        {
            ReadResult retcode = ReadResult.Suspended;

            /* NB: every possible DSTATE value should be listed in this switch */
            switch (m_global_state)
            {
                case JpegState.DSTATE_START:
                    jpeg_consume_input_start();
                    retcode = jpeg_consume_input_inHeader();
                    break;
                case JpegState.DSTATE_INHEADER:
                    retcode = jpeg_consume_input_inHeader();
                    break;
                case JpegState.DSTATE_READY:
                    /* Can't advance past first SOS until start_decompress is called */
                    retcode = ReadResult.Reached_SOS;
                    break;
                case JpegState.DSTATE_PRELOAD:
                case JpegState.DSTATE_PRESCAN:
                case JpegState.DSTATE_SCANNING:
                case JpegState.DSTATE_RAW_OK:
                case JpegState.DSTATE_BUFIMAGE:
                case JpegState.DSTATE_BUFPOST:
                case JpegState.DSTATE_STOPPING:
                    retcode = m_inputctl.consume_input();
                    break;
                default:
                    throw new Exception(String.Format("Improper call to JPEG library in state {0}", (int)m_global_state));
            }
            return retcode;
        }

        /// <summary>
        /// Pre-calculate output image dimensions and related values for current decompression parameters.
        /// </summary>
        /// <remarks>This is allowed for possible use by application. Hence it mustn't do anything 
        /// that can't be done twice. Also note that it may be called before the master module is initialized!
        /// </remarks>
        public void jpeg_calc_output_dimensions()
        {
            // Do computations that are needed before master selection phase
            /* Prevent application from calling me at wrong times */
            if (m_global_state != JpegState.DSTATE_READY)
                throw new Exception(String.Format("Improper call to JPEG library in state {0}", (int)m_global_state));

            /* Compute actual output image dimensions and DCT scaling choices. */
            if (m_scale_num * 8 <= m_scale_denom)
            {
                /* Provide 1/8 scaling */
                m_output_width = JpegUtils.jdiv_round_up(m_image_width, 8);
                m_output_height = JpegUtils.jdiv_round_up(m_image_height, 8);
                m_min_DCT_scaled_size = 1;
            }
            else if (m_scale_num * 4 <= m_scale_denom)
            {
                /* Provide 1/4 scaling */
                m_output_width = JpegUtils.jdiv_round_up(m_image_width, 4);
                m_output_height = JpegUtils.jdiv_round_up(m_image_height, 4);
                m_min_DCT_scaled_size = 2;
            }
            else if (m_scale_num * 2 <= m_scale_denom)
            {
                /* Provide 1/2 scaling */
                m_output_width = JpegUtils.jdiv_round_up(m_image_width, 2);
                m_output_height = JpegUtils.jdiv_round_up(m_image_height, 2);
                m_min_DCT_scaled_size = 4;
            }
            else
            {
                /* Provide 1/1 scaling */
                m_output_width = m_image_width;
                m_output_height = m_image_height;
                m_min_DCT_scaled_size = JpegConstants.DCTSize;
            }

            /* In selecting the actual DCT scaling for each component, we try to
            * scale up the chroma components via IDCT scaling rather than upsampling.
            * This saves time if the upsampler gets to use 1:1 scaling.
            * Note this code assumes that the supported DCT scalings are powers of 2.
            */
            for (int ci = 0; ci < m_num_components; ci++)
            {
                int ssize = m_min_DCT_scaled_size;
                while (ssize < JpegConstants.DCTSize &&
                    (m_comp_info[ci].H_samp_factor * ssize * 2 <= m_max_h_samp_factor * m_min_DCT_scaled_size) &&
                    (m_comp_info[ci].V_samp_factor * ssize * 2 <= m_max_v_samp_factor * m_min_DCT_scaled_size))
                {
                    ssize = ssize * 2;
                }

                m_comp_info[ci].DCT_scaled_size = ssize;
            }

            /* Recompute downsampled dimensions of components;
            * application needs to know these if using raw downsampled data.
            */
            for (int ci = 0; ci < m_num_components; ci++)
            {
                /* Size in samples, after IDCT scaling */
                m_comp_info[ci].downsampled_width = JpegUtils.jdiv_round_up(
                    m_image_width * m_comp_info[ci].H_samp_factor * m_comp_info[ci].DCT_scaled_size,
                    m_max_h_samp_factor * JpegConstants.DCTSize);

                m_comp_info[ci].downsampled_height = JpegUtils.jdiv_round_up(
                    m_image_height * m_comp_info[ci].V_samp_factor * m_comp_info[ci].DCT_scaled_size,
                    m_max_v_samp_factor * JpegConstants.DCTSize);
            }

            /* Report number of components in selected colorspace. */
            /* Probably this should be in the color conversion module... */
            switch (m_out_color_space)
            {
                case ColorSpace.Grayscale:
                    m_out_color_components = 1;
                    break;
                case ColorSpace.RGB:
                case ColorSpace.YCbCr:
                    m_out_color_components = 3;
                    break;
                case ColorSpace.CMYK:
                case ColorSpace.YCCK:
                    m_out_color_components = 4;
                    break;
                default:
                    /* else must be same colorspace as in file */
                    m_out_color_components = m_num_components;
                    break;
            }

            m_output_components = (m_quantize_colors ? 1 : m_out_color_components);

            /* See if up-sampler will want to emit more than one row at a time */
            if (use_merged_upsample())
                m_rec_outbuf_height = m_max_v_samp_factor;
            else
                m_rec_outbuf_height = 1;
        }

        /// <summary>
        /// Read or write the raw DCT coefficient arrays from a JPEG file (useful for lossless transcoding).
        /// </summary>
        /// <returns>Returns <c>null</c> if suspended. This case need be checked only 
        /// if a suspending data source is used.
        /// </returns>
        /// <remarks>
        /// <see cref="JpegDecompressor.jpeg_read_header">jpeg_read_header</see> must be completed before calling this.<br/>
        /// 
        /// The entire image is read into a set of virtual coefficient-block arrays, one per component.
        /// The return value is an array of virtual-array descriptors.<br/>
        /// 
        /// An alternative usage is to simply obtain access to the coefficient arrays during a 
        /// <see href="6dba59c5-d32e-4dfc-87fe-f9eff7004146.htm" target="_self">buffered-image mode</see> decompression operation. This is allowed after any 
        /// <see cref="JpegDecompressor.jpeg_finish_output">jpeg_finish_output</see> call. The arrays can be accessed 
        /// until <see cref="JpegDecompressor.jpeg_finish_decompress">jpeg_finish_decompress</see> is called. 
        /// Note that any call to the library may reposition the arrays, 
        /// so don't rely on <see cref="JpegVirtualArray{T}.Access"/> results to stay valid across library calls.
        /// </remarks>
        public JpegVirtualArray<JpegBlock>[] jpeg_read_coefficients()
        {
            if (m_global_state == JpegState.DSTATE_READY)
            {
                /* First call: initialize active modules */
                transdecode_master_selection();
                m_global_state = JpegState.DSTATE_RDCOEFS;
            }

            if (m_global_state == JpegState.DSTATE_RDCOEFS)
            {
                /* Absorb whole file into the coef buffer */
                for (; ; )
                {
                    ReadResult retcode;

                    /* Absorb some more input */
                    retcode = m_inputctl.consume_input();
                    if (retcode == ReadResult.Suspended)
                        return null;

                    if (retcode == ReadResult.Reached_EOI)
                        break;
                }

                /* Set state so that jpeg_finish_decompress does the right thing */
                m_global_state = JpegState.DSTATE_STOPPING;
            }

            /* At this point we should be in state DSTATE_STOPPING if being used
            * standalone, or in state DSTATE_BUFIMAGE if being invoked to get access
            * to the coefficients during a full buffered-image-mode decompression.
            */
            if ((m_global_state == JpegState.DSTATE_STOPPING || m_global_state == JpegState.DSTATE_BUFIMAGE) && m_buffered_image)
                return m_coef.GetCoefArrays();

            /* Oops, improper usage */
            throw new Exception(String.Format("Improper call to JPEG library in state {0}", (int)m_global_state));
        }

        /// <summary>
        /// Initializes the compression object with default parameters, then copy from the source object 
        /// all parameters needed for lossless transcoding.
        /// </summary>
        /// <param name="dstinfo">Target JPEG compression object.</param>
        /// <remarks>Parameters that can be varied without loss (such as scan script and 
        /// Huffman optimization) are left in their default states.</remarks>
        public void jpeg_copy_critical_parameters(JpegCompressor dstinfo)
        {
            /* Safety check to ensure start_compress not called yet. */
            if (dstinfo.m_global_state != JpegState.CSTATE_START)
                throw new Exception(String.Format("Improper call to JPEG library in state {0}", (int)m_global_state));

            /* Copy fundamental image dimensions */
            dstinfo.m_image_width = m_image_width;
            dstinfo.m_image_height = m_image_height;
            dstinfo.m_input_components = m_num_components;
            dstinfo.m_in_color_space = m_jpeg_color_space;

            /* Initialize all parameters to default values */
            dstinfo.jpeg_set_defaults();

            /* jpeg_set_defaults may choose wrong colorspace, eg YCbCr if input is RGB.
            * Fix it to get the right header markers for the image colorspace.
            */
            dstinfo.jpeg_set_colorspace(m_jpeg_color_space);
            dstinfo.m_data_precision = m_data_precision;
            dstinfo.m_CCIR601_sampling = m_CCIR601_sampling;

            /* Copy the source's quantization tables. */
            for (int tblno = 0; tblno < JpegConstants.NumberOfQuantTables; tblno++)
            {
                if (m_quant_tbl_ptrs[tblno] != null)
                {
                    if (dstinfo.m_quant_tbl_ptrs[tblno] == null)
                        dstinfo.m_quant_tbl_ptrs[tblno] = new JpegQuantizationTable();

                    Buffer.BlockCopy(m_quant_tbl_ptrs[tblno].quantval, 0,
                        dstinfo.m_quant_tbl_ptrs[tblno].quantval, 0,
                        dstinfo.m_quant_tbl_ptrs[tblno].quantval.Length * sizeof(short));

                    dstinfo.m_quant_tbl_ptrs[tblno].Sent_table = false;
                }
            }

            /* Copy the source's per-component info.
            * Note we assume jpeg_set_defaults has allocated the dest comp_info array.
            */
            dstinfo.m_num_components = m_num_components;
            if (dstinfo.m_num_components < 1 || dstinfo.m_num_components > JpegConstants.MaxComponents)
                throw new Exception(String.Format("Too many color components: {0}, max {1}", dstinfo.m_num_components, JpegConstants.MaxComponents));

            for (int ci = 0; ci < dstinfo.m_num_components; ci++)
            {
                dstinfo.Component_info[ci].Component_id = m_comp_info[ci].Component_id;
                dstinfo.Component_info[ci].H_samp_factor = m_comp_info[ci].H_samp_factor;
                dstinfo.Component_info[ci].V_samp_factor = m_comp_info[ci].V_samp_factor;
                dstinfo.Component_info[ci].Quant_tbl_no = m_comp_info[ci].Quant_tbl_no;

                /* Make sure saved quantization table for component matches the qtable
                * slot.  If not, the input file re-used this qtable slot.
                * IJG encoder currently cannot duplicate this.
                */
                int tblno = dstinfo.Component_info[ci].Quant_tbl_no;
                if (tblno < 0 || tblno >= JpegConstants.NumberOfQuantTables || m_quant_tbl_ptrs[tblno] == null)
                    throw new Exception(String.Format("Quantization table 0x{0:X2} was not defined", tblno));

                JpegQuantizationTable c_quant = m_comp_info[ci].quant_table;
                if (c_quant != null)
                {
                    JpegQuantizationTable slot_quant = m_quant_tbl_ptrs[tblno];
                    for (int coefi = 0; coefi < JpegConstants.DCTSize2; coefi++)
                    {
                        if (c_quant.quantval[coefi] != slot_quant.quantval[coefi])
                            throw new Exception(String.Format("Cannot transcode due to multiple use of quantization table {0}", tblno));
                    }
                }
                /* Note: we do not copy the source's Huffman table assignments;
                * instead we rely on jpeg_set_colorspace to have made a suitable choice.
                */
            }

            /* Also copy JFIF version and resolution information, if available.
            * Strictly speaking this isn't "critical" info, but it's nearly
            * always appropriate to copy it if available.  In particular,
            * if the application chooses to copy JFIF 1.02 extension markers from
            * the source file, we need to copy the version to make sure we don't
            * emit a file that has 1.02 extensions but a claimed version of 1.01.
            * We will *not*, however, copy version info from mislabeled "2.01" files.
            */
            if (m_saw_JFIF_marker)
            {
                if (m_JFIF_major_version == 1)
                {
                    dstinfo.m_JFIF_major_version = m_JFIF_major_version;
                    dstinfo.m_JFIF_minor_version = m_JFIF_minor_version;
                }

                dstinfo.m_density_unit = m_density_unit;
                dstinfo.m_X_density = (short)m_X_density;
                dstinfo.m_Y_density = (short)m_Y_density;
            }
        }

        /// <summary>
        /// Aborts processing of a JPEG decompression operation.
        /// </summary>
        /// <seealso cref="JpegCommonBase.jpeg_abort"/>
        public void jpeg_abort_decompress()
        {
            jpeg_abort();
        }

        /// <summary>
        /// Sets processor for special marker.
        /// </summary>
        /// <param name="marker_code">The marker code.</param>
        /// <param name="routine">The processor.</param>
        /// <remarks>Allows you to supply your own routine to process 
        /// COM and/or APPn markers on-the-fly as they are read.
        /// </remarks>
        /// <seealso href="81c88818-a5d7-4550-9ce5-024a768f7b1e.htm" target="_self">Special markers</seealso>
        public void jpeg_set_marker_processor(int marker_code, jpeg_marker_parser_method routine)
        {
            m_marker.jpeg_set_marker_processor(marker_code, routine);
        }

        /// <summary>
        /// Control saving of COM and APPn markers into <see cref="JpegDecompressor.Marker_list">Marker_list</see>.
        /// </summary>
        /// <param name="marker_code">The marker type to save (see JpegMarkerType enumeration).<br/>
        /// To arrange to save all the special marker types, you need to call this 
        /// routine 17 times, for COM and APP0-APP15 markers.</param>
        /// <param name="length_limit">If the incoming marker is longer than <c>length_limit</c> data bytes, 
        /// only <c>length_limit</c> bytes will be saved; this parameter allows you to avoid chewing up memory 
        /// when you only need to see the first few bytes of a potentially large marker. If you want to save 
        /// all the data, set <c>length_limit</c> to 0xFFFF; that is enough since marker lengths are only 16 bits. 
        /// As a special case, setting <c>length_limit</c> to 0 prevents that marker type from being saved at all. 
        /// (That is the default behavior, in fact.)
        /// </param>
        /// <seealso cref="JpegDecompressor.Marker_list"/>
        /// <seealso href="81c88818-a5d7-4550-9ce5-024a768f7b1e.htm" target="_self">Special markers</seealso>
        public void jpeg_save_markers(int marker_code, int length_limit)
        {
            m_marker.jpeg_save_markers(marker_code, length_limit);
        }

        /// <summary>
        /// Determine whether merged upsample/color conversion should be used.
        /// CRUCIAL: this must match the actual capabilities of merged upsampler!
        /// </summary>
        internal bool use_merged_upsample()
        {
            /* Merging is the equivalent of plain box-filter upsampling */
            if (m_do_fancy_upsampling || m_CCIR601_sampling)
                return false;

            /* UpsamplerImpl only supports YCC=>RGB color conversion */
            if (m_jpeg_color_space != ColorSpace.YCbCr || m_num_components != 3 ||
                m_out_color_space != ColorSpace.RGB || m_out_color_components != JpegConstants.RGB_PixelLength)
            {
                return false;
            }

            /* and it only handles 2h1v or 2h2v sampling ratios */
            if (m_comp_info[0].H_samp_factor != 2 || m_comp_info[1].H_samp_factor != 1 ||
                m_comp_info[2].H_samp_factor != 1 || m_comp_info[0].V_samp_factor > 2 ||
                m_comp_info[1].V_samp_factor != 1 || m_comp_info[2].V_samp_factor != 1)
            {
                return false;
            }

            /* furthermore, it doesn't work if we've scaled the IDCTs differently */
            if (m_comp_info[0].DCT_scaled_size != m_min_DCT_scaled_size ||
                m_comp_info[1].DCT_scaled_size != m_min_DCT_scaled_size ||
                m_comp_info[2].DCT_scaled_size != m_min_DCT_scaled_size)
            {
                return false;
            }

            /* ??? also need to test for upsample-time rescaling, when & if supported */
            /* by golly, it'll work... */
            return true;
        }

        /// <summary>
        /// Initialization of JPEG compression objects.
        /// The error manager must already be set up (in case memory manager fails).
        /// </summary>
        private void initialize()
        {
            /* Zero out pointers to permanent structures. */
            m_src = null;

            for (int i = 0; i < JpegConstants.NumberOfQuantTables; i++)
                m_quant_tbl_ptrs[i] = null;

            for (int i = 0; i < JpegConstants.NumberOfHuffmanTables; i++)
            {
                m_dc_huff_tbl_ptrs[i] = null;
                m_ac_huff_tbl_ptrs[i] = null;
            }

            /* Initialize marker processor so application can override methods
            * for COM, APPn markers before calling jpeg_read_header.
            */
            m_marker_list = new List<JpegMarker>();
            m_marker = new JpegMarkerReader(this);

            /* And initialize the overall input controller. */
            m_inputctl = new JpegInputController(this);

            /* OK, I'm ready */
            m_global_state = JpegState.DSTATE_START;
        }

        /// <summary>
        /// Master selection of decompression modules for transcoding (that is, reading 
        /// raw DCT coefficient arrays from an input JPEG file.)
        /// This substitutes for initialization of the full decompressor.
        /// </summary>
        private void transdecode_master_selection()
        {
            /* This is effectively a buffered-image operation. */
            m_buffered_image = true;

            if (m_progressive_mode)
                m_entropy = new ProgressiveHuffmanDecoder(this);
            else
                m_entropy = new HuffEntropyDecoder(this);

            /* Always get a full-image coefficient buffer. */
            m_coef = new JpegDecompressorCoefController(this, true);

            /* Initialize input side of decompressor to consume first scan. */
            m_inputctl.start_input_pass();
        }

        /// <summary>
        /// Set up for an output pass, and perform any dummy pass(es) needed.
        /// Common subroutine for jpeg_start_decompress and jpeg_start_output.
        /// Entry: global_state = DSTATE_PRESCAN only if previously suspended.
        /// Exit: If done, returns true and sets global_state for proper output mode.
        ///       If suspended, returns false and sets global_state = DSTATE_PRESCAN.
        /// </summary>
        private bool output_pass_setup()
        {
            if (m_global_state != JpegState.DSTATE_PRESCAN)
            {
                /* First call: do pass setup */
                m_master.prepare_for_output_pass();
                m_output_scanline = 0;
                m_global_state = JpegState.DSTATE_PRESCAN;
            }

            /* Loop over any required dummy passes */
            while (m_master.IsDummyPass())
            {
                /* Crank through the dummy pass */
                while (m_output_scanline < m_output_height)
                {
                    int last_scanline;

                    /* Process some data */
                    last_scanline = m_output_scanline;
                    m_main.process_data(null, ref m_output_scanline, 0);
                    if (m_output_scanline == last_scanline)
                    {
                        /* No progress made, must suspend */
                        return false;
                    }
                }

                /* Finish up dummy pass, and set up for another one */
                m_master.finish_output_pass();
                m_master.prepare_for_output_pass();
                m_output_scanline = 0;
            }

            /* Ready for application to drive output pass through
            * jpeg_read_scanlines or jpeg_read_raw_data.
            */
            m_global_state = m_raw_data_out ? JpegState.DSTATE_RAW_OK : JpegState.DSTATE_SCANNING;
            return true;
        }

        /// <summary>
        /// Set default decompression parameters.
        /// </summary>
        private void default_decompress_parms()
        {
            /* Guess the input colorspace, and set output colorspace accordingly. */
            /* (Wish JPEG committee had provided a real way to specify this...) */
            /* Note application may override our guesses. */
            switch (m_num_components)
            {
                case 1:
                    m_jpeg_color_space = ColorSpace.Grayscale;
                    m_out_color_space = ColorSpace.Grayscale;
                    break;

                case 3:
                    if (m_saw_JFIF_marker)
                    {
                        /* JFIF implies YCbCr */
                        m_jpeg_color_space = ColorSpace.YCbCr;
                    }
                    else if (m_saw_Adobe_marker)
                    {
                        switch (m_Adobe_transform)
                        {
                            case 0:
                                m_jpeg_color_space = ColorSpace.RGB;
                                break;
                            case 1:
                                m_jpeg_color_space = ColorSpace.YCbCr;
                                break;
                            default:
                                m_jpeg_color_space = ColorSpace.YCbCr; /* assume it's YCbCr */
                                break;
                        }
                    }
                    else
                    {
                        /* Saw no special markers, try to guess from the component IDs */
                        int cid0 = m_comp_info[0].Component_id;
                        int cid1 = m_comp_info[1].Component_id;
                        int cid2 = m_comp_info[2].Component_id;

                        if (cid0 == 1 && cid1 == 2 && cid2 == 3)
                        {
                            /* assume JFIF w/out marker */
                            m_jpeg_color_space = ColorSpace.YCbCr;
                        }
                        else if (cid0 == 82 && cid1 == 71 && cid2 == 66)
                        {
                            /* ASCII 'R', 'G', 'B' */
                            m_jpeg_color_space = ColorSpace.RGB;
                        }
                        else
                        {
                            /* assume it's YCbCr */
                            m_jpeg_color_space = ColorSpace.YCbCr;
                        }
                    }
                    /* Always guess RGB is proper output colorspace. */
                    m_out_color_space = ColorSpace.RGB;
                    break;

                case 4:
                    if (m_saw_Adobe_marker)
                    {
                        switch (m_Adobe_transform)
                        {
                            case 0:
                                m_jpeg_color_space = ColorSpace.CMYK;
                                break;
                            case 2:
                                m_jpeg_color_space = ColorSpace.YCCK;
                                break;
                            default:
                                /* assume it's YCCK */
                                m_jpeg_color_space = ColorSpace.YCCK;
                                break;
                        }
                    }
                    else
                    {
                        /* No special markers, assume straight CMYK. */
                        m_jpeg_color_space = ColorSpace.CMYK;
                    }

                    m_out_color_space = ColorSpace.CMYK;
                    break;

                default:
                    m_jpeg_color_space = ColorSpace.Unknown;
                    m_out_color_space = ColorSpace.Unknown;
                    break;
            }

            /* Set defaults for other decompression parameters. */
            m_scale_num = 1;       /* 1:1 scaling */
            m_scale_denom = 1;
            m_buffered_image = false;
            m_raw_data_out = false;
            m_dct_method = JpegConstants.DefaultDCTMethod;
            m_do_fancy_upsampling = true;
            m_do_block_smoothing = true;
            m_quantize_colors = false;

            /* We set these in case application only sets quantize_colors. */
            m_dither_mode = DitherMode.FloydStein;
            m_two_pass_quantize = true;
            m_desired_number_of_colors = 256;
            m_colormap = null;

            /* Initialize for no mode change in buffered-image mode. */
            m_enable_1pass_quant = false;
            m_enable_external_quant = false;
            m_enable_2pass_quant = false;
        }

        private void jpeg_consume_input_start()
        {
            /* Start-of-datastream actions: reset appropriate modules */
            m_inputctl.reset_input_controller();

            /* Initialize application's data source module */
            m_src.init_source();
            m_global_state = JpegState.DSTATE_INHEADER;
        }

        private ReadResult jpeg_consume_input_inHeader()
        {
            ReadResult retcode = m_inputctl.consume_input();
            if (retcode == ReadResult.Reached_SOS)
            {
                /* Found SOS, prepare to decompress */
                /* Set up default parameters based on header data */
                default_decompress_parms();

                /* Set global state: ready for start_decompress */
                m_global_state = JpegState.DSTATE_READY;
            }

            return retcode;
        }
    }
    #endregion

    #region JpegDecompressorCoefController
    /// <summary>
    /// Coefficient buffer control
    /// 
    /// This code applies interblock smoothing as described by section K.8
    /// of the JPEG standard: the first 5 AC coefficients are estimated from
    /// the DC values of a DCT block and its 8 neighboring blocks.
    /// We apply smoothing only for progressive JPEG decoding, and only if
    /// the coefficients it can estimate are not yet known to full precision.
    /// </summary>
    class JpegDecompressorCoefController
    {
        private const int SAVED_COEFS = 6; /* we save coef_bits[0..5] */

        /* Natural-order array positions of the first 5 zigzag-order coefficients */
        private const int Q01_POS = 1;
        private const int Q10_POS = 8;
        private const int Q20_POS = 16;
        private const int Q11_POS = 9;
        private const int Q02_POS = 2;

        private enum DecompressorType
        {
            Ordinary,
            Smooth,
            OnePass
        }

        private JpegDecompressor m_cinfo;
        private bool m_useDummyConsumeData;
        private DecompressorType m_decompressor;

        /* These variables keep track of the current location of the input side. */
        /* cinfo.input_iMCU_row is also used for this. */
        private int m_MCU_ctr;     /* counts MCUs processed in current row */
        private int m_MCU_vert_offset;        /* counts MCU rows within iMCU row */
        private int m_MCU_rows_per_iMCU_row;  /* number of such rows needed */

        /* The output side's location is represented by cinfo.output_iMCU_row. */

        /* In single-pass modes, it's sufficient to buffer just one MCU.
        * We allocate a workspace of DecompressorMaxBlocksInMCU coefficient blocks,
        * and let the entropy decoder write into that workspace each time.
        * (On 80x86, the workspace is FAR even though it's not really very big;
        * this is to keep the module interfaces unchanged when a large coefficient
        * buffer is necessary.)
        * In multi-pass modes, this array points to the current MCU's blocks
        * within the virtual arrays; it is used only by the input side.
        */
        private JpegBlock[] m_MCU_buffer = new JpegBlock[JpegConstants.DecompressorMaxBlocksInMCU];

        /* In multi-pass modes, we need a virtual block array for each component. */
        private JpegVirtualArray<JpegBlock>[] m_whole_image = new JpegVirtualArray<JpegBlock>[JpegConstants.MaxComponents];
        private JpegVirtualArray<JpegBlock>[] m_coef_arrays;

        /* When doing block smoothing, we latch coefficient Al values here */
        private int[] m_coef_bits_latch;
        private int m_coef_bits_savedOffset;

        public JpegDecompressorCoefController(JpegDecompressor cinfo, bool need_full_buffer)
        {
            m_cinfo = cinfo;

            /* Create the coefficient buffer. */
            if (need_full_buffer)
            {
                /* Allocate a full-image virtual array for each component, */
                /* padded to a multiple of samp_factor DCT blocks in each direction. */
                /* Note we ask for a pre-zeroed array. */
                for (int ci = 0; ci < cinfo.m_num_components; ci++)
                {
                    m_whole_image[ci] = JpegCommonBase.CreateBlocksArray(
                        JpegUtils.jround_up(cinfo.Comp_info[ci].Width_in_blocks, cinfo.Comp_info[ci].H_samp_factor),
                        JpegUtils.jround_up(cinfo.Comp_info[ci].height_in_blocks, cinfo.Comp_info[ci].V_samp_factor));
                    m_whole_image[ci].ErrorProcessor = cinfo;
                }

                m_useDummyConsumeData = false;
                m_decompressor = DecompressorType.Ordinary;
                m_coef_arrays = m_whole_image; /* link to virtual arrays */
            }
            else
            {
                /* We only need a single-MCU buffer. */
                JpegBlock[] buffer = new JpegBlock[JpegConstants.DecompressorMaxBlocksInMCU];
                for (int i = 0; i < JpegConstants.DecompressorMaxBlocksInMCU; i++)
                {
                    buffer[i] = new JpegBlock();
                    for (int ii = 0; ii < buffer[i].data.Length; ii++)
                        buffer[i].data[ii] = -12851;

                    m_MCU_buffer[i] = buffer[i];
                }

                m_useDummyConsumeData = true;
                m_decompressor = DecompressorType.OnePass;
                m_coef_arrays = null; /* flag for no virtual arrays */
            }
        }

        /// <summary>
        /// Initialize for an input processing pass.
        /// </summary>
        public void start_input_pass()
        {
            m_cinfo.m_input_iMCU_row = 0;
            start_iMCU_row();
        }

        /// <summary>
        /// Consume input data and store it in the full-image coefficient buffer.
        /// We read as much as one fully interleaved MCU row ("iMCU" row) per call,
        /// ie, v_samp_factor block rows for each component in the scan.
        /// </summary>
        public ReadResult consume_data()
        {
            if (m_useDummyConsumeData)
                return ReadResult.Suspended;  /* Always indicate nothing was done */

            JpegBlock[][][] buffer = new JpegBlock[JpegConstants.MaxComponentsInScan][][];

            /* Align the virtual buffers for the components used in this scan. */
            for (int ci = 0; ci < m_cinfo.m_comps_in_scan; ci++)
            {
                JpegComponent componentInfo = m_cinfo.Comp_info[m_cinfo.m_cur_comp_info[ci]];

                buffer[ci] = m_whole_image[componentInfo.Component_index].Access(
                    m_cinfo.m_input_iMCU_row * componentInfo.V_samp_factor, componentInfo.V_samp_factor);

                /* Note: entropy decoder expects buffer to be zeroed,
                 * but this is handled automatically by the memory manager
                 * because we requested a pre-zeroed array.
                 */
            }

            /* Loop to process one whole iMCU row */
            for (int yoffset = m_MCU_vert_offset; yoffset < m_MCU_rows_per_iMCU_row; yoffset++)
            {
                for (int MCU_col_num = m_MCU_ctr; MCU_col_num < m_cinfo.m_MCUs_per_row; MCU_col_num++)
                {
                    /* Construct list of pointers to DCT blocks belonging to this MCU */
                    int blkn = 0;           /* index of current DCT block within MCU */
                    for (int ci = 0; ci < m_cinfo.m_comps_in_scan; ci++)
                    {
                        JpegComponent componentInfo = m_cinfo.Comp_info[m_cinfo.m_cur_comp_info[ci]];
                        int start_col = MCU_col_num * componentInfo.MCU_width;
                        for (int yindex = 0; yindex < componentInfo.MCU_height; yindex++)
                        {
                            for (int xindex = 0; xindex < componentInfo.MCU_width; xindex++)
                            {
                                m_MCU_buffer[blkn] = buffer[ci][yindex + yoffset][start_col + xindex];
                                blkn++;
                            }
                        }
                    }

                    /* Try to fetch the MCU. */
                    if (!m_cinfo.m_entropy.decode_mcu(m_MCU_buffer))
                    {
                        /* Suspension forced; update state counters and exit */
                        m_MCU_vert_offset = yoffset;
                        m_MCU_ctr = MCU_col_num;
                        return ReadResult.Suspended;
                    }
                }

                /* Completed an MCU row, but perhaps not an iMCU row */
                m_MCU_ctr = 0;
            }

            /* Completed the iMCU row, advance counters for next one */
            m_cinfo.m_input_iMCU_row++;
            if (m_cinfo.m_input_iMCU_row < m_cinfo.m_total_iMCU_rows)
            {
                start_iMCU_row();
                return ReadResult.Row_Completed;
            }

            /* Completed the scan */
            m_cinfo.m_inputctl.finish_input_pass();
            return ReadResult.Scan_Completed;
        }

        /// <summary>
        /// Initialize for an output processing pass.
        /// </summary>
        public void start_output_pass()
        {
            /* If multipass, check to see whether to use block smoothing on this pass */
            if (m_coef_arrays != null)
            {
                if (m_cinfo.m_do_block_smoothing && smoothing_ok())
                    m_decompressor = DecompressorType.Smooth;
                else
                    m_decompressor = DecompressorType.Ordinary;
            }

            m_cinfo.m_output_iMCU_row = 0;
        }

        public ReadResult decompress_data(ComponentBuffer[] output_buf)
        {
            switch (m_decompressor)
            {
                case DecompressorType.Ordinary:
                    return decompress_data_ordinary(output_buf);

                case DecompressorType.Smooth:
                    return decompress_smooth_data(output_buf);

                case DecompressorType.OnePass:
                    return decompress_onepass(output_buf);
            }

            throw new Exception("Not implemented yet");
        }

        /* Pointer to array of coefficient virtual arrays, or null if none */
        public JpegVirtualArray<JpegBlock>[] GetCoefArrays()
        {
            return m_coef_arrays;
        }

        /// <summary>
        /// Decompress and return some data in the single-pass case.
        /// Always attempts to emit one fully interleaved MCU row ("iMCU" row).
        /// Input and output must run in lockstep since we have only a one-MCU buffer.
        /// Return value is JPEG_ROW_COMPLETED, JPEG_SCAN_COMPLETED, or JPEG_SUSPENDED.
        /// 
        /// NB: output_buf contains a plane for each component in image,
        /// which we index according to the component's SOF position.
        /// </summary>
        private ReadResult decompress_onepass(ComponentBuffer[] output_buf)
        {
            int last_MCU_col = m_cinfo.m_MCUs_per_row - 1;
            int last_iMCU_row = m_cinfo.m_total_iMCU_rows - 1;

            /* Loop to process as much as one whole iMCU row */
            for (int yoffset = m_MCU_vert_offset; yoffset < m_MCU_rows_per_iMCU_row; yoffset++)
            {
                for (int MCU_col_num = m_MCU_ctr; MCU_col_num <= last_MCU_col; MCU_col_num++)
                {
                    /* Try to fetch an MCU.  Entropy decoder expects buffer to be zeroed. */
                    for (int i = 0; i < m_cinfo.m_blocks_in_MCU; i++)
                        Array.Clear(m_MCU_buffer[i].data, 0, m_MCU_buffer[i].data.Length);

                    if (!m_cinfo.m_entropy.decode_mcu(m_MCU_buffer))
                    {
                        /* Suspension forced; update state counters and exit */
                        m_MCU_vert_offset = yoffset;
                        m_MCU_ctr = MCU_col_num;
                        return ReadResult.Suspended;
                    }

                    /* Determine where data should go in output_buf and do the IDCT thing.
                     * We skip dummy blocks at the right and bottom edges (but blkn gets
                     * incremented past them!).  Note the inner loop relies on having
                     * allocated the MCU_buffer[] blocks sequentially.
                     */
                    int blkn = 0;           /* index of current DCT block within MCU */
                    for (int ci = 0; ci < m_cinfo.m_comps_in_scan; ci++)
                    {
                        JpegComponent componentInfo = m_cinfo.Comp_info[m_cinfo.m_cur_comp_info[ci]];

                        /* Don't bother to IDCT an uninteresting component. */
                        if (!componentInfo.component_needed)
                        {
                            blkn += componentInfo.MCU_blocks;
                            continue;
                        }

                        int useful_width = (MCU_col_num < last_MCU_col) ? componentInfo.MCU_width : componentInfo.last_col_width;
                        int outputIndex = yoffset * componentInfo.DCT_scaled_size;
                        int start_col = MCU_col_num * componentInfo.MCU_sample_width;
                        for (int yindex = 0; yindex < componentInfo.MCU_height; yindex++)
                        {
                            if (m_cinfo.m_input_iMCU_row < last_iMCU_row || yoffset + yindex < componentInfo.last_row_height)
                            {
                                int output_col = start_col;
                                for (int xindex = 0; xindex < useful_width; xindex++)
                                {
                                    m_cinfo.m_idct.inverse(componentInfo.Component_index,
                                        m_MCU_buffer[blkn + xindex].data, output_buf[componentInfo.Component_index],
                                        outputIndex, output_col);

                                    output_col += componentInfo.DCT_scaled_size;
                                }
                            }

                            blkn += componentInfo.MCU_width;
                            outputIndex += componentInfo.DCT_scaled_size;
                        }
                    }
                }

                /* Completed an MCU row, but perhaps not an iMCU row */
                m_MCU_ctr = 0;
            }

            /* Completed the iMCU row, advance counters for next one */
            m_cinfo.m_output_iMCU_row++;
            m_cinfo.m_input_iMCU_row++;
            if (m_cinfo.m_input_iMCU_row < m_cinfo.m_total_iMCU_rows)
            {
                start_iMCU_row();
                return ReadResult.Row_Completed;
            }

            /* Completed the scan */
            m_cinfo.m_inputctl.finish_input_pass();
            return ReadResult.Scan_Completed;
        }

        /// <summary>
        /// Decompress and return some data in the multi-pass case.
        /// Always attempts to emit one fully interleaved MCU row ("iMCU" row).
        /// Return value is JPEG_ROW_COMPLETED, JPEG_SCAN_COMPLETED, or JPEG_SUSPENDED.
        /// 
        /// NB: output_buf contains a plane for each component in image.
        /// </summary>
        private ReadResult decompress_data_ordinary(ComponentBuffer[] output_buf)
        {
            /* Force some input to be done if we are getting ahead of the input. */
            while (m_cinfo.m_input_scan_number < m_cinfo.m_output_scan_number ||
                   (m_cinfo.m_input_scan_number == m_cinfo.m_output_scan_number &&
                    m_cinfo.m_input_iMCU_row <= m_cinfo.m_output_iMCU_row))
            {
                if (m_cinfo.m_inputctl.consume_input() == ReadResult.Suspended)
                    return ReadResult.Suspended;
            }

            int last_iMCU_row = m_cinfo.m_total_iMCU_rows - 1;

            /* OK, output from the virtual arrays. */
            for (int ci = 0; ci < m_cinfo.m_num_components; ci++)
            {
                JpegComponent componentInfo = m_cinfo.Comp_info[ci];

                /* Don't bother to IDCT an uninteresting component. */
                if (!componentInfo.component_needed)
                    continue;

                /* Align the virtual buffer for this component. */
                JpegBlock[][] buffer = m_whole_image[ci].Access(m_cinfo.m_output_iMCU_row * componentInfo.V_samp_factor,
                    componentInfo.V_samp_factor);

                /* Count non-dummy DCT block rows in this iMCU row. */
                int block_rows;
                if (m_cinfo.m_output_iMCU_row < last_iMCU_row)
                    block_rows = componentInfo.V_samp_factor;
                else
                {
                    /* NB: can't use last_row_height here; it is input-side-dependent! */
                    block_rows = componentInfo.height_in_blocks % componentInfo.V_samp_factor;
                    if (block_rows == 0)
                        block_rows = componentInfo.V_samp_factor;
                }

                /* Loop over all DCT blocks to be processed. */
                int rowIndex = 0;
                for (int block_row = 0; block_row < block_rows; block_row++)
                {
                    int output_col = 0;
                    for (int block_num = 0; block_num < componentInfo.Width_in_blocks; block_num++)
                    {
                        m_cinfo.m_idct.inverse(componentInfo.Component_index,
                            buffer[block_row][block_num].data, output_buf[ci], rowIndex, output_col);

                        output_col += componentInfo.DCT_scaled_size;
                    }

                    rowIndex += componentInfo.DCT_scaled_size;
                }
            }

            m_cinfo.m_output_iMCU_row++;
            if (m_cinfo.m_output_iMCU_row < m_cinfo.m_total_iMCU_rows)
                return ReadResult.Row_Completed;

            return ReadResult.Scan_Completed;
        }

        /// <summary>
        /// Variant of decompress_data for use when doing block smoothing.
        /// </summary>
        private ReadResult decompress_smooth_data(ComponentBuffer[] output_buf)
        {
            /* Force some input to be done if we are getting ahead of the input. */
            while (m_cinfo.m_input_scan_number <= m_cinfo.m_output_scan_number && !m_cinfo.m_inputctl.EOIReached())
            {
                if (m_cinfo.m_input_scan_number == m_cinfo.m_output_scan_number)
                {
                    /* If input is working on current scan, we ordinarily want it to
                     * have completed the current row.  But if input scan is DC,
                     * we want it to keep one row ahead so that next block row's DC
                     * values are up to date.
                     */
                    int delta = (m_cinfo.m_Ss == 0) ? 1 : 0;
                    if (m_cinfo.m_input_iMCU_row > m_cinfo.m_output_iMCU_row + delta)
                        break;
                }

                if (m_cinfo.m_inputctl.consume_input() == ReadResult.Suspended)
                    return ReadResult.Suspended;
            }

            int last_iMCU_row = m_cinfo.m_total_iMCU_rows - 1;

            /* OK, output from the virtual arrays. */
            for (int ci = 0; ci < m_cinfo.m_num_components; ci++)
            {
                JpegComponent componentInfo = m_cinfo.Comp_info[ci];

                /* Don't bother to IDCT an uninteresting component. */
                if (!componentInfo.component_needed)
                    continue;

                int block_rows;
                int access_rows;
                bool last_row;
                /* Count non-dummy DCT block rows in this iMCU row. */
                if (m_cinfo.m_output_iMCU_row < last_iMCU_row)
                {
                    block_rows = componentInfo.V_samp_factor;
                    access_rows = block_rows * 2; /* this and next iMCU row */
                    last_row = false;
                }
                else
                {
                    /* NB: can't use last_row_height here; it is input-side-dependent! */
                    block_rows = componentInfo.height_in_blocks % componentInfo.V_samp_factor;
                    if (block_rows == 0)
                        block_rows = componentInfo.V_samp_factor;
                    access_rows = block_rows; /* this iMCU row only */
                    last_row = true;
                }

                /* Align the virtual buffer for this component. */
                JpegBlock[][] buffer = null;
                bool first_row;
                int bufferRowOffset = 0;
                if (m_cinfo.m_output_iMCU_row > 0)
                {
                    access_rows += componentInfo.V_samp_factor; /* prior iMCU row too */
                    buffer = m_whole_image[ci].Access((m_cinfo.m_output_iMCU_row - 1) * componentInfo.V_samp_factor, access_rows);
                    bufferRowOffset = componentInfo.V_samp_factor; /* point to current iMCU row */
                    first_row = false;
                }
                else
                {
                    buffer = m_whole_image[ci].Access(0, access_rows);
                    first_row = true;
                }

                /* Fetch component-dependent info */
                int coefBitsOffset = ci * SAVED_COEFS;
                int Q00 = componentInfo.quant_table.quantval[0];
                int Q01 = componentInfo.quant_table.quantval[Q01_POS];
                int Q10 = componentInfo.quant_table.quantval[Q10_POS];
                int Q20 = componentInfo.quant_table.quantval[Q20_POS];
                int Q11 = componentInfo.quant_table.quantval[Q11_POS];
                int Q02 = componentInfo.quant_table.quantval[Q02_POS];
                int outputIndex = ci;

                /* Loop over all DCT blocks to be processed. */
                for (int block_row = 0; block_row < block_rows; block_row++)
                {
                    int bufferIndex = bufferRowOffset + block_row;

                    int prev_block_row = 0;
                    if (first_row && block_row == 0)
                        prev_block_row = bufferIndex;
                    else
                        prev_block_row = bufferIndex - 1;

                    int next_block_row = 0;
                    if (last_row && block_row == block_rows - 1)
                        next_block_row = bufferIndex;
                    else
                        next_block_row = bufferIndex + 1;

                    /* We fetch the surrounding DC values using a sliding-register approach.
                     * Initialize all nine here so as to do the right thing on narrow pics.
                     */
                    int DC1 = buffer[prev_block_row][0][0];
                    int DC2 = DC1;
                    int DC3 = DC1;

                    int DC4 = buffer[bufferIndex][0][0];
                    int DC5 = DC4;
                    int DC6 = DC4;

                    int DC7 = buffer[next_block_row][0][0];
                    int DC8 = DC7;
                    int DC9 = DC7;

                    int output_col = 0;
                    int last_block_column = componentInfo.Width_in_blocks - 1;
                    for (int block_num = 0; block_num <= last_block_column; block_num++)
                    {
                        /* Fetch current DCT block into workspace so we can modify it. */
                        JpegBlock workspace = new JpegBlock();
                        Buffer.BlockCopy(buffer[bufferIndex][0].data, 0, workspace.data, 0, workspace.data.Length * sizeof(short));

                        /* Update DC values */
                        if (block_num < last_block_column)
                        {
                            DC3 = buffer[prev_block_row][1][0];
                            DC6 = buffer[bufferIndex][1][0];
                            DC9 = buffer[next_block_row][1][0];
                        }

                        /* Compute coefficient estimates per K.8.
                         * An estimate is applied only if coefficient is still zero,
                         * and is not known to be fully accurate.
                         */
                        /* AC01 */
                        int Al = m_coef_bits_latch[m_coef_bits_savedOffset + coefBitsOffset + 1];
                        if (Al != 0 && workspace[1] == 0)
                        {
                            int pred;
                            int num = 36 * Q00 * (DC4 - DC6);
                            if (num >= 0)
                            {
                                pred = ((Q01 << 7) + num) / (Q01 << 8);
                                if (Al > 0 && pred >= (1 << Al))
                                    pred = (1 << Al) - 1;
                            }
                            else
                            {
                                pred = ((Q01 << 7) - num) / (Q01 << 8);
                                if (Al > 0 && pred >= (1 << Al))
                                    pred = (1 << Al) - 1;
                                pred = -pred;
                            }
                            workspace[1] = (short)pred;
                        }

                        /* AC10 */
                        Al = m_coef_bits_latch[m_coef_bits_savedOffset + coefBitsOffset + 2];
                        if (Al != 0 && workspace[8] == 0)
                        {
                            int pred;
                            int num = 36 * Q00 * (DC2 - DC8);
                            if (num >= 0)
                            {
                                pred = ((Q10 << 7) + num) / (Q10 << 8);
                                if (Al > 0 && pred >= (1 << Al))
                                    pred = (1 << Al) - 1;
                            }
                            else
                            {
                                pred = ((Q10 << 7) - num) / (Q10 << 8);
                                if (Al > 0 && pred >= (1 << Al))
                                    pred = (1 << Al) - 1;
                                pred = -pred;
                            }
                            workspace[8] = (short)pred;
                        }

                        /* AC20 */
                        Al = m_coef_bits_latch[m_coef_bits_savedOffset + coefBitsOffset + 3];
                        if (Al != 0 && workspace[16] == 0)
                        {
                            int pred;
                            int num = 9 * Q00 * (DC2 + DC8 - 2 * DC5);
                            if (num >= 0)
                            {
                                pred = ((Q20 << 7) + num) / (Q20 << 8);
                                if (Al > 0 && pred >= (1 << Al))
                                    pred = (1 << Al) - 1;
                            }
                            else
                            {
                                pred = ((Q20 << 7) - num) / (Q20 << 8);
                                if (Al > 0 && pred >= (1 << Al))
                                    pred = (1 << Al) - 1;
                                pred = -pred;
                            }
                            workspace[16] = (short)pred;
                        }

                        /* AC11 */
                        Al = m_coef_bits_latch[m_coef_bits_savedOffset + coefBitsOffset + 4];
                        if (Al != 0 && workspace[9] == 0)
                        {
                            int pred;
                            int num = 5 * Q00 * (DC1 - DC3 - DC7 + DC9);
                            if (num >= 0)
                            {
                                pred = ((Q11 << 7) + num) / (Q11 << 8);
                                if (Al > 0 && pred >= (1 << Al))
                                    pred = (1 << Al) - 1;
                            }
                            else
                            {
                                pred = ((Q11 << 7) - num) / (Q11 << 8);
                                if (Al > 0 && pred >= (1 << Al))
                                    pred = (1 << Al) - 1;
                                pred = -pred;
                            }
                            workspace[9] = (short)pred;
                        }

                        /* AC02 */
                        Al = m_coef_bits_latch[m_coef_bits_savedOffset + coefBitsOffset + 5];
                        if (Al != 0 && workspace[2] == 0)
                        {
                            int pred;
                            int num = 9 * Q00 * (DC4 + DC6 - 2 * DC5);
                            if (num >= 0)
                            {
                                pred = ((Q02 << 7) + num) / (Q02 << 8);
                                if (Al > 0 && pred >= (1 << Al))
                                    pred = (1 << Al) - 1;
                            }
                            else
                            {
                                pred = ((Q02 << 7) - num) / (Q02 << 8);
                                if (Al > 0 && pred >= (1 << Al))
                                    pred = (1 << Al) - 1;
                                pred = -pred;
                            }
                            workspace[2] = (short)pred;
                        }

                        /* OK, do the IDCT */
                        m_cinfo.m_idct.inverse(componentInfo.Component_index, workspace.data, output_buf[outputIndex], 0, output_col);

                        /* Advance for next column */
                        DC1 = DC2;
                        DC2 = DC3;
                        DC4 = DC5;
                        DC5 = DC6;
                        DC7 = DC8;
                        DC8 = DC9;

                        bufferIndex++;
                        prev_block_row++;
                        next_block_row++;

                        output_col += componentInfo.DCT_scaled_size;
                    }

                    outputIndex += componentInfo.DCT_scaled_size;
                }
            }

            m_cinfo.m_output_iMCU_row++;
            if (m_cinfo.m_output_iMCU_row < m_cinfo.m_total_iMCU_rows)
                return ReadResult.Row_Completed;

            return ReadResult.Scan_Completed;
        }

        /// <summary>
        /// Determine whether block smoothing is applicable and safe.
        /// We also latch the current states of the coef_bits[] entries for the
        /// AC coefficients; otherwise, if the input side of the decompressor
        /// advances into a new scan, we might think the coefficients are known
        /// more accurately than they really are.
        /// </summary>
        private bool smoothing_ok()
        {
            if (!m_cinfo.m_progressive_mode || m_cinfo.m_coef_bits == null)
                return false;

            /* Allocate latch area if not already done */
            if (m_coef_bits_latch == null)
            {
                m_coef_bits_latch = new int[m_cinfo.m_num_components * SAVED_COEFS];
                m_coef_bits_savedOffset = 0;
            }

            bool smoothing_useful = false;
            for (int ci = 0; ci < m_cinfo.m_num_components; ci++)
            {
                /* All components' quantization values must already be latched. */
                JpegQuantizationTable qtable = m_cinfo.Comp_info[ci].quant_table;
                if (qtable == null)
                    return false;

                /* Verify DC & first 5 AC quantizers are nonzero to avoid zero-divide. */
                if (qtable.quantval[0] == 0 || qtable.quantval[Q01_POS] == 0 ||
                    qtable.quantval[Q10_POS] == 0 || qtable.quantval[Q20_POS] == 0 ||
                    qtable.quantval[Q11_POS] == 0 || qtable.quantval[Q02_POS] == 0)
                {
                    return false;
                }

                /* DC values must be at least partly known for all components. */
                if (m_cinfo.m_coef_bits[ci][0] < 0)
                    return false;

                /* Block smoothing is helpful if some AC coefficients remain inaccurate. */
                for (int coefi = 1; coefi <= 5; coefi++)
                {
                    m_coef_bits_latch[m_coef_bits_savedOffset + coefi] = m_cinfo.m_coef_bits[ci][coefi];
                    if (m_cinfo.m_coef_bits[ci][coefi] != 0)
                        smoothing_useful = true;
                }

                m_coef_bits_savedOffset += SAVED_COEFS;
            }

            return smoothing_useful;
        }

        /// <summary>
        /// Reset within-iMCU-row counters for a new row (input side)
        /// </summary>
        private void start_iMCU_row()
        {
            /* In an interleaved scan, an MCU row is the same as an iMCU row.
             * In a noninterleaved scan, an iMCU row has v_samp_factor MCU rows.
             * But at the bottom of the image, process only what's left.
             */
            if (m_cinfo.m_comps_in_scan > 1)
            {
                m_MCU_rows_per_iMCU_row = 1;
            }
            else
            {
                JpegComponent componentInfo = m_cinfo.Comp_info[m_cinfo.m_cur_comp_info[0]];

                if (m_cinfo.m_input_iMCU_row < (m_cinfo.m_total_iMCU_rows - 1))
                    m_MCU_rows_per_iMCU_row = componentInfo.V_samp_factor;
                else
                    m_MCU_rows_per_iMCU_row = componentInfo.last_row_height;
            }

            m_MCU_ctr = 0;
            m_MCU_vert_offset = 0;
        }
    }
    #endregion

    #region JpegDecompressorMainController
    /// <summary>
    /// Main buffer control (downsampled-data buffer)
    /// 
    /// In the current system design, the main buffer need never be a full-image
    /// buffer; any full-height buffers will be found inside the coefficient or
    /// postprocessing controllers.  Nonetheless, the main controller is not
    /// trivial.  Its responsibility is to provide context rows for upsampling/
    /// rescaling, and doing this in an efficient fashion is a bit tricky.
    /// 
    /// Postprocessor input data is counted in "row groups".  A row group
    /// is defined to be (v_samp_factor * DCT_scaled_size / min_DCT_scaled_size)
    /// sample rows of each component.  (We require DCT_scaled_size values to be
    /// chosen such that these numbers are integers.  In practice DCT_scaled_size
    /// values will likely be powers of two, so we actually have the stronger
    /// condition that DCT_scaled_size / min_DCT_scaled_size is an integer.)
    /// Upsampling will typically produce max_v_samp_factor pixel rows from each
    /// row group (times any additional scale factor that the upsampler is
    /// applying).
    /// 
    /// The coefficient controller will deliver data to us one iMCU row at a time;
    /// each iMCU row contains v_samp_factor * DCT_scaled_size sample rows, or
    /// exactly min_DCT_scaled_size row groups.  (This amount of data corresponds
    /// to one row of MCUs when the image is fully interleaved.)  Note that the
    /// number of sample rows varies across components, but the number of row
    /// groups does not.  Some garbage sample rows may be included in the last iMCU
    /// row at the bottom of the image.
    /// 
    /// Depending on the vertical scaling algorithm used, the upsampler may need
    /// access to the sample row(s) above and below its current input row group.
    /// The upsampler is required to set need_context_rows true at global selection
    /// time if so.  When need_context_rows is false, this controller can simply
    /// obtain one iMCU row at a time from the coefficient controller and dole it
    /// out as row groups to the postprocessor.
    /// 
    /// When need_context_rows is true, this controller guarantees that the buffer
    /// passed to postprocessing contains at least one row group's worth of samples
    /// above and below the row group(s) being processed.  Note that the context
    /// rows "above" the first passed row group appear at negative row offsets in
    /// the passed buffer.  At the top and bottom of the image, the required
    /// context rows are manufactured by duplicating the first or last real sample
    /// row; this avoids having special cases in the upsampling inner loops.
    /// 
    /// The amount of context is fixed at one row group just because that's a
    /// convenient number for this controller to work with.  The existing
    /// upsamplers really only need one sample row of context.  An upsampler
    /// supporting arbitrary output rescaling might wish for more than one row
    /// group of context when shrinking the image; tough, we don't handle that.
    /// (This is justified by the assumption that downsizing will be handled mostly
    /// by adjusting the DCT_scaled_size values, so that the actual scale factor at
    /// the upsample step needn't be much less than one.)
    /// 
    /// To provide the desired context, we have to retain the last two row groups
    /// of one iMCU row while reading in the next iMCU row.  (The last row group
    /// can't be processed until we have another row group for its below-context,
    /// and so we have to save the next-to-last group too for its above-context.)
    /// We could do this most simply by copying data around in our buffer, but
    /// that'd be very slow.  We can avoid copying any data by creating a rather
    /// strange pointer structure.  Here's how it works.  We allocate a workspace
    /// consisting of M+2 row groups (where M = min_DCT_scaled_size is the number
    /// of row groups per iMCU row).  We create two sets of redundant pointers to
    /// the workspace.  Labeling the physical row groups 0 to M+1, the synthesized
    /// pointer lists look like this:
    ///                   M+1                          M-1
    ///                   master pointer --> 0         master pointer --> 0
    ///                   1                            1
    ///                   ...                          ...
    ///                   M-3                          M-3
    ///                   M-2                           M
    ///                   M-1                          M+1
    ///                    M                           M-2
    ///                   M+1                          M-1
    ///                    0                            0
    /// We read alternate iMCU rows using each master pointer; thus the last two
    /// row groups of the previous iMCU row remain un-overwritten in the workspace.
    /// The pointer lists are set up so that the required context rows appear to
    /// be adjacent to the proper places when we pass the pointer lists to the
    /// upsampler.
    /// 
    /// The above pictures describe the normal state of the pointer lists.
    /// At top and bottom of the image, we diddle the pointer lists to duplicate
    /// the first or last sample row as necessary (this is cheaper than copying
    /// sample rows around).
    /// 
    /// This scheme breaks down if M less than 2, ie, min_DCT_scaled_size is 1.  In that
    /// situation each iMCU row provides only one row group so the buffering logic
    /// must be different (eg, we must read two iMCU rows before we can emit the
    /// first row group).  For now, we simply do not support providing context
    /// rows when min_DCT_scaled_size is 1.  That combination seems unlikely to
    /// be worth providing --- if someone wants a 1/8th-size preview, they probably
    /// want it quick and dirty, so a context-free upsampler is sufficient.
    /// </summary>
    class JpegDecompressorMainController
    {
        private enum DataProcessor
        {
            context_main,
            simple_main,
            crank_post
        }

        /* context_state values: */
        private const int CTX_PREPARE_FOR_IMCU = 0;   /* need to prepare for MCU row */
        private const int CTX_PROCESS_IMCU = 1;   /* feeding iMCU to postprocessor */
        private const int CTX_POSTPONED_ROW = 2;   /* feeding postponed row group */

        private DataProcessor m_dataProcessor;
        private JpegDecompressor m_cinfo;

        /* Pointer to allocated workspace (M or M+2 row groups). */
        private byte[][][] m_buffer = new byte[JpegConstants.MaxComponents][][];

        private bool m_buffer_full;       /* Have we gotten an iMCU row from decoder? */
        private int m_rowgroup_ctr;    /* counts row groups output to postprocessor */

        /* Remaining fields are only used in the context case. */

        private int[][][] m_funnyIndices = new int[2][][] { new int[JpegConstants.MaxComponents][], new int[JpegConstants.MaxComponents][] };
        private int[] m_funnyOffsets = new int[JpegConstants.MaxComponents];
        private int m_whichFunny;           /* indicates which funny indices set is now in use */

        private int m_context_state;      /* process_data state machine status */
        private int m_rowgroups_avail; /* row groups available to postprocessor */
        private int m_iMCU_row_ctr;    /* counts iMCU rows to detect image top/bot */

        public JpegDecompressorMainController(JpegDecompressor cinfo)
        {
            m_cinfo = cinfo;

            /* Allocate the workspace.
            * ngroups is the number of row groups we need.
            */
            int ngroups = cinfo.m_min_DCT_scaled_size;
            if (cinfo.m_upsample.NeedContextRows())
            {
                if (cinfo.m_min_DCT_scaled_size < 2) /* unsupported, see comments above */
                    throw new Exception("Not implemented yet");

                alloc_funny_pointers(); /* Alloc space for xbuffer[] lists */
                ngroups = cinfo.m_min_DCT_scaled_size + 2;
            }

            for (int ci = 0; ci < cinfo.m_num_components; ci++)
            {
                /* height of a row group of component */
                int rgroup = (cinfo.Comp_info[ci].V_samp_factor * cinfo.Comp_info[ci].DCT_scaled_size) / cinfo.m_min_DCT_scaled_size;

                m_buffer[ci] = JpegCommonBase.AllocJpegSamples(
                    cinfo.Comp_info[ci].Width_in_blocks * cinfo.Comp_info[ci].DCT_scaled_size,
                    rgroup * ngroups);
            }
        }

        /// <summary>
        /// Initialize for a processing pass.
        /// </summary>
        public void start_pass(BufferMode pass_mode)
        {
            switch (pass_mode)
            {
                case BufferMode.PassThru:
                    if (m_cinfo.m_upsample.NeedContextRows())
                    {
                        m_dataProcessor = DataProcessor.context_main;
                        make_funny_pointers(); /* Create the xbuffer[] lists */
                        m_whichFunny = 0; /* Read first iMCU row into xbuffer[0] */
                        m_context_state = CTX_PREPARE_FOR_IMCU;
                        m_iMCU_row_ctr = 0;
                    }
                    else
                    {
                        /* Simple case with no context needed */
                        m_dataProcessor = DataProcessor.simple_main;
                    }
                    m_buffer_full = false;  /* Mark buffer empty */
                    m_rowgroup_ctr = 0;
                    break;
                case BufferMode.CrankDest:
                    /* For last pass of 2-pass quantization, just crank the postprocessor */
                    m_dataProcessor = DataProcessor.crank_post;
                    break;
                default:
                    throw new Exception("Bogus buffer control mode");
            }
        }

        public void process_data(byte[][] output_buf, ref int out_row_ctr, int out_rows_avail)
        {
            switch (m_dataProcessor)
            {
                case DataProcessor.simple_main:
                    process_data_simple_main(output_buf, ref out_row_ctr, out_rows_avail);
                    break;

                case DataProcessor.context_main:
                    process_data_context_main(output_buf, ref out_row_ctr, out_rows_avail);
                    break;

                case DataProcessor.crank_post:
                    process_data_crank_post(output_buf, ref out_row_ctr, out_rows_avail);
                    break;

                default:
                    throw new Exception("Not implemented yet");
            }
        }

        /// <summary>
        /// Process some data.
        /// This handles the simple case where no context is required.
        /// </summary>
        private void process_data_simple_main(byte[][] output_buf, ref int out_row_ctr, int out_rows_avail)
        {
            ComponentBuffer[] cb = new ComponentBuffer[JpegConstants.MaxComponents];
            for (int i = 0; i < JpegConstants.MaxComponents; i++)
            {
                cb[i] = new ComponentBuffer();
                cb[i].SetBuffer(m_buffer[i], null, 0);
            }

            /* Read input data if we haven't filled the main buffer yet */
            if (!m_buffer_full)
            {
                if (m_cinfo.m_coef.decompress_data(cb) == ReadResult.Suspended)
                {
                    /* suspension forced, can do nothing more */
                    return;
                }

                /* OK, we have an iMCU row to work with */
                m_buffer_full = true;
            }

            /* There are always min_DCT_scaled_size row groups in an iMCU row. */
            int rowgroups_avail = m_cinfo.m_min_DCT_scaled_size;

            /* Note: at the bottom of the image, we may pass extra garbage row groups
             * to the postprocessor.  The postprocessor has to check for bottom
             * of image anyway (at row resolution), so no point in us doing it too.
             */

            /* Feed the postprocessor */
            m_cinfo.m_post.post_process_data(cb, ref m_rowgroup_ctr, rowgroups_avail, output_buf, ref out_row_ctr, out_rows_avail);

            /* Has postprocessor consumed all the data yet? If so, mark buffer empty */
            if (m_rowgroup_ctr >= rowgroups_avail)
            {
                m_buffer_full = false;
                m_rowgroup_ctr = 0;
            }
        }

        /// <summary>
        /// Process some data.
        /// This handles the case where context rows must be provided.
        /// </summary>
        private void process_data_context_main(byte[][] output_buf, ref int out_row_ctr, int out_rows_avail)
        {
            ComponentBuffer[] cb = new ComponentBuffer[m_cinfo.m_num_components];
            for (int i = 0; i < m_cinfo.m_num_components; i++)
            {
                cb[i] = new ComponentBuffer();
                cb[i].SetBuffer(m_buffer[i], m_funnyIndices[m_whichFunny][i], m_funnyOffsets[i]);
            }

            /* Read input data if we haven't filled the main buffer yet */
            if (!m_buffer_full)
            {
                if (m_cinfo.m_coef.decompress_data(cb) == ReadResult.Suspended)
                {
                    /* suspension forced, can do nothing more */
                    return;
                }

                /* OK, we have an iMCU row to work with */
                m_buffer_full = true;

                /* count rows received */
                m_iMCU_row_ctr++;
            }

            /* Postprocessor typically will not swallow all the input data it is handed
             * in one call (due to filling the output buffer first).  Must be prepared
             * to exit and restart.
     
     
             This switch lets us keep track of how far we got.
             * Note that each case falls through to the next on successful completion.
             */
            if (m_context_state == CTX_POSTPONED_ROW)
            {
                /* Call postprocessor using previously set pointers for postponed row */
                m_cinfo.m_post.post_process_data(cb, ref m_rowgroup_ctr,
                    m_rowgroups_avail, output_buf, ref out_row_ctr, out_rows_avail);

                if (m_rowgroup_ctr < m_rowgroups_avail)
                {
                    /* Need to suspend */
                    return;
                }

                m_context_state = CTX_PREPARE_FOR_IMCU;

                if (out_row_ctr >= out_rows_avail)
                {
                    /* Postprocessor exactly filled output buf */
                    return;
                }
            }

            if (m_context_state == CTX_PREPARE_FOR_IMCU)
            {
                /* Prepare to process first M-1 row groups of this iMCU row */
                m_rowgroup_ctr = 0;
                m_rowgroups_avail = m_cinfo.m_min_DCT_scaled_size - 1;

                /* Check for bottom of image: if so, tweak pointers to "duplicate"
                 * the last sample row, and adjust rowgroups_avail to ignore padding rows.
                 */
                if (m_iMCU_row_ctr == m_cinfo.m_total_iMCU_rows)
                    set_bottom_pointers();

                m_context_state = CTX_PROCESS_IMCU;
            }

            if (m_context_state == CTX_PROCESS_IMCU)
            {
                /* Call postprocessor using previously set pointers */
                m_cinfo.m_post.post_process_data(cb, ref m_rowgroup_ctr,
                    m_rowgroups_avail, output_buf, ref out_row_ctr, out_rows_avail);

                if (m_rowgroup_ctr < m_rowgroups_avail)
                {
                    /* Need to suspend */
                    return;
                }

                /* After the first iMCU, change wraparound pointers to normal state */
                if (m_iMCU_row_ctr == 1)
                    set_wraparound_pointers();

                /* Prepare to load new iMCU row using other xbuffer list */
                m_whichFunny ^= 1;    /* 0=>1 or 1=>0 */
                m_buffer_full = false;

                /* Still need to process last row group of this iMCU row, */
                /* which is saved at index M+1 of the other xbuffer */
                m_rowgroup_ctr = m_cinfo.m_min_DCT_scaled_size + 1;
                m_rowgroups_avail = m_cinfo.m_min_DCT_scaled_size + 2;
                m_context_state = CTX_POSTPONED_ROW;
            }
        }

        /// <summary>
        /// Process some data.
        /// Final pass of two-pass quantization: just call the postprocessor.
        /// Source data will be the postprocessor controller's internal buffer.
        /// </summary>
        private void process_data_crank_post(byte[][] output_buf, ref int out_row_ctr, int out_rows_avail)
        {
            int dummy = 0;
            m_cinfo.m_post.post_process_data(null, ref dummy, 0, output_buf, ref out_row_ctr, out_rows_avail);
        }

        /// <summary>
        /// Allocate space for the funny pointer lists.
        /// This is done only once, not once per pass.
        /// </summary>
        private void alloc_funny_pointers()
        {
            int M = m_cinfo.m_min_DCT_scaled_size;
            for (int ci = 0; ci < m_cinfo.m_num_components; ci++)
            {
                /* height of a row group of component */
                int rgroup = (m_cinfo.Comp_info[ci].V_samp_factor * m_cinfo.Comp_info[ci].DCT_scaled_size) / m_cinfo.m_min_DCT_scaled_size;

                /* Get space for pointer lists --- M+4 row groups in each list.
                 */
                m_funnyIndices[0][ci] = new int[rgroup * (M + 4)];
                m_funnyIndices[1][ci] = new int[rgroup * (M + 4)];
                m_funnyOffsets[ci] = rgroup;
            }
        }

        /// <summary>
        /// Create the funny pointer lists discussed in the comments above.
        /// The actual workspace is already allocated (in main.buffer),
        /// and the space for the pointer lists is allocated too.
        /// This routine just fills in the curiously ordered lists.
        /// This will be repeated at the beginning of each pass.
        /// </summary>
        private void make_funny_pointers()
        {
            int M = m_cinfo.m_min_DCT_scaled_size;
            for (int ci = 0; ci < m_cinfo.m_num_components; ci++)
            {
                /* height of a row group of component */
                int rgroup = (m_cinfo.Comp_info[ci].V_samp_factor * m_cinfo.Comp_info[ci].DCT_scaled_size) / m_cinfo.m_min_DCT_scaled_size;

                int[] ind0 = m_funnyIndices[0][ci];
                int[] ind1 = m_funnyIndices[1][ci];

                /* First copy the workspace pointers as-is */
                for (int i = 0; i < rgroup * (M + 2); i++)
                {
                    ind0[i + rgroup] = i;
                    ind1[i + rgroup] = i;
                }

                /* In the second list, put the last four row groups in swapped order */
                for (int i = 0; i < rgroup * 2; i++)
                {
                    ind1[rgroup * (M - 1) + i] = rgroup * M + i;
                    ind1[rgroup * (M + 1) + i] = rgroup * (M - 2) + i;
                }

                /* The wraparound pointers at top and bottom will be filled later
                 * (see set_wraparound_pointers, below).  Initially we want the "above"
                 * pointers to duplicate the first actual data line.  This only needs
                 * to happen in xbuffer[0].
                 */
                for (int i = 0; i < rgroup; i++)
                    ind0[i] = ind0[rgroup];
            }
        }

        /// <summary>
        /// Set up the "wraparound" pointers at top and bottom of the pointer lists.
        /// This changes the pointer list state from top-of-image to the normal state.
        /// </summary>
        private void set_wraparound_pointers()
        {
            int M = m_cinfo.m_min_DCT_scaled_size;
            for (int ci = 0; ci < m_cinfo.m_num_components; ci++)
            {
                /* height of a row group of component */
                int rgroup = (m_cinfo.Comp_info[ci].V_samp_factor * m_cinfo.Comp_info[ci].DCT_scaled_size) / m_cinfo.m_min_DCT_scaled_size;

                int[] ind0 = m_funnyIndices[0][ci];
                int[] ind1 = m_funnyIndices[1][ci];

                for (int i = 0; i < rgroup; i++)
                {
                    ind0[i] = ind0[rgroup * (M + 2) + i];
                    ind1[i] = ind1[rgroup * (M + 2) + i];

                    ind0[rgroup * (M + 3) + i] = ind0[i + rgroup];
                    ind1[rgroup * (M + 3) + i] = ind1[i + rgroup];
                }
            }
        }

        /// <summary>
        /// Change the pointer lists to duplicate the last sample row at the bottom
        /// of the image.  m_whichFunny indicates which m_funnyIndices holds the final iMCU row.
        /// Also sets rowgroups_avail to indicate number of nondummy row groups in row.
        /// </summary>
        private void set_bottom_pointers()
        {
            for (int ci = 0; ci < m_cinfo.m_num_components; ci++)
            {
                /* Count sample rows in one iMCU row and in one row group */
                int iMCUheight = m_cinfo.Comp_info[ci].V_samp_factor * m_cinfo.Comp_info[ci].DCT_scaled_size;
                int rgroup = iMCUheight / m_cinfo.m_min_DCT_scaled_size;

                /* Count nondummy sample rows remaining for this component */
                int rows_left = m_cinfo.Comp_info[ci].downsampled_height % iMCUheight;
                if (rows_left == 0)
                    rows_left = iMCUheight;

                /* Count nondummy row groups.  Should get same answer for each component,
                 * so we need only do it once.
                 */
                if (ci == 0)
                    m_rowgroups_avail = (rows_left - 1) / rgroup + 1;

                /* Duplicate the last real sample row rgroup*2 times; this pads out the
                 * last partial rowgroup and ensures at least one full rowgroup of context.
                 */
                for (int i = 0; i < rgroup * 2; i++)
                    m_funnyIndices[m_whichFunny][ci][rows_left + i + rgroup] = m_funnyIndices[m_whichFunny][ci][rows_left - 1 + rgroup];
            }
        }
    }
    #endregion

    #region JpegDecompressorMaster
    /// <summary>
    /// Master control module
    /// </summary>
    class JpegDecompressorMaster
    {
        private JpegDecompressor m_cinfo;

        private int m_pass_number;        /* # of passes completed */
        private bool m_is_dummy_pass; /* True during 1st pass for 2-pass quant */

        private bool m_using_merged_upsample; /* true if using merged upsample/cconvert */

        /* Saved references to initialized quantizer modules,
        * in case we need to switch modes.
        */
        private ColorQuantizer m_quantizer_1pass;
        private ColorQuantizer m_quantizer_2pass;

        public JpegDecompressorMaster(JpegDecompressor cinfo)
        {
            m_cinfo = cinfo;
            master_selection();
        }

        /// <summary>
        /// Per-pass setup.
        /// This is called at the beginning of each output pass.  We determine which
        /// modules will be active during this pass and give them appropriate
        /// start_pass calls.  We also set is_dummy_pass to indicate whether this
        /// is a "real" output pass or a dummy pass for color quantization.
        /// (In the latter case, we will crank the pass to completion.)
        /// </summary>
        public void prepare_for_output_pass()
        {
            if (m_is_dummy_pass)
            {
                /* Final pass of 2-pass quantization */
                m_is_dummy_pass = false;
                m_cinfo.m_cquantize.start_pass(false);
                m_cinfo.m_post.start_pass(BufferMode.CrankDest);
                m_cinfo.m_main.start_pass(BufferMode.CrankDest);
            }
            else
            {
                if (m_cinfo.m_quantize_colors && m_cinfo.m_colormap == null)
                {
                    /* Select new quantization method */
                    if (m_cinfo.m_two_pass_quantize && m_cinfo.m_enable_2pass_quant)
                    {
                        m_cinfo.m_cquantize = m_quantizer_2pass;
                        m_is_dummy_pass = true;
                    }
                    else if (m_cinfo.m_enable_1pass_quant)
                        m_cinfo.m_cquantize = m_quantizer_1pass;
                    else
                        throw new Exception("Invalid color quantization mode change");
                }

                m_cinfo.m_idct.start_pass();
                m_cinfo.m_coef.start_output_pass();

                if (!m_cinfo.m_raw_data_out)
                {
                    m_cinfo.m_upsample.start_pass();

                    if (m_cinfo.m_quantize_colors)
                        m_cinfo.m_cquantize.start_pass(m_is_dummy_pass);

                    m_cinfo.m_post.start_pass((m_is_dummy_pass ? BufferMode.SaveAndPass : BufferMode.PassThru));
                    m_cinfo.m_main.start_pass(BufferMode.PassThru);
                }
            }
        }

        /// <summary>
        /// Finish up at end of an output pass.
        /// </summary>
        public void finish_output_pass()
        {
            if (m_cinfo.m_quantize_colors)
                m_cinfo.m_cquantize.finish_pass();

            m_pass_number++;
        }

        public bool IsDummyPass()
        {
            return m_is_dummy_pass;
        }

        /// <summary>
        /// Master selection of decompression modules.
        /// This is done once at jpeg_start_decompress time.  We determine
        /// which modules will be used and give them appropriate initialization calls.
        /// We also initialize the decompressor input side to begin consuming data.
        /// 
        /// Since jpeg_read_header has finished, we know what is in the SOF
        /// and (first) SOS markers.  We also have all the application parameter
        /// settings.
        /// </summary>
        private void master_selection()
        {
            /* Initialize dimensions and other stuff */
            m_cinfo.jpeg_calc_output_dimensions();
            prepare_range_limit_table();

            /* Width of an output scanline must be representable as int. */
            long samplesperrow = m_cinfo.m_output_width * m_cinfo.m_out_color_components;
            int jd_samplesperrow = (int)samplesperrow;
            if ((long)jd_samplesperrow != samplesperrow)
                throw new Exception("Image too wide for this implementation");

            /* Initialize my private state */
            m_pass_number = 0;
            m_using_merged_upsample = m_cinfo.use_merged_upsample();

            /* Color quantizer selection */
            m_quantizer_1pass = null;
            m_quantizer_2pass = null;

            /* No mode changes if not using buffered-image mode. */
            if (!m_cinfo.m_quantize_colors || !m_cinfo.m_buffered_image)
            {
                m_cinfo.m_enable_1pass_quant = false;
                m_cinfo.m_enable_external_quant = false;
                m_cinfo.m_enable_2pass_quant = false;
            }

            if (m_cinfo.m_quantize_colors)
            {
                if (m_cinfo.m_raw_data_out)
                    throw new Exception("Not implemented yet");

                /* 2-pass quantizer only works in 3-component color space. */
                if (m_cinfo.m_out_color_components != 3)
                {
                    m_cinfo.m_enable_1pass_quant = true;
                    m_cinfo.m_enable_external_quant = false;
                    m_cinfo.m_enable_2pass_quant = false;
                    m_cinfo.m_colormap = null;
                }
                else if (m_cinfo.m_colormap != null)
                    m_cinfo.m_enable_external_quant = true;
                else if (m_cinfo.m_two_pass_quantize)
                    m_cinfo.m_enable_2pass_quant = true;
                else
                    m_cinfo.m_enable_1pass_quant = true;

                if (m_cinfo.m_enable_1pass_quant)
                {
                    m_cinfo.m_cquantize = new Pass1ColorQuantizer(m_cinfo);
                    m_quantizer_1pass = m_cinfo.m_cquantize;
                }

                /* We use the 2-pass code to map to external colormaps. */
                if (m_cinfo.m_enable_2pass_quant || m_cinfo.m_enable_external_quant)
                {
                    m_cinfo.m_cquantize = new Pass2ColorQuantizer(m_cinfo);
                    m_quantizer_2pass = m_cinfo.m_cquantize;
                }
                /* If both quantizers are initialized, the 2-pass one is left active;
                 * this is necessary for starting with quantization to an external map.
                 */
            }

            /* Post-processing: in particular, color conversion first */
            if (!m_cinfo.m_raw_data_out)
            {
                if (m_using_merged_upsample)
                {
                    /* does color conversion too */
                    m_cinfo.m_upsample = new MergedUpsampler(m_cinfo);
                }
                else
                {
                    m_cinfo.m_cconvert = new ColorDeconverter(m_cinfo);
                    m_cinfo.m_upsample = new UpsamplerImpl(m_cinfo);
                }

                m_cinfo.m_post = new JpegDecompressorPostController(m_cinfo, m_cinfo.m_enable_2pass_quant);
            }

            /* Inverse DCT */
            m_cinfo.m_idct = new JpegInverseDCT(m_cinfo);

            if (m_cinfo.m_progressive_mode)
                m_cinfo.m_entropy = new ProgressiveHuffmanDecoder(m_cinfo);
            else
                m_cinfo.m_entropy = new HuffEntropyDecoder(m_cinfo);

            /* Initialize principal buffer controllers. */
            bool use_c_buffer = m_cinfo.m_inputctl.HasMultipleScans() || m_cinfo.m_buffered_image;
            m_cinfo.m_coef = new JpegDecompressorCoefController(m_cinfo, use_c_buffer);

            if (!m_cinfo.m_raw_data_out)
                m_cinfo.m_main = new JpegDecompressorMainController(m_cinfo);

            /* Initialize input side of decompressor to consume first scan. */
            m_cinfo.m_inputctl.start_input_pass();
        }

        /// <summary>
        /// Allocate and fill in the sample_range_limit table.
        /// 
        /// Several decompression processes need to range-limit values to the range
        /// 0..MaxSampleValue; the input value may fall somewhat outside this range
        /// due to noise introduced by quantization, roundoff error, etc. These
        /// processes are inner loops and need to be as fast as possible. On most
        /// machines, particularly CPUs with pipelines or instruction prefetch,
        /// a (subscript-check-less) C table lookup
        ///     x = sample_range_limit[x];
        /// is faster than explicit tests
        /// <c>
        ///     if (x &amp; 0)
        ///        x = 0;
        ///     else if (x > MaxSampleValue)
        ///        x = MaxSampleValue;
        /// </c>
        /// These processes all use a common table prepared by the routine below.
        /// 
        /// For most steps we can mathematically guarantee that the initial value
        /// of x is within MaxSampleValue + 1 of the legal range, so a table running from
        /// -(MaxSampleValue + 1) to 2 * MaxSampleValue + 1 is sufficient.  But for the initial
        /// limiting step (just after the IDCT), a wildly out-of-range value is 
        /// possible if the input data is corrupt.  To avoid any chance of indexing
        /// off the end of memory and getting a bad-pointer trap, we perform the
        /// post-IDCT limiting thus: <c>x = range_limit[x &amp; Mask];</c>
        /// where Mask is 2 bits wider than legal sample data, ie 10 bits for 8-bit
        /// samples.  Under normal circumstances this is more than enough range and
        /// a correct output will be generated; with bogus input data the mask will
        /// cause wraparound, and we will safely generate a bogus-but-in-range output.
        /// For the post-IDCT step, we want to convert the data from signed to unsigned
        /// representation by adding MediumSampleValue at the same time that we limit it.
        /// So the post-IDCT limiting table ends up looking like this:
        /// <pre>
        ///     MediumSampleValue, MediumSampleValue + 1, ..., MaxSampleValue,
        ///     MaxSampleValue (repeat 2 * (MaxSampleValue + 1) - MediumSampleValue times),
        ///     0          (repeat 2 * (MaxSampleValue + 1) - MediumSampleValue times),
        ///     0, 1, ..., MediumSampleValue - 1
        /// </pre>
        /// Negative inputs select values from the upper half of the table after
        /// masking.
        /// 
        /// We can save some space by overlapping the start of the post-IDCT table
        /// with the simpler range limiting table.  The post-IDCT table begins at
        /// sample_range_limit + MediumSampleValue.
        /// 
        /// Note that the table is allocated in near data space on PCs; it's small
        /// enough and used often enough to justify this.
        /// </summary>
        private void prepare_range_limit_table()
        {
            byte[] table = new byte[5 * (JpegConstants.MaxSampleValue + 1) + JpegConstants.MediumSampleValue];

            /* allow negative subscripts of simple table */
            int tableOffset = JpegConstants.MaxSampleValue + 1;
            m_cinfo.m_sample_range_limit = table;
            m_cinfo.m_sampleRangeLimitOffset = tableOffset;

            /* First segment of "simple" table: limit[x] = 0 for x < 0 */
            Array.Clear(table, 0, JpegConstants.MaxSampleValue + 1);

            /* Main part of "simple" table: limit[x] = x */
            for (int i = 0; i <= JpegConstants.MaxSampleValue; i++)
                table[tableOffset + i] = (byte)i;

            tableOffset += JpegConstants.MediumSampleValue; /* Point to where post-IDCT table starts */

            /* End of simple table, rest of first half of post-IDCT table */
            for (int i = JpegConstants.MediumSampleValue; i < 2 * (JpegConstants.MaxSampleValue + 1); i++)
                table[tableOffset + i] = JpegConstants.MaxSampleValue;

            /* Second half of post-IDCT table */
            Array.Clear(table, tableOffset + 2 * (JpegConstants.MaxSampleValue + 1),
                2 * (JpegConstants.MaxSampleValue + 1) - JpegConstants.MediumSampleValue);

            Buffer.BlockCopy(m_cinfo.m_sample_range_limit, 0, table,
                tableOffset + 4 * (JpegConstants.MaxSampleValue + 1) - JpegConstants.MediumSampleValue, JpegConstants.MediumSampleValue);
        }
    }
    #endregion

    #region JpegDecompressorPostController
    /// <summary>
    /// Decompression postprocessing (color quantization buffer control)
    /// </summary>
    class JpegDecompressorPostController
    {
        private enum ProcessorType
        {
            OnePass,
            PrePass,
            Upsample,
            SecondPass
        }

        private ProcessorType m_processor;

        private JpegDecompressor m_cinfo;

        /* Color quantization source buffer: this holds output data from
        * the upsample/color conversion step to be passed to the quantizer.
        * For two-pass color quantization, we need a full-image buffer;
        * for one-pass operation, a strip buffer is sufficient.
        */
        private JpegVirtualArray<byte> m_whole_image;  /* virtual array, or null if one-pass */
        private byte[][] m_buffer;       /* strip buffer, or current strip of virtual */
        private int m_strip_height;    /* buffer size in rows */
        /* for two-pass mode only: */
        private int m_starting_row;    /* row # of first row in current strip */
        private int m_next_row;        /* index of next row to fill/empty in strip */

        /// <summary>
        /// Initialize postprocessing controller.
        /// </summary>
        public JpegDecompressorPostController(JpegDecompressor cinfo, bool need_full_buffer)
        {
            m_cinfo = cinfo;

            /* Create the quantization buffer, if needed */
            if (cinfo.m_quantize_colors)
            {
                /* The buffer strip height is max_v_samp_factor, which is typically
                * an efficient number of rows for upsampling to return.
                * (In the presence of output rescaling, we might want to be smarter?)
                */
                m_strip_height = cinfo.m_max_v_samp_factor;

                if (need_full_buffer)
                {
                    /* Two-pass color quantization: need full-image storage. */
                    /* We round up the number of rows to a multiple of the strip height. */
                    m_whole_image = JpegCommonBase.CreateSamplesArray(
                        cinfo.m_output_width * cinfo.m_out_color_components,
                        JpegUtils.jround_up(cinfo.m_output_height, m_strip_height));
                    m_whole_image.ErrorProcessor = cinfo;
                }
                else
                {
                    /* One-pass color quantization: just make a strip buffer. */
                    m_buffer = JpegCommonBase.AllocJpegSamples(
                        cinfo.m_output_width * cinfo.m_out_color_components, m_strip_height);
                }
            }
        }

        /// <summary>
        /// Initialize for a processing pass.
        /// </summary>
        public void start_pass(BufferMode pass_mode)
        {
            switch (pass_mode)
            {
                case BufferMode.PassThru:
                    if (m_cinfo.m_quantize_colors)
                    {
                        /* Single-pass processing with color quantization. */
                        m_processor = ProcessorType.OnePass;
                        /* We could be doing buffered-image output before starting a 2-pass
                         * color quantization; in that case, jinit_d_post_controller did not
                         * allocate a strip buffer.  Use the virtual-array buffer as workspace.
                         */
                        if (m_buffer == null)
                            m_buffer = m_whole_image.Access(0, m_strip_height);
                    }
                    else
                    {
                        /* For single-pass processing without color quantization,
                         * I have no work to do; just call the upsampler directly.
                         */
                        m_processor = ProcessorType.Upsample;
                    }
                    break;
                case BufferMode.SaveAndPass:
                    /* First pass of 2-pass quantization */
                    if (m_whole_image == null)
                        throw new Exception("Bogus buffer control mode");

                    m_processor = ProcessorType.PrePass;
                    break;
                case BufferMode.CrankDest:
                    /* Second pass of 2-pass quantization */
                    if (m_whole_image == null)
                        throw new Exception("Bogus buffer control mode");

                    m_processor = ProcessorType.SecondPass;
                    break;
                default:
                    throw new Exception("Bogus buffer control mode");
            }
            m_starting_row = m_next_row = 0;
        }

        public void post_process_data(ComponentBuffer[] input_buf, ref int in_row_group_ctr, int in_row_groups_avail, byte[][] output_buf, ref int out_row_ctr, int out_rows_avail)
        {
            switch (m_processor)
            {
                case ProcessorType.OnePass:
                    post_process_1pass(input_buf, ref in_row_group_ctr, in_row_groups_avail, output_buf, ref out_row_ctr, out_rows_avail);
                    break;
                case ProcessorType.PrePass:
                    post_process_prepass(input_buf, ref in_row_group_ctr, in_row_groups_avail, ref out_row_ctr);
                    break;
                case ProcessorType.Upsample:
                    m_cinfo.m_upsample.upsample(input_buf, ref in_row_group_ctr, in_row_groups_avail, output_buf, ref out_row_ctr, out_rows_avail);
                    break;
                case ProcessorType.SecondPass:
                    post_process_2pass(output_buf, ref out_row_ctr, out_rows_avail);
                    break;
                default:
                    throw new Exception("Not implemented yet");
            }
        }

        /// <summary>
        /// Process some data in the one-pass (strip buffer) case.
        /// This is used for color precision reduction as well as one-pass quantization.
        /// </summary>
        private void post_process_1pass(ComponentBuffer[] input_buf, ref int in_row_group_ctr, int in_row_groups_avail, byte[][] output_buf, ref int out_row_ctr, int out_rows_avail)
        {
            /* Fill the buffer, but not more than what we can dump out in one go. */
            /* Note we rely on the upsampler to detect bottom of image. */
            int max_rows = out_rows_avail - out_row_ctr;
            if (max_rows > m_strip_height)
                max_rows = m_strip_height;

            int num_rows = 0;
            m_cinfo.m_upsample.upsample(input_buf, ref in_row_group_ctr, in_row_groups_avail, m_buffer, ref num_rows, max_rows);

            /* Quantize and emit data. */
            m_cinfo.m_cquantize.color_quantize(m_buffer, 0, output_buf, out_row_ctr, num_rows);
            out_row_ctr += num_rows;
        }

        /// <summary>
        /// Process some data in the first pass of 2-pass quantization.
        /// </summary>
        private void post_process_prepass(ComponentBuffer[] input_buf, ref int in_row_group_ctr, int in_row_groups_avail, ref int out_row_ctr)
        {
            int old_next_row, num_rows;

            /* Reposition virtual buffer if at start of strip. */
            if (m_next_row == 0)
                m_buffer = m_whole_image.Access(m_starting_row, m_strip_height);

            /* Upsample some data (up to a strip height's worth). */
            old_next_row = m_next_row;
            m_cinfo.m_upsample.upsample(input_buf, ref in_row_group_ctr, in_row_groups_avail, m_buffer, ref m_next_row, m_strip_height);

            /* Allow quantizer to scan new data.  No data is emitted, */
            /* but we advance out_row_ctr so outer loop can tell when we're done. */
            if (m_next_row > old_next_row)
            {
                num_rows = m_next_row - old_next_row;
                m_cinfo.m_cquantize.color_quantize(m_buffer, old_next_row, null, 0, num_rows);
                out_row_ctr += num_rows;
            }

            /* Advance if we filled the strip. */
            if (m_next_row >= m_strip_height)
            {
                m_starting_row += m_strip_height;
                m_next_row = 0;
            }
        }

        /// <summary>
        /// Process some data in the second pass of 2-pass quantization.
        /// </summary>
        private void post_process_2pass(byte[][] output_buf, ref int out_row_ctr, int out_rows_avail)
        {
            int num_rows, max_rows;

            /* Reposition virtual buffer if at start of strip. */
            if (m_next_row == 0)
                m_buffer = m_whole_image.Access(m_starting_row, m_strip_height);

            /* Determine number of rows to emit. */
            num_rows = m_strip_height - m_next_row; /* available in strip */
            max_rows = out_rows_avail - out_row_ctr; /* available in output area */
            if (num_rows > max_rows)
                num_rows = max_rows;

            /* We have to check bottom of image here, can't depend on upsampler. */
            max_rows = m_cinfo.m_output_height - m_starting_row;
            if (num_rows > max_rows)
                num_rows = max_rows;

            /* Quantize and emit data. */
            m_cinfo.m_cquantize.color_quantize(m_buffer, m_next_row, output_buf, out_row_ctr, num_rows);
            out_row_ctr += num_rows;

            /* Advance if we filled the strip. */
            m_next_row += num_rows;
            if (m_next_row >= m_strip_height)
            {
                m_starting_row += m_strip_height;
                m_next_row = 0;
            }
        }
    }
    #endregion

    #region JpegCommonBase
    /// <summary>Base class for both JPEG compressor and decompresor.</summary>
    /// <remarks>
    /// Routines that are to be used by both halves of the library are declared
    /// to receive an instance of this class. There are no actual instances of 
    /// <see cref="JpegCommonBase"/>, only of <see cref="JpegCompressor"/> 
    /// and <see cref="JpegDecompressor"/>
    /// </remarks>
    public abstract class JpegCommonBase
    {
        internal enum JpegState
        {
            DESTROYED = 0,
            CSTATE_START = 100,     /* after create_compress */
            CSTATE_SCANNING = 101,  /* start_compress done, write_scanlines OK */
            CSTATE_RAW_OK = 102,    /* start_compress done, write_raw_data OK */
            CSTATE_WRCOEFS = 103,   /* jpeg_write_coefficients done */
            DSTATE_START = 200,     /* after create_decompress */
            DSTATE_INHEADER = 201,  /* reading header markers, no SOS yet */
            DSTATE_READY = 202,     /* found SOS, ready for start_decompress */
            DSTATE_PRELOAD = 203,   /* reading multi-scan file in start_decompress*/
            DSTATE_PRESCAN = 204,   /* performing dummy pass for 2-pass quant */
            DSTATE_SCANNING = 205,  /* start_decompress done, read_scanlines OK */
            DSTATE_RAW_OK = 206,    /* start_decompress done, read_raw_data OK */
            DSTATE_BUFIMAGE = 207,  /* expecting jpeg_start_output */
            DSTATE_BUFPOST = 208,   /* looking for SOS/EOI in jpeg_finish_output */
            DSTATE_RDCOEFS = 209,   /* reading file in jpeg_read_coefficients */
            DSTATE_STOPPING = 210   /* looking for EOI in jpeg_finish_decompress */
        }

        internal JpegState m_global_state;     /* For checking call sequence validity */

        /// <summary>
        /// Base constructor.
        /// </summary>
        /// <seealso cref="JpegCompressor"/>
        /// <seealso cref="JpegDecompressor"/>
        public JpegCommonBase()
        {
        }

        /// <summary>
        /// Gets a value indicating whether this instance is Jpeg decompressor.
        /// </summary>
        /// <value>
        /// 	<c>true</c> if this is Jpeg decompressor; otherwise, <c>false</c>.
        /// </value>
        public abstract bool IsDecompressor
        {
            get;
        }

        /// <summary>
        /// Gets the version of LibJpeg.
        /// </summary>
        /// <value>The version of LibJpeg.</value>
        public static string Version
        {
            get
            {
                return "Special Compilation for Cosmos. Based on version 1.2.300.0 of LibJpeg.Net";
            }
        }

        /// <summary>
        /// Gets the LibJpeg's copyright.
        /// </summary>
        /// <value>The copyright.</value>
        public static string Copyright
        {
            get
            {
                return "Copyright (C) 2008-2011, Bit Miracle";
            }
        }

        /// <summary>
        /// Creates the array of samples.
        /// </summary>
        /// <param name="samplesPerRow">The number of samples in row.</param>
        /// <param name="numberOfRows">The number of rows.</param>
        /// <returns>The array of samples.</returns>
        public static JpegVirtualArray<byte> CreateSamplesArray(int samplesPerRow, int numberOfRows)
        {
            return new JpegVirtualArray<byte>(samplesPerRow, numberOfRows, AllocJpegSamples);
        }

        /// <summary>
        /// Creates the array of blocks.
        /// </summary>
        /// <param name="blocksPerRow">The number of blocks in row.</param>
        /// <param name="numberOfRows">The number of rows.</param>
        /// <returns>The array of blocks.</returns>
        /// <seealso cref="JpegBlock"/>
        public static JpegVirtualArray<JpegBlock> CreateBlocksArray(int blocksPerRow, int numberOfRows)
        {
            return new JpegVirtualArray<JpegBlock>(blocksPerRow, numberOfRows, allocJpegBlocks);
        }

        /// <summary>
        /// Creates 2-D sample array.
        /// </summary>
        /// <param name="samplesPerRow">The number of samples per row.</param>
        /// <param name="numberOfRows">The number of rows.</param>
        /// <returns>The array of samples.</returns>
        public static byte[][] AllocJpegSamples(int samplesPerRow, int numberOfRows)
        {
            byte[][] result = new byte[numberOfRows][];
            for (int i = 0; i < numberOfRows; ++i)
                result[i] = new byte[samplesPerRow];

            return result;
        }

        // Creation of 2-D block arrays.
        private static JpegBlock[][] allocJpegBlocks(int blocksPerRow, int numberOfRows)
        {
            JpegBlock[][] result = new JpegBlock[numberOfRows][];
            for (int i = 0; i < numberOfRows; ++i)
            {
                result[i] = new JpegBlock[blocksPerRow];
                for (int j = 0; j < blocksPerRow; ++j)
                    result[i][j] = new JpegBlock();
            }
            return result;
        }

        // Generic versions of jpeg_abort and jpeg_destroy that work on either
        // flavor of JPEG object.  These may be more convenient in some places.

        /// <summary>
        /// Abort processing of a JPEG compression or decompression operation,
        /// but don't destroy the object itself.
        /// 
        /// Closing a data source or destination, if necessary, is the 
        /// application's responsibility.
        /// </summary>
        public void jpeg_abort()
        {
            /* Reset overall state for possible reuse of object */
            if (IsDecompressor)
            {
                m_global_state = JpegState.DSTATE_START;

                /* Try to keep application from accessing now-deleted marker list.
                 * A bit kludgy to do it here, but this is the most central place.
                 */
                JpegDecompressor s = this as JpegDecompressor;
                if (s != null)
                    s.m_marker_list = null;
            }
            else
            {
                m_global_state = JpegState.CSTATE_START;
            }
        }

        /// <summary>
        /// Destruction of a JPEG object. 
        /// 
        /// Closing a data source or destination, if necessary, is the 
        /// application's responsibility.
        /// </summary>
        public void jpeg_destroy()
        {
            // mark it destroyed
            m_global_state = JpegState.DESTROYED;
        }
    }
    #endregion

    #region JpegComponent
    /// <summary>
    /// Basic info about one component (color channel).
    /// </summary>
    public class JpegComponent
    {
        /* These values are fixed over the whole image. */
        /* For compression, they must be supplied by parameter setup; */
        /* for decompression, they are read from the SOF marker. */

        private int component_id;
        private int component_index;
        private int h_samp_factor;
        private int v_samp_factor;
        private int quant_tbl_no;

        /* These values may vary between scans. */
        /* For compression, they must be supplied by parameter setup; */
        /* for decompression, they are read from the SOS marker. */
        /* The decompressor output side may not use these variables. */
        private int dc_tbl_no;
        private int ac_tbl_no;

        /* Remaining fields should be treated as private by applications. */

        /* These values are computed during compression or decompression startup: */
        /* Component's size in DCT blocks.
         * Any dummy blocks added to complete an MCU are not counted; therefore
         * these values do not depend on whether a scan is interleaved or not.
         */
        private int width_in_blocks;
        internal int height_in_blocks;
        /* Size of a DCT block in samples.  Always DCTSize for compression.
         * For decompression this is the size of the output from one DCT block,
         * reflecting any scaling we choose to apply during the IDCT step.
         * Values of 1,2,4,8 are likely to be supported.  Note that different
         * components may receive different IDCT scalings.
         */
        internal int DCT_scaled_size;
        /* The downsampled dimensions are the component's actual, unpadded number
         * of samples at the main buffer (preprocessing/compression interface), thus
         * downsampled_width = ceil(image_width * Hi/Hmax)
         * and similarly for height.  For decompression, IDCT scaling is included, so
         * downsampled_width = ceil(image_width * Hi/Hmax * DCT_scaled_size/DCTSize)
         */
        internal int downsampled_width;    /* actual width in samples */

        internal int downsampled_height; /* actual height in samples */
        /* This flag is used only for decompression.  In cases where some of the
         * components will be ignored (eg grayscale output from YCbCr image),
         * we can skip most computations for the unused components.
         */
        internal bool component_needed;  /* do we need the value of this component? */

        /* These values are computed before starting a scan of the component. */
        /* The decompressor output side may not use these variables. */
        internal int MCU_width;      /* number of blocks per MCU, horizontally */
        internal int MCU_height;     /* number of blocks per MCU, vertically */
        internal int MCU_blocks;     /* MCU_width * MCU_height */
        internal int MCU_sample_width;       /* MCU width in samples, MCU_width*DCT_scaled_size */
        internal int last_col_width;     /* # of non-dummy blocks across in last MCU */
        internal int last_row_height;        /* # of non-dummy blocks down in last MCU */

        /* Saved quantization table for component; null if none yet saved.
         * See JpegInputController comments about the need for this information.
         * This field is currently used only for decompression.
         */
        internal JpegQuantizationTable quant_table;

        internal JpegComponent()
        {
        }

        internal void Assign(JpegComponent ci)
        {
            component_id = ci.component_id;
            component_index = ci.component_index;
            h_samp_factor = ci.h_samp_factor;
            v_samp_factor = ci.v_samp_factor;
            quant_tbl_no = ci.quant_tbl_no;
            dc_tbl_no = ci.dc_tbl_no;
            ac_tbl_no = ci.ac_tbl_no;
            width_in_blocks = ci.width_in_blocks;
            height_in_blocks = ci.height_in_blocks;
            DCT_scaled_size = ci.DCT_scaled_size;
            downsampled_width = ci.downsampled_width;
            downsampled_height = ci.downsampled_height;
            component_needed = ci.component_needed;
            MCU_width = ci.MCU_width;
            MCU_height = ci.MCU_height;
            MCU_blocks = ci.MCU_blocks;
            MCU_sample_width = ci.MCU_sample_width;
            last_col_width = ci.last_col_width;
            last_row_height = ci.last_row_height;
            quant_table = ci.quant_table;
        }

        /// <summary>
        /// Identifier for this component (0..255)
        /// </summary>
        /// <value>The component ID.</value>
        public int Component_id
        {
            get { return component_id; }
            set { component_id = value; }
        }

        /// <summary>
        /// Its index in SOF or <see cref="JpegDecompressor.Comp_info"/>.
        /// </summary>
        /// <value>The component index.</value>
        public int Component_index
        {
            get { return component_index; }
            set { component_index = value; }
        }

        /// <summary>
        /// Horizontal sampling factor (1..4)
        /// </summary>
        /// <value>The horizontal sampling factor.</value>
        public int H_samp_factor
        {
            get { return h_samp_factor; }
            set { h_samp_factor = value; }
        }

        /// <summary>
        /// Vertical sampling factor (1..4)
        /// </summary>
        /// <value>The vertical sampling factor.</value>
        public int V_samp_factor
        {
            get { return v_samp_factor; }
            set { v_samp_factor = value; }
        }

        /// <summary>
        /// Quantization table selector (0..3)
        /// </summary>
        /// <value>The quantization table selector.</value>
        public int Quant_tbl_no
        {
            get { return quant_tbl_no; }
            set { quant_tbl_no = value; }
        }

        /// <summary>
        /// DC entropy table selector (0..3)
        /// </summary>
        /// <value>The DC entropy table selector.</value>
        public int Dc_tbl_no
        {
            get { return dc_tbl_no; }
            set { dc_tbl_no = value; }
        }

        /// <summary>
        /// AC entropy table selector (0..3)
        /// </summary>
        /// <value>The AC entropy table selector.</value>
        public int Ac_tbl_no
        {
            get { return ac_tbl_no; }
            set { ac_tbl_no = value; }
        }

        /// <summary>
        /// Gets or sets the width in blocks.
        /// </summary>
        /// <value>The width in blocks.</value>
        public int Width_in_blocks
        {
            get { return width_in_blocks; }
            set { width_in_blocks = value; }
        }

        /// <summary>
        /// Gets the downsampled width.
        /// </summary>
        /// <value>The downsampled width.</value>
        public int Downsampled_width
        {
            get { return downsampled_width; }
        }

        internal static JpegComponent[] createArrayOfComponents(int length)
        {
            if (length < 0)
                throw new ArgumentOutOfRangeException("length");

            JpegComponent[] result = new JpegComponent[length];
            for (int i = 0; i < result.Length; ++i)
                result[i] = new JpegComponent();

            return result;
        }
    }
    #endregion

    #region JpegConstants
    /// <summary>
    /// Defines some JPEG constants.
    /// </summary>
    public static class JpegConstants
    {
        //////////////////////////////////////////////////////////////////////////
        // All of these are specified by the JPEG standard, so don't change them
        // if you want to be compatible.
        //

        /// <summary>
        /// The basic DCT block is 8x8 samples
        /// </summary>
        public const int DCTSize = 8;

        /// <summary>
        /// DCTSize squared; the number of elements in a block. 
        /// </summary>
        public const int DCTSize2 = DCTSize * DCTSize;

        /// <summary>
        /// Quantization tables are numbered 0..3 
        /// </summary>
        public const int NumberOfQuantTables = 4;

        /// <summary>
        /// Huffman tables are numbered 0..3
        /// </summary>
        public const int NumberOfHuffmanTables = 4;

        /// <summary>
        /// JPEG limit on the number of components in one scan.
        /// </summary>
        public const int MaxComponentsInScan = 4;

        // compressor's limit on blocks per MCU
        //
        // Unfortunately, some bozo at Adobe saw no reason to be bound by the standard;
        // the PostScript DCT filter can emit files with many more than 10 blocks/MCU.
        // If you happen to run across such a file, you can up DecompressorMaxBlocksInMCU
        // to handle it.  We even let you do this from the jconfig.h file. However,
        // we strongly discourage changing CompressorMaxBlocksInMCU; just because Adobe
        // sometimes emits noncompliant files doesn't mean you should too.

        /// <summary>
        /// Compressor's limit on blocks per MCU.
        /// </summary>
        public const int CompressorMaxBlocksInMCU = 10;

        /// <summary>
        /// Decompressor's limit on blocks per MCU.
        /// </summary>
        public const int DecompressorMaxBlocksInMCU = 10;

        /// <summary>
        /// JPEG limit on sampling factors.
        /// </summary>
        public const int MaxSamplingFactor = 4;


        //////////////////////////////////////////////////////////////////////////
        // implementation-specific constants
        //

        // Maximum number of components (color channels) allowed in JPEG image.
        // To meet the letter of the JPEG spec, set this to 255.  However, darn
        // few applications need more than 4 channels (maybe 5 for CMYK + alpha
        // mask).  We recommend 10 as a reasonable compromise; use 4 if you are
        // really short on memory.  (Each allowed component costs a hundred or so
        // bytes of storage, whether actually used in an image or not.)

        /// <summary>
        /// Maximum number of color channels allowed in JPEG image.
        /// </summary>
        public const int MaxComponents = 10;



        /// <summary>
        /// The size of sample.
        /// </summary>
        /// <remarks>Is either:
        /// 8 - for 8-bit sample values (the usual setting)<br/>
        /// 12 - for 12-bit sample values (not supported by this version)<br/>
        /// Only 8 and 12 are legal data precisions for lossy JPEG according to the JPEG standard.
        /// Although original IJG code claims it supports 12 bit images, our code does not support 
        /// anything except 8-bit images.</remarks>
        public const int BitsInSample = 8;

        /// <summary>
        /// DCT method used by default.
        /// </summary>
        public static DCTMethod DefaultDCTMethod = DCTMethod.IntSlow;

        /// <summary>
        /// Fastest DCT method.
        /// </summary>
        public static DCTMethod FastestDCTMethod = DCTMethod.IntFast;

        /// <summary>
        /// A tad under 64K to prevent overflows. 
        /// </summary>
        public const int JpegMaxDimention = 65500;

        /// <summary>
        /// The maximum sample value.
        /// </summary>
        public const int MaxSampleValue = 255;

        /// <summary>
        /// The medium sample value.
        /// </summary>
        public const int MediumSampleValue = 128;

        // Ordering of RGB data in scanlines passed to or from the application.
        // RESTRICTIONS:
        // 1. These macros only affect RGB<=>YCbCr color conversion, so they are not
        // useful if you are using JPEG color spaces other than YCbCr or grayscale.
        // 2. The color quantizer modules will not behave desirably if RGB_PixelLength
        // is not 3 (they don't understand about dummy color components!).  So you
        // can't use color quantization if you change that value.

        /// <summary>
        /// Offset of Red in an RGB scanline element. 
        /// </summary>
        public const int Offset_RGB_Red = 0;

        /// <summary>
        /// Offset of Green in an RGB scanline element. 
        /// </summary>
        public const int Offset_RGB_Green = 1;

        /// <summary>
        /// Offset of Blue in an RGB scanline element. 
        /// </summary>
        public const int Offset_RGB_Blue = 2;

        /// <summary>
        /// Bytes per RGB scanline element.
        /// </summary>
        public const int RGB_PixelLength = 3;

        /// <summary>
        /// The number of bits of lookahead.
        /// </summary>
        public const int HuffmanLookaheadDistance = 8;
    }
    #endregion

    #region JpegDownsampler
    /// <summary>
    /// Downsampling
    /// </summary>
    class JpegDownsampler
    {
        private enum downSampleMethod
        {
            fullsize_smooth_downsampler,
            fullsize_downsampler,
            h2v1_downsampler,
            h2v2_smooth_downsampler,
            h2v2_downsampler,
            int_downsampler
        };

        /* Downsamplers, one per component */
        private downSampleMethod[] m_downSamplers = new downSampleMethod[JpegConstants.MaxComponents];

        private JpegCompressor m_cinfo;
        private bool m_need_context_rows; /* true if need rows above & below */

        public JpegDownsampler(JpegCompressor cinfo)
        {
            m_cinfo = cinfo;
            m_need_context_rows = false;

            if (cinfo.m_CCIR601_sampling)
                throw new Exception("CCIR601 sampling not implemented yet");

            /* Verify we can handle the sampling factors, and set up method pointers */
            for (int ci = 0; ci < cinfo.m_num_components; ci++)
            {
                JpegComponent componentInfo = cinfo.Component_info[ci];

                if (componentInfo.H_samp_factor == cinfo.m_max_h_samp_factor &&
                    componentInfo.V_samp_factor == cinfo.m_max_v_samp_factor)
                {
                    if (cinfo.m_smoothing_factor != 0)
                    {
                        m_downSamplers[ci] = downSampleMethod.fullsize_smooth_downsampler;
                        m_need_context_rows = true;
                    }
                    else
                    {
                        m_downSamplers[ci] = downSampleMethod.fullsize_downsampler;
                    }
                }
                else if (componentInfo.H_samp_factor * 2 == cinfo.m_max_h_samp_factor &&
                         componentInfo.V_samp_factor == cinfo.m_max_v_samp_factor)
                {
                    m_downSamplers[ci] = downSampleMethod.h2v1_downsampler;
                }
                else if (componentInfo.H_samp_factor * 2 == cinfo.m_max_h_samp_factor &&
                         componentInfo.V_samp_factor * 2 == cinfo.m_max_v_samp_factor)
                {
                    if (cinfo.m_smoothing_factor != 0)
                    {
                        m_downSamplers[ci] = downSampleMethod.h2v2_smooth_downsampler;
                        m_need_context_rows = true;
                    }
                    else
                    {
                        m_downSamplers[ci] = downSampleMethod.h2v2_downsampler;
                    }
                }
                else if ((cinfo.m_max_h_samp_factor % componentInfo.H_samp_factor) == 0 &&
                         (cinfo.m_max_v_samp_factor % componentInfo.V_samp_factor) == 0)
                {
                    m_downSamplers[ci] = downSampleMethod.int_downsampler;
                }
                else
                    throw new Exception("Fractional sampling not implemented yet");
            }
        }

        /// <summary>
        /// Do downsampling for a whole row group (all components).
        /// 
        /// In this version we simply downsample each component independently.
        /// </summary>
        public void downsample(byte[][][] input_buf, int in_row_index, byte[][][] output_buf, int out_row_group_index)
        {
            for (int ci = 0; ci < m_cinfo.m_num_components; ci++)
            {
                int outIndex = out_row_group_index * m_cinfo.Component_info[ci].V_samp_factor;
                switch (m_downSamplers[ci])
                {
                    case downSampleMethod.fullsize_smooth_downsampler:
                        fullsize_smooth_downsample(ci, input_buf[ci], in_row_index, output_buf[ci], outIndex);
                        break;

                    case downSampleMethod.fullsize_downsampler:
                        fullsize_downsample(ci, input_buf[ci], in_row_index, output_buf[ci], outIndex);
                        break;

                    case downSampleMethod.h2v1_downsampler:
                        h2v1_downsample(ci, input_buf[ci], in_row_index, output_buf[ci], outIndex);
                        break;

                    case downSampleMethod.h2v2_smooth_downsampler:
                        h2v2_smooth_downsample(ci, input_buf[ci], in_row_index, output_buf[ci], outIndex);
                        break;

                    case downSampleMethod.h2v2_downsampler:
                        h2v2_downsample(ci, input_buf[ci], in_row_index, output_buf[ci], outIndex);
                        break;

                    case downSampleMethod.int_downsampler:
                        int_downsample(ci, input_buf[ci], in_row_index, output_buf[ci], outIndex);
                        break;
                };
            }
        }

        public bool NeedContextRows()
        {
            return m_need_context_rows;
        }

        /// <summary>
        /// Downsample pixel values of a single component.
        /// One row group is processed per call.
        /// This version handles arbitrary integral sampling ratios, without smoothing.
        /// Note that this version is not actually used for customary sampling ratios.
        /// </summary>
        private void int_downsample(int componentIndex, byte[][] input_data, int startInputRow, byte[][] output_data, int startOutRow)
        {
            /* Expand input data enough to let all the output samples be generated
             * by the standard loop.  Special-casing padded output would be more
             * efficient.
             */
            int output_cols = m_cinfo.Component_info[componentIndex].Width_in_blocks * JpegConstants.DCTSize;
            int h_expand = m_cinfo.m_max_h_samp_factor / m_cinfo.Component_info[componentIndex].H_samp_factor;
            expand_right_edge(input_data, startInputRow, m_cinfo.m_max_v_samp_factor, m_cinfo.m_image_width, output_cols * h_expand);

            int v_expand = m_cinfo.m_max_v_samp_factor / m_cinfo.Component_info[componentIndex].V_samp_factor;
            int numpix = h_expand * v_expand;
            int numpix2 = numpix / 2;
            int inrow = 0;
            for (int outrow = 0; outrow < m_cinfo.Component_info[componentIndex].V_samp_factor; outrow++)
            {
                for (int outcol = 0, outcol_h = 0; outcol < output_cols; outcol++, outcol_h += h_expand)
                {
                    int outvalue = 0;
                    for (int v = 0; v < v_expand; v++)
                    {
                        for (int h = 0; h < h_expand; h++)
                            outvalue += input_data[startInputRow + inrow + v][outcol_h + h];
                    }

                    output_data[startOutRow + outrow][outcol] = (byte)((outvalue + numpix2) / numpix);
                }

                inrow += v_expand;
            }
        }

        /// <summary>
        /// Downsample pixel values of a single component.
        /// This version handles the special case of a full-size component,
        /// without smoothing.
        /// </summary>
        private void fullsize_downsample(int componentIndex, byte[][] input_data, int startInputRow, byte[][] output_data, int startOutRow)
        {
            /* Copy the data */
            JpegUtils.jcopy_sample_rows(input_data, startInputRow, output_data, startOutRow, m_cinfo.m_max_v_samp_factor, m_cinfo.m_image_width);

            /* Edge-expand */
            expand_right_edge(output_data, startOutRow, m_cinfo.m_max_v_samp_factor, m_cinfo.m_image_width, m_cinfo.Component_info[componentIndex].Width_in_blocks * JpegConstants.DCTSize);
        }

        /// <summary>
        /// Downsample pixel values of a single component.
        /// This version handles the common case of 2:1 horizontal and 1:1 vertical,
        /// without smoothing.
        /// 
        /// A note about the "bias" calculations: when rounding fractional values to
        /// integer, we do not want to always round 0.5 up to the next integer.
        /// If we did that, we'd introduce a noticeable bias towards larger values.
        /// Instead, this code is arranged so that 0.5 will be rounded up or down at
        /// alternate pixel locations (a simple ordered dither pattern).
        /// </summary>
        private void h2v1_downsample(int componentIndex, byte[][] input_data, int startInputRow, byte[][] output_data, int startOutRow)
        {
            /* Expand input data enough to let all the output samples be generated
             * by the standard loop.  Special-casing padded output would be more
             * efficient.
             */
            int output_cols = m_cinfo.Component_info[componentIndex].Width_in_blocks * JpegConstants.DCTSize;
            expand_right_edge(input_data, startInputRow, m_cinfo.m_max_v_samp_factor, m_cinfo.m_image_width, output_cols * 2);

            for (int outrow = 0; outrow < m_cinfo.Component_info[componentIndex].V_samp_factor; outrow++)
            {
                /* bias = 0,1,0,1,... for successive samples */
                int bias = 0;
                int inputColumn = 0;
                for (int outcol = 0; outcol < output_cols; outcol++)
                {
                    output_data[startOutRow + outrow][outcol] = (byte)(
                        ((int)input_data[startInputRow + outrow][inputColumn] +
                        (int)input_data[startInputRow + outrow][inputColumn + 1] + bias) >> 1);

                    bias ^= 1;      /* 0=>1, 1=>0 */
                    inputColumn += 2;
                }
            }
        }

        /// <summary>
        /// Downsample pixel values of a single component.
        /// This version handles the standard case of 2:1 horizontal and 2:1 vertical,
        /// without smoothing.
        /// </summary>
        private void h2v2_downsample(int componentIndex, byte[][] input_data, int startInputRow, byte[][] output_data, int startOutRow)
        {
            /* Expand input data enough to let all the output samples be generated
             * by the standard loop.  Special-casing padded output would be more
             * efficient.
             */
            int output_cols = m_cinfo.Component_info[componentIndex].Width_in_blocks * JpegConstants.DCTSize;
            expand_right_edge(input_data, startInputRow, m_cinfo.m_max_v_samp_factor, m_cinfo.m_image_width, output_cols * 2);

            int inrow = 0;
            for (int outrow = 0; outrow < m_cinfo.Component_info[componentIndex].V_samp_factor; outrow++)
            {
                /* bias = 1,2,1,2,... for successive samples */
                int bias = 1;
                int inputColumn = 0;
                for (int outcol = 0; outcol < output_cols; outcol++)
                {
                    output_data[startOutRow + outrow][outcol] = (byte)((
                        (int)input_data[startInputRow + inrow][inputColumn] +
                        (int)input_data[startInputRow + inrow][inputColumn + 1] +
                        (int)input_data[startInputRow + inrow + 1][inputColumn] +
                        (int)input_data[startInputRow + inrow + 1][inputColumn + 1] + bias) >> 2);

                    bias ^= 3;      /* 1=>2, 2=>1 */
                    inputColumn += 2;
                }

                inrow += 2;
            }
        }

        /// <summary>
        /// Downsample pixel values of a single component.
        /// This version handles the standard case of 2:1 horizontal and 2:1 vertical,
        /// with smoothing.  One row of context is required.
        /// </summary>
        private void h2v2_smooth_downsample(int componentIndex, byte[][] input_data, int startInputRow, byte[][] output_data, int startOutRow)
        {
            /* Expand input data enough to let all the output samples be generated
             * by the standard loop.  Special-casing padded output would be more
             * efficient.
             */
            int output_cols = m_cinfo.Component_info[componentIndex].Width_in_blocks * JpegConstants.DCTSize;
            expand_right_edge(input_data, startInputRow - 1, m_cinfo.m_max_v_samp_factor + 2, m_cinfo.m_image_width, output_cols * 2);

            /* We don't bother to form the individual "smoothed" input pixel values;
             * we can directly compute the output which is the average of the four
             * smoothed values.  Each of the four member pixels contributes a fraction
             * (1-8*SF) to its own smoothed image and a fraction SF to each of the three
             * other smoothed pixels, therefore a total fraction (1-5*SF)/4 to the final
             * output.  The four corner-adjacent neighbor pixels contribute a fraction
             * SF to just one smoothed pixel, or SF/4 to the final output; while the
             * eight edge-adjacent neighbors contribute SF to each of two smoothed
             * pixels, or SF/2 overall.  In order to use integer arithmetic, these
             * factors are scaled by 2^16 = 65536.
             * Also recall that SF = smoothing_factor / 1024.
             */

            int memberscale = 16384 - m_cinfo.m_smoothing_factor * 80; /* scaled (1-5*SF)/4 */
            int neighscale = m_cinfo.m_smoothing_factor * 16; /* scaled SF/4 */

            for (int inrow = 0, outrow = 0; outrow < m_cinfo.Component_info[componentIndex].V_samp_factor; outrow++)
            {
                int outIndex = 0;
                int inIndex0 = 0;
                int inIndex1 = 0;
                int aboveIndex = 0;
                int belowIndex = 0;

                /* Special case for first column: pretend column -1 is same as column 0 */
                int membersum = input_data[startInputRow + inrow][inIndex0] +
                    input_data[startInputRow + inrow][inIndex0 + 1] +
                    input_data[startInputRow + inrow + 1][inIndex1] +
                    input_data[startInputRow + inrow + 1][inIndex1 + 1];

                int neighsum = input_data[startInputRow + inrow - 1][aboveIndex] +
                    input_data[startInputRow + inrow - 1][aboveIndex + 1] +
                    input_data[startInputRow + inrow + 2][belowIndex] +
                    input_data[startInputRow + inrow + 2][belowIndex + 1] +
                    input_data[startInputRow + inrow][inIndex0] +
                    input_data[startInputRow + inrow][inIndex0 + 2] +
                    input_data[startInputRow + inrow + 1][inIndex1] +
                    input_data[startInputRow + inrow + 1][inIndex1 + 2];

                neighsum += neighsum;
                neighsum += input_data[startInputRow + inrow - 1][aboveIndex] +
                    input_data[startInputRow + inrow - 1][aboveIndex + 2] +
                    input_data[startInputRow + inrow + 2][belowIndex] +
                    input_data[startInputRow + inrow + 2][belowIndex + 2];

                membersum = membersum * memberscale + neighsum * neighscale;
                output_data[startOutRow + outrow][outIndex] = (byte)((membersum + 32768) >> 16);
                outIndex++;

                inIndex0 += 2;
                inIndex1 += 2;
                aboveIndex += 2;
                belowIndex += 2;

                for (int colctr = output_cols - 2; colctr > 0; colctr--)
                {
                    /* sum of pixels directly mapped to this output element */
                    membersum = input_data[startInputRow + inrow][inIndex0] +
                        input_data[startInputRow + inrow][inIndex0 + 1] +
                        input_data[startInputRow + inrow + 1][inIndex1] +
                        input_data[startInputRow + inrow + 1][inIndex1 + 1];

                    /* sum of edge-neighbor pixels */
                    neighsum = input_data[startInputRow + inrow - 1][aboveIndex] +
                        input_data[startInputRow + inrow - 1][aboveIndex + 1] +
                        input_data[startInputRow + inrow + 2][belowIndex] +
                        input_data[startInputRow + inrow + 2][belowIndex + 1] +
                        input_data[startInputRow + inrow][inIndex0 - 1] +
                        input_data[startInputRow + inrow][inIndex0 + 2] +
                        input_data[startInputRow + inrow + 1][inIndex1 - 1] +
                        input_data[startInputRow + inrow + 1][inIndex1 + 2];

                    /* The edge-neighbors count twice as much as corner-neighbors */
                    neighsum += neighsum;

                    /* Add in the corner-neighbors */
                    neighsum += input_data[startInputRow + inrow - 1][aboveIndex - 1] +
                        input_data[startInputRow + inrow - 1][aboveIndex + 2] +
                        input_data[startInputRow + inrow + 2][belowIndex - 1] +
                        input_data[startInputRow + inrow + 2][belowIndex + 2];

                    /* form final output scaled up by 2^16 */
                    membersum = membersum * memberscale + neighsum * neighscale;

                    /* round, descale and output it */
                    output_data[startOutRow + outrow][outIndex] = (byte)((membersum + 32768) >> 16);
                    outIndex++;

                    inIndex0 += 2;
                    inIndex1 += 2;
                    aboveIndex += 2;
                    belowIndex += 2;
                }

                /* Special case for last column */
                membersum = input_data[startInputRow + inrow][inIndex0] +
                    input_data[startInputRow + inrow][inIndex0 + 1] +
                    input_data[startInputRow + inrow + 1][inIndex1] +
                    input_data[startInputRow + inrow + 1][inIndex1 + 1];

                neighsum = input_data[startInputRow + inrow - 1][aboveIndex] +
                    input_data[startInputRow + inrow - 1][aboveIndex + 1] +
                    input_data[startInputRow + inrow + 2][belowIndex] +
                    input_data[startInputRow + inrow + 2][belowIndex + 1] +
                    input_data[startInputRow + inrow][inIndex0 - 1] +
                    input_data[startInputRow + inrow][inIndex0 + 1] +
                    input_data[startInputRow + inrow + 1][inIndex1 - 1] +
                    input_data[startInputRow + inrow + 1][inIndex1 + 1];

                neighsum += neighsum;
                neighsum += input_data[startInputRow + inrow - 1][aboveIndex - 1] +
                    input_data[startInputRow + inrow - 1][aboveIndex + 1] +
                    input_data[startInputRow + inrow + 2][belowIndex - 1] +
                    input_data[startInputRow + inrow + 2][belowIndex + 1];

                membersum = membersum * memberscale + neighsum * neighscale;
                output_data[startOutRow + outrow][outIndex] = (byte)((membersum + 32768) >> 16);

                inrow += 2;
            }
        }

        /// <summary>
        /// Downsample pixel values of a single component.
        /// This version handles the special case of a full-size component,
        /// with smoothing.  One row of context is required.
        /// </summary>
        private void fullsize_smooth_downsample(int componentIndex, byte[][] input_data, int startInputRow, byte[][] output_data, int startOutRow)
        {
            /* Expand input data enough to let all the output samples be generated
             * by the standard loop.  Special-casing padded output would be more
             * efficient.
             */
            int output_cols = m_cinfo.Component_info[componentIndex].Width_in_blocks * JpegConstants.DCTSize;
            expand_right_edge(input_data, startInputRow - 1, m_cinfo.m_max_v_samp_factor + 2, m_cinfo.m_image_width, output_cols);

            /* Each of the eight neighbor pixels contributes a fraction SF to the
             * smoothed pixel, while the main pixel contributes (1-8*SF).  In order
             * to use integer arithmetic, these factors are multiplied by 2^16 = 65536.
             * Also recall that SF = smoothing_factor / 1024.
             */

            int memberscale = 65536 - m_cinfo.m_smoothing_factor * 512; /* scaled 1-8*SF */
            int neighscale = m_cinfo.m_smoothing_factor * 64; /* scaled SF */

            for (int outrow = 0; outrow < m_cinfo.Component_info[componentIndex].V_samp_factor; outrow++)
            {
                int outIndex = 0;
                int inIndex = 0;
                int aboveIndex = 0;
                int belowIndex = 0;

                /* Special case for first column */
                int colsum = input_data[startInputRow + outrow - 1][aboveIndex] +
                    input_data[startInputRow + outrow + 1][belowIndex] +
                    input_data[startInputRow + outrow][inIndex];

                aboveIndex++;
                belowIndex++;

                int membersum = input_data[startInputRow + outrow][inIndex];
                inIndex++;

                int nextcolsum = input_data[startInputRow + outrow - 1][aboveIndex] +
                    input_data[startInputRow + outrow + 1][belowIndex] +
                    input_data[startInputRow + outrow][inIndex];

                int neighsum = colsum + (colsum - membersum) + nextcolsum;

                membersum = membersum * memberscale + neighsum * neighscale;
                output_data[startOutRow + outrow][outIndex] = (byte)((membersum + 32768) >> 16);
                outIndex++;

                int lastcolsum = colsum;
                colsum = nextcolsum;

                for (int colctr = output_cols - 2; colctr > 0; colctr--)
                {
                    membersum = input_data[startInputRow + outrow][inIndex];

                    inIndex++;
                    aboveIndex++;
                    belowIndex++;

                    nextcolsum = input_data[startInputRow + outrow - 1][aboveIndex] +
                        input_data[startInputRow + outrow + 1][belowIndex] +
                        input_data[startInputRow + outrow][inIndex];

                    neighsum = lastcolsum + (colsum - membersum) + nextcolsum;
                    membersum = membersum * memberscale + neighsum * neighscale;

                    output_data[startOutRow + outrow][outIndex] = (byte)((membersum + 32768) >> 16);
                    outIndex++;

                    lastcolsum = colsum;
                    colsum = nextcolsum;
                }

                /* Special case for last column */
                membersum = input_data[startInputRow + outrow][inIndex];
                neighsum = lastcolsum + (colsum - membersum) + colsum;
                membersum = membersum * memberscale + neighsum * neighscale;
                output_data[startOutRow + outrow][outIndex] = (byte)((membersum + 32768) >> 16);
            }
        }

        /// <summary>
        /// Expand a component horizontally from width input_cols to width output_cols,
        /// by duplicating the rightmost samples.
        /// </summary>
        private static void expand_right_edge(byte[][] image_data, int startInputRow, int num_rows, int input_cols, int output_cols)
        {
            int numcols = output_cols - input_cols;
            if (numcols > 0)
            {
                for (int row = startInputRow; row < (startInputRow + num_rows); row++)
                {
                    /* don't need GETJSAMPLE() here */
                    byte pixval = image_data[row][input_cols - 1];
                    for (int count = 0; count < numcols; count++)
                        image_data[row][input_cols + count] = pixval;
                }
            }
        }
    }
    #endregion

    #region JpegEntropyDecoder
    /// <summary>
    /// Entropy decoding
    /// </summary>
    abstract class JpegEntropyDecoder
    {
        // Figure F.12: extend sign bit.
        // entry n is 2**(n-1)
        private static int[] extend_test = 
        { 
            0, 0x0001, 0x0002, 0x0004, 0x0008, 0x0010, 0x0020, 
            0x0040, 0x0080, 0x0100, 0x0200, 0x0400, 0x0800, 
            0x1000, 0x2000, 0x4000 
        };

        // entry n is (-1 << n) + 1
        private static int[] extend_offset = 
        { 
            0, (-1 << 1) + 1, (-1 << 2) + 1, 
            (-1 << 3) + 1, (-1 << 4) + 1, (-1 << 5) + 1,
            (-1 << 6) + 1, (-1 << 7) + 1, (-1 << 8) + 1,
            (-1 << 9) + 1, (-1 << 10) + 1,
            (-1 << 11) + 1, (-1 << 12) + 1,
            (-1 << 13) + 1, (-1 << 14) + 1,
            (-1 << 15) + 1 
        };

        /* Fetching the next N bits from the input stream is a time-critical operation
        * for the Huffman decoders.  We implement it with a combination of inline
        * macros and out-of-line subroutines.  Note that N (the number of bits
        * demanded at one time) never exceeds 15 for JPEG use.
        *
        * We read source bytes into get_buffer and dole out bits as needed.
        * If get_buffer already contains enough bits, they are fetched in-line
        * by the macros CHECK_BIT_BUFFER and GET_BITS.  When there aren't enough
        * bits, jpeg_fill_bit_buffer is called; it will attempt to fill get_buffer
        * as full as possible (not just to the number of bits needed; this
        * prefetching reduces the overhead cost of calling jpeg_fill_bit_buffer).
        * Note that jpeg_fill_bit_buffer may return false to indicate suspension.
        * On true return, jpeg_fill_bit_buffer guarantees that get_buffer contains
        * at least the requested number of bits --- dummy zeroes are inserted if
        * necessary.
        */
        protected const int BIT_BUF_SIZE = 32;    /* size of buffer in bits */

        /*
        * Out-of-line code for bit fetching (shared with jdphuff.c).
        * See jdhuff.h for info about usage.
        * Note: current values of get_buffer and bits_left are passed as parameters,
        * but are returned in the corresponding fields of the state struct.
        *
        * On most machines MIN_GET_BITS should be 25 to allow the full 32-bit width
        * of get_buffer to be used.  (On machines with wider words, an even larger
        * buffer could be used.)  However, on some machines 32-bit shifts are
        * quite slow and take time proportional to the number of places shifted.
        * (This is true with most PC compilers, for instance.)  In this case it may
        * be a win to set MIN_GET_BITS to the minimum value of 15.  This reduces the
        * average shift distance at the cost of more calls to jpeg_fill_bit_buffer.
        */

        protected const int MIN_GET_BITS = BIT_BUF_SIZE - 7;

        protected JpegDecompressor m_cinfo;

        /* This is here to share code between baseline and progressive decoders; */
        /* other modules probably should not use it */
        protected bool m_insufficient_data; /* set true after emitting warning */

        public abstract void start_pass();
        public abstract bool decode_mcu(JpegBlock[] MCU_data);

        protected static int HUFF_EXTEND(int x, int s)
        {
            return ((x) < extend_test[s] ? (x) + extend_offset[s] : (x));
        }

        protected void BITREAD_LOAD_STATE(SavedBitreadState bitstate, out int get_buffer, out int bits_left, ref WorkingBitreadState br_state)
        {
            br_state.cinfo = m_cinfo;
            get_buffer = bitstate.get_buffer;
            bits_left = bitstate.bits_left;
        }

        protected static void BITREAD_SAVE_STATE(ref SavedBitreadState bitstate, int get_buffer, int bits_left)
        {
            bitstate.get_buffer = get_buffer;
            bitstate.bits_left = bits_left;
        }

        /// <summary>
        /// Expand a Huffman table definition into the derived format
        /// This routine also performs some validation checks on the table.
        /// </summary>
        protected void jpeg_make_d_derived_tbl(bool isDC, int tblno, ref DerivedTable dtbl)
        {
            /* Note that huffsize[] and huffcode[] are filled in code-length order,
            * paralleling the order of the symbols themselves in htbl.huffval[].
            */

            /* Find the input Huffman table */
            if (tblno < 0 || tblno >= JpegConstants.NumberOfHuffmanTables)
                throw new Exception(String.Format("Huffman table 0x{0:X2} was not defined", tblno));

            JpegHuffmanTable htbl = isDC ? m_cinfo.m_dc_huff_tbl_ptrs[tblno] : m_cinfo.m_ac_huff_tbl_ptrs[tblno];
            if (htbl == null)
                throw new Exception(String.Format("Huffman table 0x{0:X2} was not defined", tblno));

            /* Allocate a workspace if we haven't already done so. */
            if (dtbl == null)
                dtbl = new DerivedTable();

            dtbl.pub = htbl;       /* fill in back link */

            /* Figure C.1: make table of Huffman code length for each symbol */

            int p = 0;
            char[] huffsize = new char[257];
            for (int l = 1; l <= 16; l++)
            {
                int i = htbl.Bits[l];
                if (i < 0 || p + i > 256)    /* protect against table overrun */
                    throw new Exception("Bogus Huffman table definition");

                while ((i--) != 0)
                    huffsize[p++] = (char)l;
            }
            huffsize[p] = (char)0;
            int numsymbols = p;

            /* Figure C.2: generate the codes themselves */
            /* We also validate that the counts represent a legal Huffman code tree. */

            int code = 0;
            int si = huffsize[0];
            int[] huffcode = new int[257];
            p = 0;
            while (huffsize[p] != 0)
            {
                while (((int)huffsize[p]) == si)
                {
                    huffcode[p++] = code;
                    code++;
                }

                /* code is now 1 more than the last code used for code-length si; but
                * it must still fit in si bits, since no code is allowed to be all ones.
                */
                if (code >= (1 << si))
                    throw new Exception("Bogus Huffman table definition");
                code <<= 1;
                si++;
            }

            /* Figure F.15: generate decoding tables for bit-sequential decoding */

            p = 0;
            for (int l = 1; l <= 16; l++)
            {
                if (htbl.Bits[l] != 0)
                {
                    /* valoffset[l] = huffval[] index of 1st symbol of code length l,
                    * minus the minimum code of length l
                    */
                    dtbl.valoffset[l] = p - huffcode[p];
                    p += htbl.Bits[l];
                    dtbl.maxcode[l] = huffcode[p - 1]; /* maximum code of length l */
                }
                else
                {
                    /* -1 if no codes of this length */
                    dtbl.maxcode[l] = -1;
                }
            }
            dtbl.maxcode[17] = 0xFFFFF; /* ensures jpeg_huff_decode terminates */

            /* Compute lookahead tables to speed up decoding.
            * First we set all the table entries to 0, indicating "too long";
            * then we iterate through the Huffman codes that are short enough and
            * fill in all the entries that correspond to bit sequences starting
            * with that code.
            */

            Array.Clear(dtbl.look_nbits, 0, dtbl.look_nbits.Length);
            p = 0;
            for (int l = 1; l <= JpegConstants.HuffmanLookaheadDistance; l++)
            {
                for (int i = 1; i <= htbl.Bits[l]; i++, p++)
                {
                    /* l = current code's length, p = its index in huffcode[] & huffval[]. */
                    /* Generate left-justified code followed by all possible bit sequences */
                    int lookbits = huffcode[p] << (JpegConstants.HuffmanLookaheadDistance - l);
                    for (int ctr = 1 << (JpegConstants.HuffmanLookaheadDistance - l); ctr > 0; ctr--)
                    {
                        dtbl.look_nbits[lookbits] = l;
                        dtbl.look_sym[lookbits] = htbl.Huffval[p];
                        lookbits++;
                    }
                }
            }

            /* Validate symbols as being reasonable.
            * For AC tables, we make no check, but accept all byte values 0..255.
            * For DC tables, we require the symbols to be in range 0..15.
            * (Tighter bounds could be applied depending on the data depth and mode,
            * but this is sufficient to ensure safe decoding.)
            */
            if (isDC)
            {
                for (int i = 0; i < numsymbols; i++)
                {
                    int sym = htbl.Huffval[i];
                    if (sym < 0 || sym > 15)
                        throw new Exception("Bogus Huffman table definition");
                }
            }
        }

        /*
        * These methods provide the in-line portion of bit fetching.
        * Use CHECK_BIT_BUFFER to ensure there are N bits in get_buffer
        * before using GET_BITS, PEEK_BITS, or DROP_BITS.
        * The variables get_buffer and bits_left are assumed to be locals,
        * but the state struct might not be (jpeg_huff_decode needs this).
        *  CHECK_BIT_BUFFER(state,n,action);
        *      Ensure there are N bits in get_buffer; if suspend, take action.
        *      val = GET_BITS(n);
        *      Fetch next N bits.
        *      val = PEEK_BITS(n);
        *      Fetch next N bits without removing them from the buffer.
        *  DROP_BITS(n);
        *      Discard next N bits.
        * The value N should be a simple variable, not an expression, because it
        * is evaluated multiple times.
        */

        protected static bool CHECK_BIT_BUFFER(ref WorkingBitreadState state, int nbits, ref int get_buffer, ref int bits_left)
        {
            if (bits_left < nbits)
            {
                if (!jpeg_fill_bit_buffer(ref state, get_buffer, bits_left, nbits))
                    return false;

                get_buffer = state.get_buffer;
                bits_left = state.bits_left;
            }

            return true;
        }

        protected static int GET_BITS(int nbits, int get_buffer, ref int bits_left)
        {
            return (((int)(get_buffer >> (bits_left -= nbits))) & ((1 << nbits) - 1));
        }

        protected static int PEEK_BITS(int nbits, int get_buffer, int bits_left)
        {
            return (((int)(get_buffer >> (bits_left - nbits))) & ((1 << nbits) - 1));
        }

        protected static void DROP_BITS(int nbits, ref int bits_left)
        {
            bits_left -= nbits;
        }

        /* Load up the bit buffer to a depth of at least nbits */
        protected static bool jpeg_fill_bit_buffer(ref WorkingBitreadState state, int get_buffer, int bits_left, int nbits)
        {
            /* Attempt to load at least MIN_GET_BITS bits into get_buffer. */
            /* (It is assumed that no request will be for more than that many bits.) */
            /* We fail to do so only if we hit a marker or are forced to suspend. */

            bool noMoreBytes = false;

            if (state.cinfo.m_unread_marker == 0)
            {
                /* cannot advance past a marker */
                while (bits_left < MIN_GET_BITS)
                {
                    int c;
                    state.cinfo.m_src.GetByte(out c);

                    /* If it's 0xFF, check and discard stuffed zero byte */
                    if (c == 0xFF)
                    {
                        /* Loop here to discard any padding FF's on terminating marker,
                        * so that we can save a valid unread_marker value.  NOTE: we will
                        * accept multiple FF's followed by a 0 as meaning a single FF data
                        * byte.  This data pattern is not valid according to the standard.
                        */
                        do
                        {
                            state.cinfo.m_src.GetByte(out c);
                        }
                        while (c == 0xFF);

                        if (c == 0)
                        {
                            /* Found FF/00, which represents an FF data byte */
                            c = 0xFF;
                        }
                        else
                        {
                            /* Oops, it's actually a marker indicating end of compressed data.
                            * Save the marker code for later use.
                            * Fine point: it might appear that we should save the marker into
                            * bitread working state, not straight into permanent state.  But
                            * once we have hit a marker, we cannot need to suspend within the
                            * current MCU, because we will read no more bytes from the data
                            * source.  So it is OK to update permanent state right away.
                            */
                            state.cinfo.m_unread_marker = c;
                            /* See if we need to insert some fake zero bits. */
                            noMoreBytes = true;
                            break;
                        }
                    }

                    /* OK, load c into get_buffer */
                    get_buffer = (get_buffer << 8) | c;
                    bits_left += 8;
                } /* end while */
            }
            else
                noMoreBytes = true;

            if (noMoreBytes)
            {
                /* We get here if we've read the marker that terminates the compressed
                * data segment.  There should be enough bits in the buffer register
                * to satisfy the request; if so, no problem.
                */
                if (nbits > bits_left)
                {
                    /* Uh-oh.  Report corrupted data to user and stuff zeroes into
                    * the data stream, so that we can produce some kind of image.
                    * We use a nonvolatile flag to ensure that only one warning message
                    * appears per data segment.
                    */
                    if (!state.cinfo.m_entropy.m_insufficient_data)
                    {
                        state.cinfo.m_entropy.m_insufficient_data = true;
                    }

                    /* Fill the buffer with zero bits */
                    get_buffer <<= MIN_GET_BITS - bits_left;
                    bits_left = MIN_GET_BITS;
                }
            }

            /* Unload the local registers */
            state.get_buffer = get_buffer;
            state.bits_left = bits_left;

            return true;
        }

        /*
        * Code for extracting next Huffman-coded symbol from input bit stream.
        * Again, this is time-critical and we make the main paths be macros.
        *
        * We use a lookahead table to process codes of up to HuffmanLookaheadDistance bits
        * without looping.  Usually, more than 95% of the Huffman codes will be 8
        * or fewer bits long.  The few overlength codes are handled with a loop,
        * which need not be inline code.
        *
        * Notes about the HUFF_DECODE macro:
        * 1. Near the end of the data segment, we may fail to get enough bits
        *    for a lookahead.  In that case, we do it the hard way.
        * 2. If the lookahead table contains no entry, the next code must be
        *    more than HuffmanLookaheadDistance bits long.
        * 3. jpeg_huff_decode returns -1 if forced to suspend.
        */
        protected static bool HUFF_DECODE(out int result, ref WorkingBitreadState state, DerivedTable htbl, ref int get_buffer, ref int bits_left)
        {
            int nb = 0;
            bool doSlow = false;

            if (bits_left < JpegConstants.HuffmanLookaheadDistance)
            {
                if (!jpeg_fill_bit_buffer(ref state, get_buffer, bits_left, 0))
                {
                    result = -1;
                    return false;
                }

                get_buffer = state.get_buffer;
                bits_left = state.bits_left;
                if (bits_left < JpegConstants.HuffmanLookaheadDistance)
                {
                    nb = 1;
                    doSlow = true;
                }
            }

            if (!doSlow)
            {
                int look = PEEK_BITS(JpegConstants.HuffmanLookaheadDistance, get_buffer, bits_left);
                if ((nb = htbl.look_nbits[look]) != 0)
                {
                    DROP_BITS(nb, ref bits_left);
                    result = htbl.look_sym[look];
                    return true;
                }

                nb = JpegConstants.HuffmanLookaheadDistance + 1;
            }

            result = jpeg_huff_decode(ref state, get_buffer, bits_left, htbl, nb);
            if (result < 0)
                return false;

            get_buffer = state.get_buffer;
            bits_left = state.bits_left;

            return true;
        }

        /* Out-of-line case for Huffman code fetching */
        protected static int jpeg_huff_decode(ref WorkingBitreadState state, int get_buffer, int bits_left, DerivedTable htbl, int min_bits)
        {
            /* HUFF_DECODE has determined that the code is at least min_bits */
            /* bits long, so fetch that many bits in one swoop. */
            int l = min_bits;
            if (!CHECK_BIT_BUFFER(ref state, l, ref get_buffer, ref bits_left))
                return -1;

            int code = GET_BITS(l, get_buffer, ref bits_left);

            /* Collect the rest of the Huffman code one bit at a time. */
            /* This is per Figure F.16 in the JPEG spec. */

            while (code > htbl.maxcode[l])
            {
                code <<= 1;
                if (!CHECK_BIT_BUFFER(ref state, 1, ref get_buffer, ref bits_left))
                    return -1;

                code |= GET_BITS(1, get_buffer, ref bits_left);
                l++;
            }

            /* Unload the local registers */
            state.get_buffer = get_buffer;
            state.bits_left = bits_left;

            /* With garbage input we may reach the sentinel value l = 17. */

            if (l > 16)
            {
                /* fake a zero as the safest result */
                return 0;
            }

            return htbl.pub.Huffval[code + htbl.valoffset[l]];
        }
    }
    #endregion

    #region JpegEntropyEncoder
    /// <summary>
    /// Entropy encoding
    /// </summary>
    abstract class JpegEntropyEncoder
    {
        /* Derived data constructed for each Huffman table */
        protected class c_derived_tbl
        {
            public int[] ehufco = new int[256];   /* code for each symbol */
            public char[] ehufsi = new char[256];       /* length of code for each symbol */
            /* If no code has been allocated for a symbol S, ehufsi[S] contains 0 */
        }

        /* The legal range of a DCT coefficient is
        *  -1024 .. +1023  for 8-bit data;
        * -16384 .. +16383 for 12-bit data.
        * Hence the magnitude should always fit in 10 or 14 bits respectively.
        */
        protected static int MAX_HUFFMAN_COEF_BITS = 10;
        private static int MAX_CLEN = 32;     /* assumed maximum initial code length */

        protected JpegCompressor m_cinfo;

        public abstract void start_pass(bool gather_statistics);
        public abstract bool encode_mcu(JpegBlock[][] MCU_data);
        public abstract void finish_pass();

        /// <summary>
        /// Expand a Huffman table definition into the derived format
        /// Compute the derived values for a Huffman table.
        /// This routine also performs some validation checks on the table.
        /// </summary>
        protected void jpeg_make_c_derived_tbl(bool isDC, int tblno, ref c_derived_tbl dtbl)
        {
            /* Note that huffsize[] and huffcode[] are filled in code-length order,
            * paralleling the order of the symbols themselves in htbl.huffval[].
            */

            /* Find the input Huffman table */
            if (tblno < 0 || tblno >= JpegConstants.NumberOfHuffmanTables)
                throw new Exception(String.Format("Huffman table 0x{0:X2} was not defined", tblno));

            JpegHuffmanTable htbl = isDC ? m_cinfo.m_dc_huff_tbl_ptrs[tblno] : m_cinfo.m_ac_huff_tbl_ptrs[tblno];
            if (htbl == null)
                throw new Exception(String.Format("Huffman table 0x{0:X2} was not defined", tblno));

            /* Allocate a workspace if we haven't already done so. */
            if (dtbl == null)
                dtbl = new c_derived_tbl();

            /* Figure C.1: make table of Huffman code length for each symbol */

            int p = 0;
            char[] huffsize = new char[257];
            for (int l = 1; l <= 16; l++)
            {
                int i = htbl.Bits[l];
                if (i < 0 || p + i > 256)    /* protect against table overrun */
                    throw new Exception("Bogus Huffman table definition");

                while ((i--) != 0)
                    huffsize[p++] = (char)l;
            }
            huffsize[p] = (char)0;
            int lastp = p;

            /* Figure C.2: generate the codes themselves */
            /* We also validate that the counts represent a legal Huffman code tree. */

            int code = 0;
            int si = huffsize[0];
            p = 0;
            int[] huffcode = new int[257];
            while (huffsize[p] != 0)
            {
                while (((int)huffsize[p]) == si)
                {
                    huffcode[p++] = code;
                    code++;
                }
                /* code is now 1 more than the last code used for codelength si; but
                * it must still fit in si bits, since no code is allowed to be all ones.
                */
                if (code >= (1 << si))
                    throw new Exception("Bogus Huffman table definition");
                code <<= 1;
                si++;
            }

            /* Figure C.3: generate encoding tables */
            /* These are code and size indexed by symbol value */

            /* Set all codeless symbols to have code length 0;
            * this lets us detect duplicate VAL entries here, and later
            * allows emit_bits to detect any attempt to emit such symbols.
            */
            Array.Clear(dtbl.ehufsi, 0, dtbl.ehufsi.Length);

            /* This is also a convenient place to check for out-of-range
            * and duplicated VAL entries.  We allow 0..255 for AC symbols
            * but only 0..15 for DC.  (We could constrain them further
            * based on data depth and mode, but this seems enough.)
            */
            int maxsymbol = isDC ? 15 : 255;

            for (p = 0; p < lastp; p++)
            {
                int i = htbl.Huffval[p];
                if (i < 0 || i > maxsymbol || dtbl.ehufsi[i] != 0)
                    throw new Exception("Bogus Huffman table definition");

                dtbl.ehufco[i] = huffcode[p];
                dtbl.ehufsi[i] = huffsize[p];
            }
        }

        /// <summary>
        /// Generate the best Huffman code table for the given counts, fill htbl.
        /// 
        /// The JPEG standard requires that no symbol be assigned a codeword of all
        /// one bits (so that padding bits added at the end of a compressed segment
        /// can't look like a valid code).  Because of the canonical ordering of
        /// codewords, this just means that there must be an unused slot in the
        /// longest codeword length category.  Section K.2 of the JPEG spec suggests
        /// reserving such a slot by pretending that symbol 256 is a valid symbol
        /// with count 1.  In theory that's not optimal; giving it count zero but
        /// including it in the symbol set anyway should give a better Huffman code.
        /// But the theoretically better code actually seems to come out worse in
        /// practice, because it produces more all-ones bytes (which incur stuffed
        /// zero bytes in the final file).  In any case the difference is tiny.
        /// 
        /// The JPEG standard requires Huffman codes to be no more than 16 bits long.
        /// If some symbols have a very small but nonzero probability, the Huffman tree
        /// must be adjusted to meet the code length restriction.  We currently use
        /// the adjustment method suggested in JPEG section K.2.  This method is *not*
        /// optimal; it may not choose the best possible limited-length code.  But
        /// typically only very-low-frequency symbols will be given less-than-optimal
        /// lengths, so the code is almost optimal.  Experimental comparisons against
        /// an optimal limited-length-code algorithm indicate that the difference is
        /// microscopic --- usually less than a hundredth of a percent of total size.
        /// So the extra complexity of an optimal algorithm doesn't seem worthwhile.
        /// </summary>
        protected void jpeg_gen_optimal_table(JpegHuffmanTable htbl, long[] freq)
        {
            byte[] bits = new byte[MAX_CLEN + 1];   /* bits[k] = # of symbols with code length k */
            int[] codesize = new int[257];      /* codesize[k] = code length of symbol k */
            int[] others = new int[257];        /* next symbol in current branch of tree */
            int c1, c2;
            int p, i, j;
            long v;

            /* This algorithm is explained in section K.2 of the JPEG standard */
            for (i = 0; i < 257; i++)
                others[i] = -1;     /* init links to empty */

            freq[256] = 1;      /* make sure 256 has a nonzero count */
            /* Including the pseudo-symbol 256 in the Huffman procedure guarantees
            * that no real symbol is given code-value of all ones, because 256
            * will be placed last in the largest codeword category.
            */

            /* Huffman's basic algorithm to assign optimal code lengths to symbols */

            for (; ; )
            {
                /* Find the smallest nonzero frequency, set c1 = its symbol */
                /* In case of ties, take the larger symbol number */
                c1 = -1;
                v = 1000000000L;
                for (i = 0; i <= 256; i++)
                {
                    if (freq[i] != 0 && freq[i] <= v)
                    {
                        v = freq[i];
                        c1 = i;
                    }
                }

                /* Find the next smallest nonzero frequency, set c2 = its symbol */
                /* In case of ties, take the larger symbol number */
                c2 = -1;
                v = 1000000000L;
                for (i = 0; i <= 256; i++)
                {
                    if (freq[i] != 0 && freq[i] <= v && i != c1)
                    {
                        v = freq[i];
                        c2 = i;
                    }
                }

                /* Done if we've merged everything into one frequency */
                if (c2 < 0)
                    break;

                /* Else merge the two counts/trees */
                freq[c1] += freq[c2];
                freq[c2] = 0;

                /* Increment the codesize of everything in c1's tree branch */
                codesize[c1]++;
                while (others[c1] >= 0)
                {
                    c1 = others[c1];
                    codesize[c1]++;
                }

                others[c1] = c2;        /* chain c2 onto c1's tree branch */

                /* Increment the codesize of everything in c2's tree branch */
                codesize[c2]++;
                while (others[c2] >= 0)
                {
                    c2 = others[c2];
                    codesize[c2]++;
                }
            }

            /* Now count the number of symbols of each code length */
            for (i = 0; i <= 256; i++)
            {
                if (codesize[i] != 0)
                {
                    /* The JPEG standard seems to think that this can't happen, */
                    /* but I'm paranoid... */
                    if (codesize[i] > MAX_CLEN)
                        throw new Exception("Huffman code size table overflow");

                    bits[codesize[i]]++;
                }
            }

            /* JPEG doesn't allow symbols with code lengths over 16 bits, so if the pure
            * Huffman procedure assigned any such lengths, we must adjust the coding.
            * Here is what the JPEG spec says about how this next bit works:
            * Since symbols are paired for the longest Huffman code, the symbols are
            * removed from this length category two at a time.  The prefix for the pair
            * (which is one bit shorter) is allocated to one of the pair; then,
            * skipping the BITS entry for that prefix length, a code word from the next
            * shortest nonzero BITS entry is converted into a prefix for two code words
            * one bit longer.
            */

            for (i = MAX_CLEN; i > 16; i--)
            {
                while (bits[i] > 0)
                {
                    j = i - 2;      /* find length of new prefix to be used */
                    while (bits[j] == 0)
                        j--;

                    bits[i] -= 2;       /* remove two symbols */
                    bits[i - 1]++;      /* one goes in this length */
                    bits[j + 1] += 2;       /* two new symbols in this length */
                    bits[j]--;      /* symbol of this length is now a prefix */
                }
            }

            /* Remove the count for the pseudo-symbol 256 from the largest codelength */
            while (bits[i] == 0)        /* find largest codelength still in use */
                i--;
            bits[i]--;

            /* Return final symbol counts (only for lengths 0..16) */
            Buffer.BlockCopy(bits, 0, htbl.Bits, 0, htbl.Bits.Length);

            /* Return a list of the symbols sorted by code length */
            /* It's not real clear to me why we don't need to consider the codelength
            * changes made above, but the JPEG spec seems to think this works.
            */
            p = 0;
            for (i = 1; i <= MAX_CLEN; i++)
            {
                for (j = 0; j <= 255; j++)
                {
                    if (codesize[j] == i)
                    {
                        htbl.Huffval[p] = (byte)j;
                        p++;
                    }
                }
            }

            /* Set sent_table false so updated table will be written to JPEG file. */
            htbl.Sent_table = false;
        }
    }
    #endregion

    #region JpegFowardDCT
    /// <summary>
    /// Forward DCT (also controls coefficient quantization)
    /// 
    /// A forward DCT routine is given a pointer to a work area of type DCTELEM[];
    /// the DCT is to be performed in-place in that buffer.  Type DCTELEM is int
    /// for 8-bit samples, int for 12-bit samples.  (NOTE: Floating-point DCT
    /// implementations use an array of type float, instead.)
    /// The DCT inputs are expected to be signed (range +-MediumSampleValue).
    /// The DCT outputs are returned scaled up by a factor of 8; they therefore
    /// have a range of +-8K for 8-bit data, +-128K for 12-bit data. This
    /// convention improves accuracy in integer implementations and saves some
    /// work in floating-point ones.
    /// 
    /// Each IDCT routine has its own ideas about the best dct_table element type.
    /// </summary>
    class JpegFowardDCT
    {
        private const int FAST_INTEGER_CONST_BITS = 8;

        /* We use the following pre-calculated constants.
        * If you change FAST_INTEGER_CONST_BITS you may want to add appropriate values.
        * 
        * Convert a positive real constant to an integer scaled by CONST_SCALE.
        * static int FAST_INTEGER_FIX(double x)
        *{
        *    return ((int) ((x) * (((int) 1) << FAST_INTEGER_CONST_BITS) + 0.5));
        *}
        */
        private const int FAST_INTEGER_FIX_0_382683433 = 98;        /* FIX(0.382683433) */
        private const int FAST_INTEGER_FIX_0_541196100 = 139;       /* FIX(0.541196100) */
        private const int FAST_INTEGER_FIX_0_707106781 = 181;       /* FIX(0.707106781) */
        private const int FAST_INTEGER_FIX_1_306562965 = 334;       /* FIX(1.306562965) */

        private const int SLOW_INTEGER_CONST_BITS = 13;
        private const int SLOW_INTEGER_PASS1_BITS = 2;

        /* We use the following pre-calculated constants.
        * If you change SLOW_INTEGER_CONST_BITS you may want to add appropriate values.
        * 
        * Convert a positive real constant to an integer scaled by CONST_SCALE.
        *
        * static int SLOW_INTEGER_FIX(double x)
        * {
        *     return ((int) ((x) * (((int) 1) << SLOW_INTEGER_CONST_BITS) + 0.5));
        * }
        */
        private const int SLOW_INTEGER_FIX_0_298631336 = 2446;   /* FIX(0.298631336) */
        private const int SLOW_INTEGER_FIX_0_390180644 = 3196;   /* FIX(0.390180644) */
        private const int SLOW_INTEGER_FIX_0_541196100 = 4433;   /* FIX(0.541196100) */
        private const int SLOW_INTEGER_FIX_0_765366865 = 6270;   /* FIX(0.765366865) */
        private const int SLOW_INTEGER_FIX_0_899976223 = 7373;   /* FIX(0.899976223) */
        private const int SLOW_INTEGER_FIX_1_175875602 = 9633;   /* FIX(1.175875602) */
        private const int SLOW_INTEGER_FIX_1_501321110 = 12299;  /* FIX(1.501321110) */
        private const int SLOW_INTEGER_FIX_1_847759065 = 15137;  /* FIX(1.847759065) */
        private const int SLOW_INTEGER_FIX_1_961570560 = 16069;  /* FIX(1.961570560) */
        private const int SLOW_INTEGER_FIX_2_053119869 = 16819;  /* FIX(2.053119869) */
        private const int SLOW_INTEGER_FIX_2_562915447 = 20995;  /* FIX(2.562915447) */
        private const int SLOW_INTEGER_FIX_3_072711026 = 25172;  /* FIX(3.072711026) */

        /* For AA&N IDCT method, divisors are equal to quantization
         * coefficients scaled by scalefactor[row]*scalefactor[col], where
         *   scalefactor[0] = 1
         *   scalefactor[k] = cos(k*PI/16) * sqrt(2)    for k=1..7
         * We apply a further scale factor of 8.
         */
        private const int CONST_BITS = 14;

        /* precomputed values scaled up by 14 bits */
        private static short[] aanscales = { 
            16384, 22725, 21407, 19266, 16384, 12873, 8867, 4520, 22725, 31521, 29692, 26722, 22725, 17855,
            12299, 6270, 21407, 29692, 27969, 25172, 21407, 16819, 11585,
            5906, 19266, 26722, 25172, 22654, 19266, 15137, 10426, 5315,
            16384, 22725, 21407, 19266, 16384, 12873, 8867, 4520, 12873,
            17855, 16819, 15137, 12873, 10114, 6967, 3552, 8867, 12299,
            11585, 10426, 8867, 6967, 4799, 2446, 4520, 6270, 5906, 5315,
            4520, 3552, 2446, 1247 };

        /* For float AA&N IDCT method, divisors are equal to quantization
         * coefficients scaled by scalefactor[row]*scalefactor[col], where
         *   scalefactor[0] = 1
         *   scalefactor[k] = cos(k*PI/16) * sqrt(2)    for k=1..7
         * We apply a further scale factor of 8.
         * What's actually stored is 1/divisor so that the inner loop can
         * use a multiplication rather than a division.
         */
        private static double[] aanscalefactor = { 
            1.0, 1.387039845, 1.306562965, 1.175875602, 1.0,
            0.785694958, 0.541196100, 0.275899379 };

        private JpegCompressor m_cinfo;
        private bool m_useSlowMethod;
        private bool m_useFloatMethod;

        /* The actual post-DCT divisors --- not identical to the quant table
        * entries, because of scaling (especially for an unnormalized DCT).
        * Each table is given in normal array order.
        */
        private int[][] m_divisors = new int[JpegConstants.NumberOfQuantTables][];

        /* Same as above for the floating-point case. */
        private float[][] m_float_divisors = new float[JpegConstants.NumberOfQuantTables][];

        public JpegFowardDCT(JpegCompressor cinfo)
        {
            m_cinfo = cinfo;

            switch (cinfo.m_dct_method)
            {
                case DCTMethod.IntSlow:
                    m_useFloatMethod = false;
                    m_useSlowMethod = true;
                    break;
                case DCTMethod.IntFast:
                    m_useFloatMethod = false;
                    m_useSlowMethod = false;
                    break;
                case DCTMethod.Float:
                    m_useFloatMethod = true;
                    break;
                default:
                    throw new Exception("Unknown dct method!");
            }

            /* Mark divisor tables unallocated */
            for (int i = 0; i < JpegConstants.NumberOfQuantTables; i++)
            {
                m_divisors[i] = null;
                m_float_divisors[i] = null;
            }
        }

        /// <summary>
        /// Initialize for a processing pass.
        /// Verify that all referenced Q-tables are present, and set up
        /// the divisor table for each one.
        /// In the current implementation, DCT of all components is done during
        /// the first pass, even if only some components will be output in the
        /// first scan.  Hence all components should be examined here.
        /// </summary>
        public virtual void start_pass()
        {
            for (int ci = 0; ci < m_cinfo.m_num_components; ci++)
            {
                int qtblno = m_cinfo.Component_info[ci].Quant_tbl_no;

                /* Make sure specified quantization table is present */
                if (qtblno < 0 || qtblno >= JpegConstants.NumberOfQuantTables || m_cinfo.m_quant_tbl_ptrs[qtblno] == null)
                    throw new Exception(String.Format("Quantization table 0x{0:X2} was not defined", qtblno));

                JpegQuantizationTable qtbl = m_cinfo.m_quant_tbl_ptrs[qtblno];

                /* Compute divisors for this quant table */
                /* We may do this more than once for same table, but it's not a big deal */
                int i = 0;
                switch (m_cinfo.m_dct_method)
                {
                    case DCTMethod.IntSlow:
                        /* For LL&M IDCT method, divisors are equal to raw quantization
                         * coefficients multiplied by 8 (to counteract scaling).
                         */
                        if (m_divisors[qtblno] == null)
                            m_divisors[qtblno] = new int[JpegConstants.DCTSize2];

                        for (i = 0; i < JpegConstants.DCTSize2; i++)
                            m_divisors[qtblno][i] = ((int)qtbl.quantval[i]) << 3;

                        break;
                    case DCTMethod.IntFast:
                        if (m_divisors[qtblno] == null)
                            m_divisors[qtblno] = new int[JpegConstants.DCTSize2];

                        for (i = 0; i < JpegConstants.DCTSize2; i++)
                            m_divisors[qtblno][i] = JpegUtils.DESCALE((int)qtbl.quantval[i] * (int)aanscales[i], CONST_BITS - 3);
                        break;
                    case DCTMethod.Float:
                        if (m_float_divisors[qtblno] == null)
                            m_float_divisors[qtblno] = new float[JpegConstants.DCTSize2];

                        float[] fdtbl = m_float_divisors[qtblno];
                        i = 0;
                        for (int row = 0; row < JpegConstants.DCTSize; row++)
                        {
                            for (int col = 0; col < JpegConstants.DCTSize; col++)
                            {
                                fdtbl[i] = (float)(1.0 / (((double)qtbl.quantval[i] * aanscalefactor[row] * aanscalefactor[col] * 8.0)));
                                i++;
                            }
                        }
                        break;
                    default:
                        throw new Exception("Unknown dct method!");
                }
            }
        }

        /// <summary>
        /// Perform forward DCT on one or more blocks of a component.
        /// 
        /// The input samples are taken from the sample_data[] array starting at
        /// position start_row/start_col, and moving to the right for any additional
        /// blocks. The quantized coefficients are returned in coef_blocks[].
        /// </summary>
        public virtual void forward_DCT(int quant_tbl_no, byte[][] sample_data, JpegBlock[] coef_blocks, int start_row, int start_col, int num_blocks)
        {
            if (m_useFloatMethod)
                forwardDCTFloatImpl(quant_tbl_no, sample_data, coef_blocks, start_row, start_col, num_blocks);
            else
                forwardDCTImpl(quant_tbl_no, sample_data, coef_blocks, start_row, start_col, num_blocks);
        }

        // This version is used for integer DCT implementations.
        private void forwardDCTImpl(int quant_tbl_no, byte[][] sample_data, JpegBlock[] coef_blocks, int start_row, int start_col, int num_blocks)
        {
            /* This routine is heavily used, so it's worth coding it tightly. */
            int[] workspace = new int[JpegConstants.DCTSize2];    /* work area for FDCT subroutine */
            for (int bi = 0; bi < num_blocks; bi++, start_col += JpegConstants.DCTSize)
            {
                /* Load data into workspace, applying unsigned->signed conversion */
                int workspaceIndex = 0;
                for (int elemr = 0; elemr < JpegConstants.DCTSize; elemr++)
                {
                    for (int column = 0; column < JpegConstants.DCTSize; column++)
                    {
                        workspace[workspaceIndex] = (int)sample_data[start_row + elemr][start_col + column] - JpegConstants.MediumSampleValue;
                        workspaceIndex++;
                    }
                }

                /* Perform the DCT */
                if (m_useSlowMethod)
                    jpeg_fdct_islow(workspace);
                else
                    jpeg_fdct_ifast(workspace);

                /* Quantize/descale the coefficients, and store into coef_blocks[] */
                for (int i = 0; i < JpegConstants.DCTSize2; i++)
                {
                    int qval = m_divisors[quant_tbl_no][i];
                    int temp = workspace[i];

                    if (temp < 0)
                    {
                        temp = -temp;
                        temp += qval >> 1;  /* for rounding */

                        if (temp >= qval)
                            temp /= qval;
                        else
                            temp = 0;

                        temp = -temp;
                    }
                    else
                    {
                        temp += qval >> 1;  /* for rounding */

                        if (temp >= qval)
                            temp /= qval;
                        else
                            temp = 0;
                    }

                    coef_blocks[bi][i] = (short)temp;
                }
            }
        }

        // This version is used for floating-point DCT implementations.
        private void forwardDCTFloatImpl(int quant_tbl_no, byte[][] sample_data, JpegBlock[] coef_blocks, int start_row, int start_col, int num_blocks)
        {
            /* This routine is heavily used, so it's worth coding it tightly. */
            float[] workspace = new float[JpegConstants.DCTSize2]; /* work area for FDCT subroutine */
            for (int bi = 0; bi < num_blocks; bi++, start_col += JpegConstants.DCTSize)
            {
                /* Load data into workspace, applying unsigned->signed conversion */
                int workspaceIndex = 0;
                for (int elemr = 0; elemr < JpegConstants.DCTSize; elemr++)
                {
                    for (int column = 0; column < JpegConstants.DCTSize; column++)
                    {
                        workspace[workspaceIndex] = (float)((int)sample_data[start_row + elemr][start_col + column] - JpegConstants.MediumSampleValue);
                        workspaceIndex++;
                    }
                }

                /* Perform the DCT */
                jpeg_fdct_float(workspace);

                /* Quantize/descale the coefficients, and store into coef_blocks[] */
                for (int i = 0; i < JpegConstants.DCTSize2; i++)
                {
                    /* Apply the quantization and scaling factor */
                    float temp = workspace[i] * m_float_divisors[quant_tbl_no][i];

                    /* Round to nearest integer.
                     * Since C does not specify the direction of rounding for negative
                     * quotients, we have to force the dividend positive for portability.
                     * The maximum coefficient size is +-16K (for 12-bit data), so this
                     * code should work for either 16-bit or 32-bit ints.
                     */
                    coef_blocks[bi][i] = (short)((int)(temp + (float)16384.5) - 16384);
                }
            }
        }

        /// <summary>
        /// Perform the forward DCT on one block of samples.
        /// NOTE: this code only copes with 8x8 DCTs.
        /// 
        /// A floating-point implementation of the 
        /// forward DCT (Discrete Cosine Transform).
        /// 
        /// This implementation should be more accurate than either of the integer
        /// DCT implementations.  However, it may not give the same results on all
        /// machines because of differences in roundoff behavior.  Speed will depend
        /// on the hardware's floating point capacity.
        /// 
        /// A 2-D DCT can be done by 1-D DCT on each row followed by 1-D DCT
        /// on each column.  Direct algorithms are also available, but they are
        /// much more complex and seem not to be any faster when reduced to code.
        /// 
        /// This implementation is based on Arai, Agui, and Nakajima's algorithm for
        /// scaled DCT.  Their original paper (Trans. IEICE E-71(11):1095) is in
        /// Japanese, but the algorithm is described in the Pennebaker &amp; Mitchell
        /// JPEG textbook (see REFERENCES section in file README).  The following code
        /// is based directly on figure 4-8 in P&amp;M.
        /// While an 8-point DCT cannot be done in less than 11 multiplies, it is
        /// possible to arrange the computation so that many of the multiplies are
        /// simple scalings of the final outputs.  These multiplies can then be
        /// folded into the multiplications or divisions by the JPEG quantization
        /// table entries.  The AA&amp;N method leaves only 5 multiplies and 29 adds
        /// to be done in the DCT itself.
        /// The primary disadvantage of this method is that with a fixed-point
        /// implementation, accuracy is lost due to imprecise representation of the
        /// scaled quantization values.  However, that problem does not arise if
        /// we use floating point arithmetic.
        /// </summary>
        private static void jpeg_fdct_float(float[] data)
        {
            /* Pass 1: process rows. */
            int dataIndex = 0;
            for (int ctr = JpegConstants.DCTSize - 1; ctr >= 0; ctr--)
            {
                float tmp0 = data[dataIndex + 0] + data[dataIndex + 7];
                float tmp7 = data[dataIndex + 0] - data[dataIndex + 7];
                float tmp1 = data[dataIndex + 1] + data[dataIndex + 6];
                float tmp6 = data[dataIndex + 1] - data[dataIndex + 6];
                float tmp2 = data[dataIndex + 2] + data[dataIndex + 5];
                float tmp5 = data[dataIndex + 2] - data[dataIndex + 5];
                float tmp3 = data[dataIndex + 3] + data[dataIndex + 4];
                float tmp4 = data[dataIndex + 3] - data[dataIndex + 4];

                /* Even part */

                float tmp10 = tmp0 + tmp3;    /* phase 2 */
                float tmp13 = tmp0 - tmp3;
                float tmp11 = tmp1 + tmp2;
                float tmp12 = tmp1 - tmp2;

                data[dataIndex + 0] = tmp10 + tmp11; /* phase 3 */
                data[dataIndex + 4] = tmp10 - tmp11;

                float z1 = (tmp12 + tmp13) * ((float)0.707106781); /* c4 */
                data[dataIndex + 2] = tmp13 + z1;    /* phase 5 */
                data[dataIndex + 6] = tmp13 - z1;

                /* Odd part */

                tmp10 = tmp4 + tmp5;    /* phase 2 */
                tmp11 = tmp5 + tmp6;
                tmp12 = tmp6 + tmp7;

                /* The rotator is modified from fig 4-8 to avoid extra negations. */
                float z5 = (tmp10 - tmp12) * ((float)0.382683433); /* c6 */
                float z2 = ((float)0.541196100) * tmp10 + z5; /* c2-c6 */
                float z4 = ((float)1.306562965) * tmp12 + z5; /* c2+c6 */
                float z3 = tmp11 * ((float)0.707106781); /* c4 */

                float z11 = tmp7 + z3;        /* phase 5 */
                float z13 = tmp7 - z3;

                data[dataIndex + 5] = z13 + z2;  /* phase 6 */
                data[dataIndex + 3] = z13 - z2;
                data[dataIndex + 1] = z11 + z4;
                data[dataIndex + 7] = z11 - z4;

                dataIndex += JpegConstants.DCTSize;     /* advance pointer to next row */
            }

            /* Pass 2: process columns. */

            dataIndex = 0;
            for (int ctr = JpegConstants.DCTSize - 1; ctr >= 0; ctr--)
            {
                float tmp0 = data[dataIndex + JpegConstants.DCTSize * 0] + data[dataIndex + JpegConstants.DCTSize * 7];
                float tmp7 = data[dataIndex + JpegConstants.DCTSize * 0] - data[dataIndex + JpegConstants.DCTSize * 7];
                float tmp1 = data[dataIndex + JpegConstants.DCTSize * 1] + data[dataIndex + JpegConstants.DCTSize * 6];
                float tmp6 = data[dataIndex + JpegConstants.DCTSize * 1] - data[dataIndex + JpegConstants.DCTSize * 6];
                float tmp2 = data[dataIndex + JpegConstants.DCTSize * 2] + data[dataIndex + JpegConstants.DCTSize * 5];
                float tmp5 = data[dataIndex + JpegConstants.DCTSize * 2] - data[dataIndex + JpegConstants.DCTSize * 5];
                float tmp3 = data[dataIndex + JpegConstants.DCTSize * 3] + data[dataIndex + JpegConstants.DCTSize * 4];
                float tmp4 = data[dataIndex + JpegConstants.DCTSize * 3] - data[dataIndex + JpegConstants.DCTSize * 4];

                /* Even part */

                float tmp10 = tmp0 + tmp3;    /* phase 2 */
                float tmp13 = tmp0 - tmp3;
                float tmp11 = tmp1 + tmp2;
                float tmp12 = tmp1 - tmp2;

                data[dataIndex + JpegConstants.DCTSize * 0] = tmp10 + tmp11; /* phase 3 */
                data[dataIndex + JpegConstants.DCTSize * 4] = tmp10 - tmp11;

                float z1 = (tmp12 + tmp13) * ((float)0.707106781); /* c4 */
                data[dataIndex + JpegConstants.DCTSize * 2] = tmp13 + z1; /* phase 5 */
                data[dataIndex + JpegConstants.DCTSize * 6] = tmp13 - z1;

                /* Odd part */

                tmp10 = tmp4 + tmp5;    /* phase 2 */
                tmp11 = tmp5 + tmp6;
                tmp12 = tmp6 + tmp7;

                /* The rotator is modified from fig 4-8 to avoid extra negations. */
                float z5 = (tmp10 - tmp12) * ((float)0.382683433); /* c6 */
                float z2 = ((float)0.541196100) * tmp10 + z5; /* c2-c6 */
                float z4 = ((float)1.306562965) * tmp12 + z5; /* c2+c6 */
                float z3 = tmp11 * ((float)0.707106781); /* c4 */

                float z11 = tmp7 + z3;        /* phase 5 */
                float z13 = tmp7 - z3;

                data[dataIndex + JpegConstants.DCTSize * 5] = z13 + z2; /* phase 6 */
                data[dataIndex + JpegConstants.DCTSize * 3] = z13 - z2;
                data[dataIndex + JpegConstants.DCTSize * 1] = z11 + z4;
                data[dataIndex + JpegConstants.DCTSize * 7] = z11 - z4;

                dataIndex++;          /* advance pointer to next column */
            }
        }

        /// <summary>
        /// Perform the forward DCT on one block of samples.
        /// NOTE: this code only copes with 8x8 DCTs.
        /// This file contains a fast, not so accurate integer implementation of the
        /// forward DCT (Discrete Cosine Transform).
        /// 
        /// A 2-D DCT can be done by 1-D DCT on each row followed by 1-D DCT
        /// on each column.  Direct algorithms are also available, but they are
        /// much more complex and seem not to be any faster when reduced to code.
        /// 
        /// This implementation is based on Arai, Agui, and Nakajima's algorithm for
        /// scaled DCT.  Their original paper (Trans. IEICE E-71(11):1095) is in
        /// Japanese, but the algorithm is described in the Pennebaker &amp; Mitchell
        /// JPEG textbook (see REFERENCES section in file README).  The following code
        /// is based directly on figure 4-8 in P&amp;M.
        /// While an 8-point DCT cannot be done in less than 11 multiplies, it is
        /// possible to arrange the computation so that many of the multiplies are
        /// simple scalings of the final outputs.  These multiplies can then be
        /// folded into the multiplications or divisions by the JPEG quantization
        /// table entries.  The AA&amp;N method leaves only 5 multiplies and 29 adds
        /// to be done in the DCT itself.
        /// The primary disadvantage of this method is that with fixed-point math,
        /// accuracy is lost due to imprecise representation of the scaled
        /// quantization values.  The smaller the quantization table entry, the less
        /// precise the scaled value, so this implementation does worse with high-
        /// quality-setting files than with low-quality ones.
        /// 
        /// Scaling decisions are generally the same as in the LL&amp;M algorithm;
        /// see jpeg_fdct_islow for more details.  However, we choose to descale
        /// (right shift) multiplication products as soon as they are formed,
        /// rather than carrying additional fractional bits into subsequent additions.
        /// This compromises accuracy slightly, but it lets us save a few shifts.
        /// More importantly, 16-bit arithmetic is then adequate (for 8-bit samples)
        /// everywhere except in the multiplications proper; this saves a good deal
        /// of work on 16-bit-int machines.
        /// 
        /// Again to save a few shifts, the intermediate results between pass 1 and
        /// pass 2 are not upscaled, but are represented only to integral precision.
        /// 
        /// A final compromise is to represent the multiplicative constants to only
        /// 8 fractional bits, rather than 13.  This saves some shifting work on some
        /// machines, and may also reduce the cost of multiplication (since there
        /// are fewer one-bits in the constants).
        /// </summary>
        private static void jpeg_fdct_ifast(int[] data)
        {
            /* Pass 1: process rows. */
            int dataIndex = 0;
            for (int ctr = JpegConstants.DCTSize - 1; ctr >= 0; ctr--)
            {
                int tmp0 = data[dataIndex + 0] + data[dataIndex + 7];
                int tmp7 = data[dataIndex + 0] - data[dataIndex + 7];
                int tmp1 = data[dataIndex + 1] + data[dataIndex + 6];
                int tmp6 = data[dataIndex + 1] - data[dataIndex + 6];
                int tmp2 = data[dataIndex + 2] + data[dataIndex + 5];
                int tmp5 = data[dataIndex + 2] - data[dataIndex + 5];
                int tmp3 = data[dataIndex + 3] + data[dataIndex + 4];
                int tmp4 = data[dataIndex + 3] - data[dataIndex + 4];

                /* Even part */

                int tmp10 = tmp0 + tmp3;    /* phase 2 */
                int tmp13 = tmp0 - tmp3;
                int tmp11 = tmp1 + tmp2;
                int tmp12 = tmp1 - tmp2;

                data[dataIndex + 0] = tmp10 + tmp11; /* phase 3 */
                data[dataIndex + 4] = tmp10 - tmp11;

                int z1 = FAST_INTEGER_MULTIPLY(tmp12 + tmp13, FAST_INTEGER_FIX_0_707106781); /* c4 */
                data[dataIndex + 2] = tmp13 + z1;    /* phase 5 */
                data[dataIndex + 6] = tmp13 - z1;

                /* Odd part */

                tmp10 = tmp4 + tmp5;    /* phase 2 */
                tmp11 = tmp5 + tmp6;
                tmp12 = tmp6 + tmp7;

                /* The rotator is modified from fig 4-8 to avoid extra negations. */
                int z5 = FAST_INTEGER_MULTIPLY(tmp10 - tmp12, FAST_INTEGER_FIX_0_382683433); /* c6 */
                int z2 = FAST_INTEGER_MULTIPLY(tmp10, FAST_INTEGER_FIX_0_541196100) + z5; /* c2-c6 */
                int z4 = FAST_INTEGER_MULTIPLY(tmp12, FAST_INTEGER_FIX_1_306562965) + z5; /* c2+c6 */
                int z3 = FAST_INTEGER_MULTIPLY(tmp11, FAST_INTEGER_FIX_0_707106781); /* c4 */

                int z11 = tmp7 + z3;        /* phase 5 */
                int z13 = tmp7 - z3;

                data[dataIndex + 5] = z13 + z2;  /* phase 6 */
                data[dataIndex + 3] = z13 - z2;
                data[dataIndex + 1] = z11 + z4;
                data[dataIndex + 7] = z11 - z4;

                dataIndex += JpegConstants.DCTSize;     /* advance pointer to next row */
            }

            /* Pass 2: process columns. */

            dataIndex = 0;
            for (int ctr = JpegConstants.DCTSize - 1; ctr >= 0; ctr--)
            {
                int tmp0 = data[dataIndex + JpegConstants.DCTSize * 0] + data[dataIndex + JpegConstants.DCTSize * 7];
                int tmp7 = data[dataIndex + JpegConstants.DCTSize * 0] - data[dataIndex + JpegConstants.DCTSize * 7];
                int tmp1 = data[dataIndex + JpegConstants.DCTSize * 1] + data[dataIndex + JpegConstants.DCTSize * 6];
                int tmp6 = data[dataIndex + JpegConstants.DCTSize * 1] - data[dataIndex + JpegConstants.DCTSize * 6];
                int tmp2 = data[dataIndex + JpegConstants.DCTSize * 2] + data[dataIndex + JpegConstants.DCTSize * 5];
                int tmp5 = data[dataIndex + JpegConstants.DCTSize * 2] - data[dataIndex + JpegConstants.DCTSize * 5];
                int tmp3 = data[dataIndex + JpegConstants.DCTSize * 3] + data[dataIndex + JpegConstants.DCTSize * 4];
                int tmp4 = data[dataIndex + JpegConstants.DCTSize * 3] - data[dataIndex + JpegConstants.DCTSize * 4];

                /* Even part */

                int tmp10 = tmp0 + tmp3;    /* phase 2 */
                int tmp13 = tmp0 - tmp3;
                int tmp11 = tmp1 + tmp2;
                int tmp12 = tmp1 - tmp2;

                data[dataIndex + JpegConstants.DCTSize * 0] = tmp10 + tmp11; /* phase 3 */
                data[dataIndex + JpegConstants.DCTSize * 4] = tmp10 - tmp11;

                int z1 = FAST_INTEGER_MULTIPLY(tmp12 + tmp13, FAST_INTEGER_FIX_0_707106781); /* c4 */
                data[dataIndex + JpegConstants.DCTSize * 2] = tmp13 + z1; /* phase 5 */
                data[dataIndex + JpegConstants.DCTSize * 6] = tmp13 - z1;

                /* Odd part */

                tmp10 = tmp4 + tmp5;    /* phase 2 */
                tmp11 = tmp5 + tmp6;
                tmp12 = tmp6 + tmp7;

                /* The rotator is modified from fig 4-8 to avoid extra negations. */
                int z5 = FAST_INTEGER_MULTIPLY(tmp10 - tmp12, FAST_INTEGER_FIX_0_382683433); /* c6 */
                int z2 = FAST_INTEGER_MULTIPLY(tmp10, FAST_INTEGER_FIX_0_541196100) + z5; /* c2-c6 */
                int z4 = FAST_INTEGER_MULTIPLY(tmp12, FAST_INTEGER_FIX_1_306562965) + z5; /* c2+c6 */
                int z3 = FAST_INTEGER_MULTIPLY(tmp11, FAST_INTEGER_FIX_0_707106781); /* c4 */

                int z11 = tmp7 + z3;        /* phase 5 */
                int z13 = tmp7 - z3;

                data[dataIndex + JpegConstants.DCTSize * 5] = z13 + z2; /* phase 6 */
                data[dataIndex + JpegConstants.DCTSize * 3] = z13 - z2;
                data[dataIndex + JpegConstants.DCTSize * 1] = z11 + z4;
                data[dataIndex + JpegConstants.DCTSize * 7] = z11 - z4;

                dataIndex++;          /* advance pointer to next column */
            }
        }

        /// <summary>
        /// Perform the forward DCT on one block of samples.
        /// NOTE: this code only copes with 8x8 DCTs.
        /// 
        /// A slow-but-accurate integer implementation of the
        /// forward DCT (Discrete Cosine Transform).
        /// 
        /// A 2-D DCT can be done by 1-D DCT on each row followed by 1-D DCT
        /// on each column.  Direct algorithms are also available, but they are
        /// much more complex and seem not to be any faster when reduced to code.
        /// 
        /// This implementation is based on an algorithm described in
        /// C. Loeffler, A. Ligtenberg and G. Moschytz, "Practical Fast 1-D DCT
        /// Algorithms with 11 Multiplications", Proc. Int'l. Conf. on Acoustics,
        /// Speech, and Signal Processing 1989 (ICASSP '89), pp. 988-991.
        /// The primary algorithm described there uses 11 multiplies and 29 adds.
        /// We use their alternate method with 12 multiplies and 32 adds.
        /// The advantage of this method is that no data path contains more than one
        /// multiplication; this allows a very simple and accurate implementation in
        /// scaled fixed-point arithmetic, with a minimal number of shifts.
        /// 
        /// The poop on this scaling stuff is as follows:
        /// 
        /// Each 1-D DCT step produces outputs which are a factor of sqrt(N)
        /// larger than the true DCT outputs.  The final outputs are therefore
        /// a factor of N larger than desired; since N=8 this can be cured by
        /// a simple right shift at the end of the algorithm.  The advantage of
        /// this arrangement is that we save two multiplications per 1-D DCT,
        /// because the y0 and y4 outputs need not be divided by sqrt(N).
        /// In the IJG code, this factor of 8 is removed by the quantization 
        /// step, NOT here.
        /// 
        /// We have to do addition and subtraction of the integer inputs, which
        /// is no problem, and multiplication by fractional constants, which is
        /// a problem to do in integer arithmetic.  We multiply all the constants
        /// by CONST_SCALE and convert them to integer constants (thus retaining
        /// SLOW_INTEGER_CONST_BITS bits of precision in the constants).  After doing a
        /// multiplication we have to divide the product by CONST_SCALE, with proper
        /// rounding, to produce the correct output.  This division can be done
        /// cheaply as a right shift of SLOW_INTEGER_CONST_BITS bits.  We postpone shifting
        /// as long as possible so that partial sums can be added together with
        /// full fractional precision.
        /// 
        /// The outputs of the first pass are scaled up by SLOW_INTEGER_PASS1_BITS bits so that
        /// they are represented to better-than-integral precision.  These outputs
        /// require BitsInSample + SLOW_INTEGER_PASS1_BITS + 3 bits; this fits in a 16-bit word
        /// with the recommended scaling.  (For 12-bit sample data, the intermediate
        /// array is int anyway.)
        /// 
        /// To avoid overflow of the 32-bit intermediate results in pass 2, we must
        /// have BitsInSample + SLOW_INTEGER_CONST_BITS + SLOW_INTEGER_PASS1_BITS &lt;= 26.  Error analysis
        /// shows that the values given below are the most effective.
        /// </summary>
        private static void jpeg_fdct_islow(int[] data)
        {
            /* Pass 1: process rows. */
            /* Note results are scaled up by sqrt(8) compared to a true DCT; */
            /* furthermore, we scale the results by 2**SLOW_INTEGER_PASS1_BITS. */
            int dataIndex = 0;
            for (int ctr = JpegConstants.DCTSize - 1; ctr >= 0; ctr--)
            {
                int tmp0 = data[dataIndex + 0] + data[dataIndex + 7];
                int tmp7 = data[dataIndex + 0] - data[dataIndex + 7];
                int tmp1 = data[dataIndex + 1] + data[dataIndex + 6];
                int tmp6 = data[dataIndex + 1] - data[dataIndex + 6];
                int tmp2 = data[dataIndex + 2] + data[dataIndex + 5];
                int tmp5 = data[dataIndex + 2] - data[dataIndex + 5];
                int tmp3 = data[dataIndex + 3] + data[dataIndex + 4];
                int tmp4 = data[dataIndex + 3] - data[dataIndex + 4];

                /* Even part per LL&M figure 1 --- note that published figure is faulty;
                * rotator "sqrt(2)*c1" should be "sqrt(2)*c6".
                */

                int tmp10 = tmp0 + tmp3;
                int tmp13 = tmp0 - tmp3;
                int tmp11 = tmp1 + tmp2;
                int tmp12 = tmp1 - tmp2;

                data[dataIndex + 0] = (tmp10 + tmp11) << SLOW_INTEGER_PASS1_BITS;
                data[dataIndex + 4] = (tmp10 - tmp11) << SLOW_INTEGER_PASS1_BITS;

                int z1 = (tmp12 + tmp13) * SLOW_INTEGER_FIX_0_541196100;
                data[dataIndex + 2] = JpegUtils.DESCALE(z1 + tmp13 * SLOW_INTEGER_FIX_0_765366865,
                                                SLOW_INTEGER_CONST_BITS - SLOW_INTEGER_PASS1_BITS);
                data[dataIndex + 6] = JpegUtils.DESCALE(z1 + tmp12 * (-SLOW_INTEGER_FIX_1_847759065),
                                                SLOW_INTEGER_CONST_BITS - SLOW_INTEGER_PASS1_BITS);

                /* Odd part per figure 8 --- note paper omits factor of sqrt(2).
                * cK represents cos(K*pi/16).
                * i0..i3 in the paper are tmp4..tmp7 here.
                */

                z1 = tmp4 + tmp7;
                int z2 = tmp5 + tmp6;
                int z3 = tmp4 + tmp6;
                int z4 = tmp5 + tmp7;
                int z5 = (z3 + z4) * SLOW_INTEGER_FIX_1_175875602; /* sqrt(2) * c3 */

                tmp4 = tmp4 * SLOW_INTEGER_FIX_0_298631336; /* sqrt(2) * (-c1+c3+c5-c7) */
                tmp5 = tmp5 * SLOW_INTEGER_FIX_2_053119869; /* sqrt(2) * ( c1+c3-c5+c7) */
                tmp6 = tmp6 * SLOW_INTEGER_FIX_3_072711026; /* sqrt(2) * ( c1+c3+c5-c7) */
                tmp7 = tmp7 * SLOW_INTEGER_FIX_1_501321110; /* sqrt(2) * ( c1+c3-c5-c7) */
                z1 = z1 * (-SLOW_INTEGER_FIX_0_899976223); /* sqrt(2) * (c7-c3) */
                z2 = z2 * (-SLOW_INTEGER_FIX_2_562915447); /* sqrt(2) * (-c1-c3) */
                z3 = z3 * (-SLOW_INTEGER_FIX_1_961570560); /* sqrt(2) * (-c3-c5) */
                z4 = z4 * (-SLOW_INTEGER_FIX_0_390180644); /* sqrt(2) * (c5-c3) */

                z3 += z5;
                z4 += z5;

                data[dataIndex + 7] = JpegUtils.DESCALE(tmp4 + z1 + z3, SLOW_INTEGER_CONST_BITS - SLOW_INTEGER_PASS1_BITS);
                data[dataIndex + 5] = JpegUtils.DESCALE(tmp5 + z2 + z4, SLOW_INTEGER_CONST_BITS - SLOW_INTEGER_PASS1_BITS);
                data[dataIndex + 3] = JpegUtils.DESCALE(tmp6 + z2 + z3, SLOW_INTEGER_CONST_BITS - SLOW_INTEGER_PASS1_BITS);
                data[dataIndex + 1] = JpegUtils.DESCALE(tmp7 + z1 + z4, SLOW_INTEGER_CONST_BITS - SLOW_INTEGER_PASS1_BITS);

                dataIndex += JpegConstants.DCTSize;     /* advance pointer to next row */
            }

            /* Pass 2: process columns.
            * We remove the SLOW_INTEGER_PASS1_BITS scaling, but leave the results scaled up
            * by an overall factor of 8.
            */

            dataIndex = 0;
            for (int ctr = JpegConstants.DCTSize - 1; ctr >= 0; ctr--)
            {
                int tmp0 = data[dataIndex + JpegConstants.DCTSize * 0] + data[dataIndex + JpegConstants.DCTSize * 7];
                int tmp7 = data[dataIndex + JpegConstants.DCTSize * 0] - data[dataIndex + JpegConstants.DCTSize * 7];
                int tmp1 = data[dataIndex + JpegConstants.DCTSize * 1] + data[dataIndex + JpegConstants.DCTSize * 6];
                int tmp6 = data[dataIndex + JpegConstants.DCTSize * 1] - data[dataIndex + JpegConstants.DCTSize * 6];
                int tmp2 = data[dataIndex + JpegConstants.DCTSize * 2] + data[dataIndex + JpegConstants.DCTSize * 5];
                int tmp5 = data[dataIndex + JpegConstants.DCTSize * 2] - data[dataIndex + JpegConstants.DCTSize * 5];
                int tmp3 = data[dataIndex + JpegConstants.DCTSize * 3] + data[dataIndex + JpegConstants.DCTSize * 4];
                int tmp4 = data[dataIndex + JpegConstants.DCTSize * 3] - data[dataIndex + JpegConstants.DCTSize * 4];

                /* Even part per LL&M figure 1 --- note that published figure is faulty;
                * rotator "sqrt(2)*c1" should be "sqrt(2)*c6".
                */

                int tmp10 = tmp0 + tmp3;
                int tmp13 = tmp0 - tmp3;
                int tmp11 = tmp1 + tmp2;
                int tmp12 = tmp1 - tmp2;

                data[dataIndex + JpegConstants.DCTSize * 0] = JpegUtils.DESCALE(tmp10 + tmp11, SLOW_INTEGER_PASS1_BITS);
                data[dataIndex + JpegConstants.DCTSize * 4] = JpegUtils.DESCALE(tmp10 - tmp11, SLOW_INTEGER_PASS1_BITS);

                int z1 = (tmp12 + tmp13) * SLOW_INTEGER_FIX_0_541196100;
                data[dataIndex + JpegConstants.DCTSize * 2] = JpegUtils.DESCALE(z1 + tmp13 * SLOW_INTEGER_FIX_0_765366865,
                                                          SLOW_INTEGER_CONST_BITS + SLOW_INTEGER_PASS1_BITS);
                data[dataIndex + JpegConstants.DCTSize * 6] = JpegUtils.DESCALE(z1 + tmp12 * (-SLOW_INTEGER_FIX_1_847759065),
                                                          SLOW_INTEGER_CONST_BITS + SLOW_INTEGER_PASS1_BITS);

                /* Odd part per figure 8 --- note paper omits factor of sqrt(2).
                * cK represents cos(K*pi/16).
                * i0..i3 in the paper are tmp4..tmp7 here.
                */

                z1 = tmp4 + tmp7;
                int z2 = tmp5 + tmp6;
                int z3 = tmp4 + tmp6;
                int z4 = tmp5 + tmp7;
                int z5 = (z3 + z4) * SLOW_INTEGER_FIX_1_175875602; /* sqrt(2) * c3 */

                tmp4 = tmp4 * SLOW_INTEGER_FIX_0_298631336; /* sqrt(2) * (-c1+c3+c5-c7) */
                tmp5 = tmp5 * SLOW_INTEGER_FIX_2_053119869; /* sqrt(2) * ( c1+c3-c5+c7) */
                tmp6 = tmp6 * SLOW_INTEGER_FIX_3_072711026; /* sqrt(2) * ( c1+c3+c5-c7) */
                tmp7 = tmp7 * SLOW_INTEGER_FIX_1_501321110; /* sqrt(2) * ( c1+c3-c5-c7) */
                z1 = z1 * (-SLOW_INTEGER_FIX_0_899976223); /* sqrt(2) * (c7-c3) */
                z2 = z2 * (-SLOW_INTEGER_FIX_2_562915447); /* sqrt(2) * (-c1-c3) */
                z3 = z3 * (-SLOW_INTEGER_FIX_1_961570560); /* sqrt(2) * (-c3-c5) */
                z4 = z4 * (-SLOW_INTEGER_FIX_0_390180644); /* sqrt(2) * (c5-c3) */

                z3 += z5;
                z4 += z5;

                data[dataIndex + JpegConstants.DCTSize * 7] = JpegUtils.DESCALE(tmp4 + z1 + z3, SLOW_INTEGER_CONST_BITS + SLOW_INTEGER_PASS1_BITS);
                data[dataIndex + JpegConstants.DCTSize * 5] = JpegUtils.DESCALE(tmp5 + z2 + z4, SLOW_INTEGER_CONST_BITS + SLOW_INTEGER_PASS1_BITS);
                data[dataIndex + JpegConstants.DCTSize * 3] = JpegUtils.DESCALE(tmp6 + z2 + z3, SLOW_INTEGER_CONST_BITS + SLOW_INTEGER_PASS1_BITS);
                data[dataIndex + JpegConstants.DCTSize * 1] = JpegUtils.DESCALE(tmp7 + z1 + z4, SLOW_INTEGER_CONST_BITS + SLOW_INTEGER_PASS1_BITS);

                dataIndex++;          /* advance pointer to next column */
            }
        }

        /// <summary>
        /// Multiply a DCTELEM variable by an int constant, and immediately
        /// descale to yield a DCTELEM result.
        /// </summary>
        private static int FAST_INTEGER_MULTIPLY(int var, int c)
        {
            return (JpegUtils.DESCALE((var) * (c), FAST_INTEGER_CONST_BITS));
        }
    }
    #endregion

    #region JpegHuffmanTable
    /// <summary>
    /// Huffman coding table.
    /// </summary>
    public class JpegHuffmanTable
    {
        /* These two fields directly represent the contents of a JPEG DHT marker */
        private readonly byte[] m_bits = new byte[17];     /* bits[k] = # of symbols with codes of */

        /* length k bits; bits[0] is unused */
        private readonly byte[] m_huffval = new byte[256];     /* The symbols, in order of incr code length */

        private bool m_sent_table;        /* true when table has been output */


        internal JpegHuffmanTable()
        {
        }

        internal byte[] Bits
        {
            get { return m_bits; }
        }

        internal byte[] Huffval
        {
            get { return m_huffval; }
        }

        /// <summary>
        /// Gets or sets a value indicating whether the table has been output to file.
        /// </summary>
        /// <value>It's initialized <c>false</c> when the table is created, and set 
        /// <c>true</c> when it's been output to the file. You could suppress output 
        /// of a table by setting this to <c>true</c>.
        /// </value>
        /// <remarks>This property is used only during compression. It's initialized
        /// <c>false</c> when the table is created, and set <c>true</c> when it's been
        /// output to the file. You could suppress output of a table by setting this to
        /// <c>true</c>. (See jpeg_suppress_tables for an example.)</remarks>
        /// <seealso cref="JpegCompressor.jpeg_suppress_tables"/>
        public bool Sent_table
        {
            get { return m_sent_table; }
            set { m_sent_table = value; }
        }
    }
    #endregion

    #region JpegInputController
    /// <summary>
    /// Input control module
    /// </summary>
    class JpegInputController
    {
        private JpegDecompressor m_cinfo;
        private bool m_consumeData;
        private bool m_inheaders;     /* true until first SOS is reached */
        private bool m_has_multiple_scans;    /* True if file has multiple scans */
        private bool m_eoi_reached;       /* True when EOI has been consumed */

        /// <summary>
        /// Initialize the input controller module.
        /// This is called only once, when the decompression object is created.
        /// </summary>
        public JpegInputController(JpegDecompressor cinfo)
        {
            m_cinfo = cinfo;

            /* Initialize state: can't use reset_input_controller since we don't
            * want to try to reset other modules yet.
            */
            m_inheaders = true;
        }

        public ReadResult consume_input()
        {
            if (m_consumeData)
                return m_cinfo.m_coef.consume_data();

            return consume_markers();
        }

        /// <summary>
        /// Reset state to begin a fresh datastream.
        /// </summary>
        public void reset_input_controller()
        {
            m_consumeData = false;
            m_has_multiple_scans = false; /* "unknown" would be better */
            m_eoi_reached = false;
            m_inheaders = true;

            /* Reset other modules */
            m_cinfo.m_marker.reset_marker_reader();

            /* Reset progression state -- would be cleaner if entropy decoder did this */
            m_cinfo.m_coef_bits = null;
        }

        /// <summary>
        /// Initialize the input modules to read a scan of compressed data.
        /// The first call to this is done after initializing
        /// the entire decompressor (during jpeg_start_decompress).
        /// Subsequent calls come from consume_markers, below.
        /// </summary>
        public void start_input_pass()
        {
            per_scan_setup();
            latch_quant_tables();
            m_cinfo.m_entropy.start_pass();
            m_cinfo.m_coef.start_input_pass();
            m_consumeData = true;
        }

        /// <summary>
        /// Finish up after inputting a compressed-data scan.
        /// This is called by the coefficient controller after it's read all
        /// the expected data of the scan.
        /// </summary>
        public void finish_input_pass()
        {
            m_consumeData = false;
        }

        public bool HasMultipleScans()
        {
            return m_has_multiple_scans;
        }

        public bool EOIReached()
        {
            return m_eoi_reached;
        }

        /// <summary>
        /// Read JPEG markers before, between, or after compressed-data scans.
        /// Change state as necessary when a new scan is reached.
        /// Return value is JPEG_SUSPENDED, JPEG_REACHED_SOS, or JPEG_REACHED_EOI.
        /// 
        /// The consume_input method pointer points either here or to the
        /// coefficient controller's consume_data routine, depending on whether
        /// we are reading a compressed data segment or inter-segment markers.
        /// </summary>
        private ReadResult consume_markers()
        {
            ReadResult val;

            if (m_eoi_reached) /* After hitting EOI, read no further */
                return ReadResult.Reached_EOI;

            val = m_cinfo.m_marker.read_markers();

            switch (val)
            {
                case ReadResult.Reached_SOS:
                    /* Found SOS */
                    if (m_inheaders)
                    {
                        /* 1st SOS */
                        initial_setup();
                        m_inheaders = false;
                        /* Note: start_input_pass must be called by JpegDecompressorMaster
                         * before any more input can be consumed.
                         */
                    }
                    else
                    {
                        /* 2nd or later SOS marker */
                        if (!m_has_multiple_scans)
                        {
                            /* Oops, I wasn't expecting this! */
                            throw new Exception("Didn't expect more than one scan");
                        }

                        m_cinfo.m_inputctl.start_input_pass();
                    }
                    break;
                case ReadResult.Reached_EOI:
                    /* Found EOI */
                    m_eoi_reached = true;
                    if (m_inheaders)
                    {
                        /* Tables-only data-stream, apparently */
                        if (m_cinfo.m_marker.SawSOF())
                            throw new Exception("Invalid JPEG file structure: missing SOS marker");
                    }
                    else
                    {
                        /* Prevent infinite loop in coef ctlr's decompress_data routine
                         * if user set output_scan_number larger than number of scans.
                         */
                        if (m_cinfo.m_output_scan_number > m_cinfo.m_input_scan_number)
                            m_cinfo.m_output_scan_number = m_cinfo.m_input_scan_number;
                    }
                    break;
                case ReadResult.Suspended:
                    break;
            }

            return val;
        }

        /// <summary>
        /// Routines to calculate various quantities related to the size of the image.
        /// Called once, when first SOS marker is reached
        /// </summary>
        private void initial_setup()
        {
            /* Make sure image isn't bigger than I can handle */
            if (m_cinfo.m_image_height > JpegConstants.JpegMaxDimention ||
                m_cinfo.m_image_width > JpegConstants.JpegMaxDimention)
            {
                throw new Exception(String.Format("Maximum supported image dimension is {0} pixels", (int)JpegConstants.JpegMaxDimention));

            }

            /* For now, precision must match compiled-in value... */
            if (m_cinfo.m_data_precision != JpegConstants.BitsInSample)
                throw new Exception(String.Format("Unsupported JPEG data precision {0}", m_cinfo.m_data_precision));

            /* Check that number of components won't exceed internal array sizes */
            if (m_cinfo.m_num_components > JpegConstants.MaxComponents)
                throw new Exception(String.Format("Too many color components: {0}, max {1}", m_cinfo.m_num_components, JpegConstants.MaxComponents));

            /* Compute maximum sampling factors; check factor validity */
            m_cinfo.m_max_h_samp_factor = 1;
            m_cinfo.m_max_v_samp_factor = 1;

            for (int ci = 0; ci < m_cinfo.m_num_components; ci++)
            {
                if (m_cinfo.Comp_info[ci].H_samp_factor <= 0 || m_cinfo.Comp_info[ci].H_samp_factor > JpegConstants.MaxSamplingFactor ||
                    m_cinfo.Comp_info[ci].V_samp_factor <= 0 || m_cinfo.Comp_info[ci].V_samp_factor > JpegConstants.MaxSamplingFactor)
                {
                    throw new Exception("Bogus sampling factors");
                }

                m_cinfo.m_max_h_samp_factor = Math.Max(m_cinfo.m_max_h_samp_factor, m_cinfo.Comp_info[ci].H_samp_factor);
                m_cinfo.m_max_v_samp_factor = Math.Max(m_cinfo.m_max_v_samp_factor, m_cinfo.Comp_info[ci].V_samp_factor);
            }

            /* We initialize DCT_scaled_size and min_DCT_scaled_size to DCTSize.
             * In the full decompressor, this will be overridden JpegDecompressorMaster;
             * but in the transcoder, JpegDecompressorMaster is not used, so we must do it here.
             */
            m_cinfo.m_min_DCT_scaled_size = JpegConstants.DCTSize;

            /* Compute dimensions of components */
            for (int ci = 0; ci < m_cinfo.m_num_components; ci++)
            {
                m_cinfo.Comp_info[ci].DCT_scaled_size = JpegConstants.DCTSize;

                /* Size in DCT blocks */
                m_cinfo.Comp_info[ci].Width_in_blocks = JpegUtils.jdiv_round_up(
                    m_cinfo.m_image_width * m_cinfo.Comp_info[ci].H_samp_factor,
                    m_cinfo.m_max_h_samp_factor * JpegConstants.DCTSize);

                m_cinfo.Comp_info[ci].height_in_blocks = JpegUtils.jdiv_round_up(
                    m_cinfo.m_image_height * m_cinfo.Comp_info[ci].V_samp_factor,
                    m_cinfo.m_max_v_samp_factor * JpegConstants.DCTSize);

                /* downsampled_width and downsampled_height will also be overridden by
                 * JpegDecompressorMaster if we are doing full decompression.  The transcoder library
                 * doesn't use these values, but the calling application might.
                 */
                /* Size in samples */
                m_cinfo.Comp_info[ci].downsampled_width = JpegUtils.jdiv_round_up(
                    m_cinfo.m_image_width * m_cinfo.Comp_info[ci].H_samp_factor,
                    m_cinfo.m_max_h_samp_factor);

                m_cinfo.Comp_info[ci].downsampled_height = JpegUtils.jdiv_round_up(
                    m_cinfo.m_image_height * m_cinfo.Comp_info[ci].V_samp_factor,
                    m_cinfo.m_max_v_samp_factor);

                /* Mark component needed, until color conversion says otherwise */
                m_cinfo.Comp_info[ci].component_needed = true;

                /* Mark no quantization table yet saved for component */
                m_cinfo.Comp_info[ci].quant_table = null;
            }

            /* Compute number of fully interleaved MCU rows. */
            m_cinfo.m_total_iMCU_rows = JpegUtils.jdiv_round_up(
                m_cinfo.m_image_height, m_cinfo.m_max_v_samp_factor * JpegConstants.DCTSize);

            /* Decide whether file contains multiple scans */
            if (m_cinfo.m_comps_in_scan < m_cinfo.m_num_components || m_cinfo.m_progressive_mode)
                m_cinfo.m_inputctl.m_has_multiple_scans = true;
            else
                m_cinfo.m_inputctl.m_has_multiple_scans = false;
        }

        /// <summary>
        /// Save away a copy of the Q-table referenced by each component present
        /// in the current scan, unless already saved during a prior scan.
        /// 
        /// In a multiple-scan JPEG file, the encoder could assign different components
        /// the same Q-table slot number, but change table definitions between scans
        /// so that each component uses a different Q-table.  (The IJG encoder is not
        /// currently capable of doing this, but other encoders might.)  Since we want
        /// to be able to de-quantize all the components at the end of the file, this
        /// means that we have to save away the table actually used for each component.
        /// We do this by copying the table at the start of the first scan containing
        /// the component.
        /// The JPEG spec prohibits the encoder from changing the contents of a Q-table
        /// slot between scans of a component using that slot.  If the encoder does so
        /// anyway, this decoder will simply use the Q-table values that were current
        /// at the start of the first scan for the component.
        /// 
        /// The decompressor output side looks only at the saved quant tables,
        /// not at the current Q-table slots.
        /// </summary>
        private void latch_quant_tables()
        {
            for (int ci = 0; ci < m_cinfo.m_comps_in_scan; ci++)
            {
                JpegComponent componentInfo = m_cinfo.Comp_info[m_cinfo.m_cur_comp_info[ci]];

                /* No work if we already saved Q-table for this component */
                if (componentInfo.quant_table != null)
                    continue;

                /* Make sure specified quantization table is present */
                int qtblno = componentInfo.Quant_tbl_no;
                if (qtblno < 0 || qtblno >= JpegConstants.NumberOfQuantTables || m_cinfo.m_quant_tbl_ptrs[qtblno] == null)
                    throw new Exception(String.Format("Quantization table 0x{0:X2} was not defined", qtblno));

                /* OK, save away the quantization table */
                JpegQuantizationTable qtbl = new JpegQuantizationTable();
                Buffer.BlockCopy(m_cinfo.m_quant_tbl_ptrs[qtblno].quantval, 0,
                    qtbl.quantval, 0, qtbl.quantval.Length * sizeof(short));
                qtbl.Sent_table = m_cinfo.m_quant_tbl_ptrs[qtblno].Sent_table;
                componentInfo.quant_table = qtbl;
                m_cinfo.Comp_info[m_cinfo.m_cur_comp_info[ci]] = componentInfo;
            }
        }

        /// <summary>
        /// Do computations that are needed before processing a JPEG scan
        /// cinfo.comps_in_scan and cinfo.cur_comp_info[] were set from SOS marker
        /// </summary>
        private void per_scan_setup()
        {
            if (m_cinfo.m_comps_in_scan == 1)
            {
                /* Non-interleaved (single-component) scan */
                JpegComponent componentInfo = m_cinfo.Comp_info[m_cinfo.m_cur_comp_info[0]];

                /* Overall image size in MCUs */
                m_cinfo.m_MCUs_per_row = componentInfo.Width_in_blocks;
                m_cinfo.m_MCU_rows_in_scan = componentInfo.height_in_blocks;

                /* For non-interleaved scan, always one block per MCU */
                componentInfo.MCU_width = 1;
                componentInfo.MCU_height = 1;
                componentInfo.MCU_blocks = 1;
                componentInfo.MCU_sample_width = componentInfo.DCT_scaled_size;
                componentInfo.last_col_width = 1;

                /* For non-interleaved scans, it is convenient to define last_row_height
                 * as the number of block rows present in the last iMCU row.
                 */
                int tmp = componentInfo.height_in_blocks % componentInfo.V_samp_factor;
                if (tmp == 0)
                    tmp = componentInfo.V_samp_factor;
                componentInfo.last_row_height = tmp;
                m_cinfo.Comp_info[m_cinfo.m_cur_comp_info[0]] = componentInfo;

                /* Prepare array describing MCU composition */
                m_cinfo.m_blocks_in_MCU = 1;
                m_cinfo.m_MCU_membership[0] = 0;
            }
            else
            {
                /* Interleaved (multi-component) scan */
                if (m_cinfo.m_comps_in_scan <= 0 || m_cinfo.m_comps_in_scan > JpegConstants.MaxComponentsInScan)
                    throw new Exception(String.Format("Too many color components: {0}, max {1}", m_cinfo.m_comps_in_scan, JpegConstants.MaxComponentsInScan));

                /* Overall image size in MCUs */
                m_cinfo.m_MCUs_per_row = JpegUtils.jdiv_round_up(
                    m_cinfo.m_image_width, m_cinfo.m_max_h_samp_factor * JpegConstants.DCTSize);

                m_cinfo.m_MCU_rows_in_scan = JpegUtils.jdiv_round_up(
                    m_cinfo.m_image_height, m_cinfo.m_max_v_samp_factor * JpegConstants.DCTSize);

                m_cinfo.m_blocks_in_MCU = 0;

                for (int ci = 0; ci < m_cinfo.m_comps_in_scan; ci++)
                {
                    JpegComponent componentInfo = m_cinfo.Comp_info[m_cinfo.m_cur_comp_info[ci]];

                    /* Sampling factors give # of blocks of component in each MCU */
                    componentInfo.MCU_width = componentInfo.H_samp_factor;
                    componentInfo.MCU_height = componentInfo.V_samp_factor;
                    componentInfo.MCU_blocks = componentInfo.MCU_width * componentInfo.MCU_height;
                    componentInfo.MCU_sample_width = componentInfo.MCU_width * componentInfo.DCT_scaled_size;

                    /* Figure number of non-dummy blocks in last MCU column & row */
                    int tmp = componentInfo.Width_in_blocks % componentInfo.MCU_width;
                    if (tmp == 0)
                        tmp = componentInfo.MCU_width;
                    componentInfo.last_col_width = tmp;

                    tmp = componentInfo.height_in_blocks % componentInfo.MCU_height;
                    if (tmp == 0)
                        tmp = componentInfo.MCU_height;
                    componentInfo.last_row_height = tmp;

                    /* Prepare array describing MCU composition */
                    int mcublks = componentInfo.MCU_blocks;
                    if (m_cinfo.m_blocks_in_MCU + mcublks > JpegConstants.DecompressorMaxBlocksInMCU)
                        throw new Exception("Sampling factors too large for interleaved scan");

                    m_cinfo.Comp_info[m_cinfo.m_cur_comp_info[ci]] = componentInfo;

                    while (mcublks-- > 0)
                        m_cinfo.m_MCU_membership[m_cinfo.m_blocks_in_MCU++] = ci;
                }
            }
        }
    }
    #endregion

    #region JpegInverseDCT
    /// <summary>
    /// An inverse DCT routine is given a pointer to the input JpegBlock and a pointer
    /// to an output sample array.  The routine must dequantize the input data as
    /// well as perform the IDCT; for dequantization, it uses the multiplier table
    /// pointed to by componentInfo.dct_table.  The output data is to be placed into the
    /// sample array starting at a specified column. (Any row offset needed will
    /// be applied to the array pointer before it is passed to the IDCT code)
    /// Note that the number of samples emitted by the IDCT routine is
    /// DCT_scaled_size * DCT_scaled_size.
    /// 
    /// Each IDCT routine has its own ideas about the best dct_table element type.
    /// 
    /// The decompressor input side saves away the appropriate
    /// quantization table for each component at the start of the first scan
    /// involving that component.  (This is necessary in order to correctly
    /// decode files that reuse Q-table slots.)
    /// When we are ready to make an output pass, the saved Q-table is converted
    /// to a multiplier table that will actually be used by the IDCT routine.
    /// The multiplier table contents are IDCT-method-dependent.  To support
    /// application changes in IDCT method between scans, we can remake the
    /// multiplier tables if necessary.
    /// In buffered-image mode, the first output pass may occur before any data
    /// has been seen for some components, and thus before their Q-tables have
    /// been saved away.  To handle this case, multiplier tables are preset
    /// to zeroes; the result of the IDCT will be a neutral gray level.
    /// </summary>
    class JpegInverseDCT
    {
        private const int IFAST_SCALE_BITS = 2; /* fractional bits in scale factors */

        /*
        * Each IDCT routine is responsible for range-limiting its results and
        * converting them to unsigned form (0..MaxSampleValue).  The raw outputs could
        * be quite far out of range if the input data is corrupt, so a bulletproof
        * range-limiting step is required.  We use a mask-and-table-lookup method
        * to do the combined operations quickly.  See the comments with
        * prepare_range_limit_table (in jdmaster.c) for more info.
        */
        private const int RANGE_MASK = (JpegConstants.MaxSampleValue * 4 + 3); /* 2 bits wider than legal samples */

        private const int SLOW_INTEGER_CONST_BITS = 13;
        private const int SLOW_INTEGER_PASS1_BITS = 2;

        /* We use the following pre-calculated constants.
        * If you change SLOW_INTEGER_CONST_BITS you may want to add appropriate values.
        * 
        * Convert a positive real constant to an integer scaled by CONST_SCALE.
        * static int SLOW_INTEGER_FIX(double x)
        * {
        *  return ((int) ((x) * (((int) 1) << SLOW_INTEGER_CONST_BITS) + 0.5));
        * }
        */

        private const int SLOW_INTEGER_FIX_0_298631336 = 2446;   /* SLOW_INTEGER_FIX(0.298631336) */
        private const int SLOW_INTEGER_FIX_0_390180644 = 3196;   /* SLOW_INTEGER_FIX(0.390180644) */
        private const int SLOW_INTEGER_FIX_0_541196100 = 4433;   /* SLOW_INTEGER_FIX(0.541196100) */
        private const int SLOW_INTEGER_FIX_0_765366865 = 6270;   /* SLOW_INTEGER_FIX(0.765366865) */
        private const int SLOW_INTEGER_FIX_0_899976223 = 7373;   /* SLOW_INTEGER_FIX(0.899976223) */
        private const int SLOW_INTEGER_FIX_1_175875602 = 9633;   /* SLOW_INTEGER_FIX(1.175875602) */
        private const int SLOW_INTEGER_FIX_1_501321110 = 12299;  /* SLOW_INTEGER_FIX(1.501321110) */
        private const int SLOW_INTEGER_FIX_1_847759065 = 15137;  /* SLOW_INTEGER_FIX(1.847759065) */
        private const int SLOW_INTEGER_FIX_1_961570560 = 16069;  /* SLOW_INTEGER_FIX(1.961570560) */
        private const int SLOW_INTEGER_FIX_2_053119869 = 16819;  /* SLOW_INTEGER_FIX(2.053119869) */
        private const int SLOW_INTEGER_FIX_2_562915447 = 20995;  /* SLOW_INTEGER_FIX(2.562915447) */
        private const int SLOW_INTEGER_FIX_3_072711026 = 25172;  /* SLOW_INTEGER_FIX(3.072711026) */

        private const int FAST_INTEGER_CONST_BITS = 8;
        private const int FAST_INTEGER_PASS1_BITS = 2;

        /* We use the following pre-calculated constants.
        * If you change FAST_INTEGER_CONST_BITS you may want to add appropriate values.
        */
        private const int FAST_INTEGER_FIX_1_082392200 = 277;        /* FAST_INTEGER_FIX(1.082392200) */
        private const int FAST_INTEGER_FIX_1_414213562 = 362;        /* FAST_INTEGER_FIX(1.414213562) */
        private const int FAST_INTEGER_FIX_1_847759065 = 473;        /* FAST_INTEGER_FIX(1.847759065) */
        private const int FAST_INTEGER_FIX_2_613125930 = 669;        /* FAST_INTEGER_FIX(2.613125930) */

        private const int REDUCED_CONST_BITS = 13;
        private const int REDUCED_PASS1_BITS = 2;

        /* We use the following pre-calculated constants.
        * If you change REDUCED_CONST_BITS you may want to add appropriate values.
        * Convert a positive real constant to an integer scaled by CONST_SCALE.
        * static int REDUCED_FIX(double x)
        * {
        *   return ((int) ((x) * (((int) 1) << REDUCED_CONST_BITS) + 0.5));
        * }
        */

        private const int REDUCED_FIX_0_211164243 = 1730;    /* REDUCED_FIX(0.211164243) */
        private const int REDUCED_FIX_0_509795579 = 4176;    /* REDUCED_FIX(0.509795579) */
        private const int REDUCED_FIX_0_601344887 = 4926;    /* REDUCED_FIX(0.601344887) */
        private const int REDUCED_FIX_0_720959822 = 5906;    /* REDUCED_FIX(0.720959822) */
        private const int REDUCED_FIX_0_765366865 = 6270;    /* REDUCED_FIX(0.765366865) */
        private const int REDUCED_FIX_0_850430095 = 6967;    /* REDUCED_FIX(0.850430095) */
        private const int REDUCED_FIX_0_899976223 = 7373;    /* REDUCED_FIX(0.899976223) */
        private const int REDUCED_FIX_1_061594337 = 8697;    /* REDUCED_FIX(1.061594337) */
        private const int REDUCED_FIX_1_272758580 = 10426;   /* REDUCED_FIX(1.272758580) */
        private const int REDUCED_FIX_1_451774981 = 11893;   /* REDUCED_FIX(1.451774981) */
        private const int REDUCED_FIX_1_847759065 = 15137;   /* REDUCED_FIX(1.847759065) */
        private const int REDUCED_FIX_2_172734803 = 17799;   /* REDUCED_FIX(2.172734803) */
        private const int REDUCED_FIX_2_562915447 = 20995;   /* REDUCED_FIX(2.562915447) */
        private const int REDUCED_FIX_3_624509785 = 29692;   /* REDUCED_FIX(3.624509785) */

        /* precomputed values scaled up by 14 bits */
        private static short[] aanscales = 
        {
            16384, 22725, 21407, 19266, 16384, 12873, 8867, 4520, 22725, 31521, 29692, 26722, 22725, 17855,
            12299, 6270, 21407, 29692, 27969, 25172, 21407, 16819, 11585,
            5906, 19266, 26722, 25172, 22654, 19266, 15137, 10426, 5315,
            16384, 22725, 21407, 19266, 16384, 12873, 8867, 4520, 12873,
            17855, 16819, 15137, 12873, 10114, 6967, 3552, 8867, 12299,
            11585, 10426, 8867, 6967, 4799, 2446, 4520, 6270, 5906, 5315,
            4520, 3552, 2446, 1247 
        };

        private const int CONST_BITS = 14;

        private static double[] aanscalefactor = 
        { 
            1.0, 1.387039845, 1.306562965, 1.175875602, 1.0,
            0.785694958, 0.541196100, 0.275899379 
        };

        private enum InverseMethod
        {
            Unknown,
            idct_1x1_method,
            idct_2x2_method,
            idct_4x4_method,
            idct_islow_method,
            idct_ifast_method,
            idct_float_method
        }

        /* It is useful to allow each component to have a separate IDCT method. */
        private InverseMethod[] m_inverse_DCT_method = new InverseMethod[JpegConstants.MaxComponents];

        /* Allocated multiplier tables: big enough for any supported variant */
        private class multiplier_table
        {
            public int[] int_array = new int[JpegConstants.DCTSize2];
            public float[] float_array = new float[JpegConstants.DCTSize2];
        };

        private multiplier_table[] m_dctTables;

        private JpegDecompressor m_cinfo;

        /* This array contains the IDCT method code that each multiplier table
        * is currently set up for, or -1 if it's not yet set up.
        * The actual multiplier tables are pointed to by dct_table in the
        * per-component comp_info structures.
        */
        private int[] m_cur_method = new int[JpegConstants.MaxComponents];

        private ComponentBuffer m_componentBuffer;

        public JpegInverseDCT(JpegDecompressor cinfo)
        {
            m_cinfo = cinfo;

            m_dctTables = new multiplier_table[cinfo.m_num_components];
            for (int ci = 0; ci < cinfo.m_num_components; ci++)
            {
                /* Allocate and pre-zero a multiplier table for each component */
                m_dctTables[ci] = new multiplier_table();

                /* Mark multiplier table not yet set up for any method */
                m_cur_method[ci] = -1;
            }
        }

        /// <summary>
        /// Prepare for an output pass.
        /// Here we select the proper IDCT routine for each component and build
        /// a matching multiplier table.
        /// </summary>
        public void start_pass()
        {
            for (int ci = 0; ci < m_cinfo.m_num_components; ci++)
            {
                JpegComponent componentInfo = m_cinfo.Comp_info[ci];

                InverseMethod im = InverseMethod.Unknown;
                int method = 0;
                /* Select the proper IDCT routine for this component's scaling */
                switch (componentInfo.DCT_scaled_size)
                {
                    case 1:
                        im = InverseMethod.idct_1x1_method;
                        method = (int)DCTMethod.IntSlow;    /* jidctred uses islow-style table */
                        break;
                    case 2:
                        im = InverseMethod.idct_2x2_method;
                        method = (int)DCTMethod.IntSlow;    /* jidctred uses islow-style table */
                        break;
                    case 4:
                        im = InverseMethod.idct_4x4_method;
                        method = (int)DCTMethod.IntSlow;    /* jidctred uses islow-style table */
                        break;
                    case JpegConstants.DCTSize:
                        switch (m_cinfo.m_dct_method)
                        {
                            case DCTMethod.IntSlow:
                                im = InverseMethod.idct_islow_method;
                                method = (int)DCTMethod.IntSlow;
                                break;
                            case DCTMethod.IntFast:
                                im = InverseMethod.idct_ifast_method;
                                method = (int)DCTMethod.IntFast;
                                break;
                            case DCTMethod.Float:
                                im = InverseMethod.idct_float_method;
                                method = (int)DCTMethod.Float;
                                break;
                            default:
                                throw new Exception("Unknown DCT Method!");
                        }
                        break;
                    default:
                        throw new Exception(String.Format("IDCT output block size {0} not supported", componentInfo.DCT_scaled_size));
                }

                m_inverse_DCT_method[ci] = im;

                /* Create multiplier table from quant table.
                 * However, we can skip this if the component is uninteresting
                 * or if we already built the table.  Also, if no quant table
                 * has yet been saved for the component, we leave the
                 * multiplier table all-zero; we'll be reading zeroes from the
                 * coefficient controller's buffer anyway.
                 */
                if (!componentInfo.component_needed || m_cur_method[ci] == method)
                    continue;

                if (componentInfo.quant_table == null)
                {
                    /* happens if no data yet for component */
                    continue;
                }

                m_cur_method[ci] = method;
                switch ((DCTMethod)method)
                {
                    case DCTMethod.IntSlow:
                        /* For LL&M IDCT method, multipliers are equal to raw quantization
                         * coefficients, but are stored as ints to ensure access efficiency.
                         */
                        int[] ismtbl = m_dctTables[ci].int_array;
                        for (int i = 0; i < JpegConstants.DCTSize2; i++)
                            ismtbl[i] = componentInfo.quant_table.quantval[i];
                        break;

                    case DCTMethod.IntFast:
                        /* For AA&N IDCT method, multipliers are equal to quantization
                         * coefficients scaled by scalefactor[row]*scalefactor[col], where
                         *   scalefactor[0] = 1
                         *   scalefactor[k] = cos(k*PI/16) * sqrt(2)    for k=1..7
                         * For integer operation, the multiplier table is to be scaled by
                         * IFAST_SCALE_BITS.
                         */
                        int[] ifmtbl = m_dctTables[ci].int_array;

                        for (int i = 0; i < JpegConstants.DCTSize2; i++)
                        {
                            ifmtbl[i] = JpegUtils.DESCALE((int)componentInfo.quant_table.quantval[i] * (int)aanscales[i], CONST_BITS - IFAST_SCALE_BITS);
                        }
                        break;

                    case DCTMethod.Float:
                        /* For float AA&N IDCT method, multipliers are equal to quantization
                         * coefficients scaled by scalefactor[row]*scalefactor[col], where
                         *   scalefactor[0] = 1
                         *   scalefactor[k] = cos(k*PI/16) * sqrt(2)    for k=1..7
                         */
                        float[] fmtbl = m_dctTables[ci].float_array;
                        int ii = 0;
                        for (int row = 0; row < JpegConstants.DCTSize; row++)
                        {
                            for (int col = 0; col < JpegConstants.DCTSize; col++)
                            {
                                fmtbl[ii] = (float)((double)componentInfo.quant_table.quantval[ii] * aanscalefactor[row] * aanscalefactor[col]);
                                ii++;
                            }
                        }
                        break;

                    default:
                        throw new Exception("Unknown DCT Method!");
                }
            }
        }

        /* Inverse DCT (also performs de-quantization) */
        public void inverse(int component_index, short[] coef_block, ComponentBuffer output_buf, int output_row, int output_col)
        {
            m_componentBuffer = output_buf;
            switch (m_inverse_DCT_method[component_index])
            {
                case InverseMethod.idct_1x1_method:
                    jpeg_idct_1x1(component_index, coef_block, output_row, output_col);
                    break;
                case InverseMethod.idct_2x2_method:
                    jpeg_idct_2x2(component_index, coef_block, output_row, output_col);
                    break;
                case InverseMethod.idct_4x4_method:
                    jpeg_idct_4x4(component_index, coef_block, output_row, output_col);
                    break;
                case InverseMethod.idct_islow_method:
                    jpeg_idct_islow(component_index, coef_block, output_row, output_col);
                    break;
                case InverseMethod.idct_ifast_method:
                    jpeg_idct_ifast(component_index, coef_block, output_row, output_col);
                    break;
                case InverseMethod.idct_float_method:
                    jpeg_idct_float(component_index, coef_block, output_row, output_col);
                    break;
                case InverseMethod.Unknown:
                default:
                    throw new Exception("Unknown Inverse Method!");
            }
        }

        /// <summary>
        /// Perform de-quantization and inverse DCT on one block of coefficients.
        /// NOTE: this code only copes with 8x8 DCTs.
        /// A slow-but-accurate integer implementation of the
        /// inverse DCT (Discrete Cosine Transform).  In the IJG code, this routine
        /// must also perform de-quantization of the input coefficients.
        /// 
        /// A 2-D IDCT can be done by 1-D IDCT on each column followed by 1-D IDCT
        /// on each row (or vice versa, but it's more convenient to emit a row at
        /// a time).  Direct algorithms are also available, but they are much more
        /// complex and seem not to be any faster when reduced to code.
        /// 
        /// This implementation is based on an algorithm described in
        /// C. Loeffler, A. Ligtenberg and G. Moschytz, "Practical Fast 1-D DCT
        /// Algorithms with 11 Multiplications", Proc. Int'l. Conf. on Acoustics,
        /// Speech, and Signal Processing 1989 (ICASSP '89), pp. 988-991.
        /// The primary algorithm described there uses 11 multiplies and 29 adds.
        /// We use their alternate method with 12 multiplies and 32 adds.
        /// The advantage of this method is that no data path contains more than one
        /// multiplication; this allows a very simple and accurate implementation in
        /// scaled fixed-point arithmetic, with a minimal number of shifts.
        /// 
        /// The poop on this scaling stuff is as follows:
        /// 
        /// Each 1-D IDCT step produces outputs which are a factor of sqrt(N)
        /// larger than the true IDCT outputs.  The final outputs are therefore
        /// a factor of N larger than desired; since N=8 this can be cured by
        /// a simple right shift at the end of the algorithm.  The advantage of
        /// this arrangement is that we save two multiplications per 1-D IDCT,
        /// because the y0 and y4 inputs need not be divided by sqrt(N).
        /// 
        /// We have to do addition and subtraction of the integer inputs, which
        /// is no problem, and multiplication by fractional constants, which is
        /// a problem to do in integer arithmetic.  We multiply all the constants
        /// by CONST_SCALE and convert them to integer constants (thus retaining
        /// SLOW_INTEGER_CONST_BITS bits of precision in the constants).  After doing a
        /// multiplication we have to divide the product by CONST_SCALE, with proper
        /// rounding, to produce the correct output.  This division can be done
        /// cheaply as a right shift of SLOW_INTEGER_CONST_BITS bits.  We postpone shifting
        /// as long as possible so that partial sums can be added together with
        /// full fractional precision.
        /// 
        /// The outputs of the first pass are scaled up by SLOW_INTEGER_PASS1_BITS bits so that
        /// they are represented to better-than-integral precision.  These outputs
        /// require BitsInSample + SLOW_INTEGER_PASS1_BITS + 3 bits; this fits in a 16-bit word
        /// with the recommended scaling.  (To scale up 12-bit sample data further, an
        /// intermediate int array would be needed.)
        /// 
        /// To avoid overflow of the 32-bit intermediate results in pass 2, we must
        /// have BitsInSample + SLOW_INTEGER_CONST_BITS + SLOW_INTEGER_PASS1_BITS &lt;= 26.  Error analysis
        /// shows that the values given below are the most effective.
        /// </summary>
        private void jpeg_idct_islow(int component_index, short[] coef_block, int output_row, int output_col)
        {
            /* buffers data between passes */
            int[] workspace = new int[JpegConstants.DCTSize2];

            /* Pass 1: process columns from input, store into work array. */
            /* Note results are scaled up by sqrt(8) compared to a true IDCT; */
            /* furthermore, we scale the results by 2**SLOW_INTEGER_PASS1_BITS. */

            int coefBlockIndex = 0;

            int[] quantTable = m_dctTables[component_index].int_array;
            int quantTableIndex = 0;

            int workspaceIndex = 0;

            for (int ctr = JpegConstants.DCTSize; ctr > 0; ctr--)
            {
                /* Due to quantization, we will usually find that many of the input
                * coefficients are zero, especially the AC terms.  We can exploit this
                * by short-circuiting the IDCT calculation for any column in which all
                * the AC terms are zero.  In that case each output is equal to the
                * DC coefficient (with scale factor as needed).
                * With typical images and quantization tables, half or more of the
                * column DCT calculations can be simplified this way.
                */

                if (coef_block[coefBlockIndex + JpegConstants.DCTSize * 1] == 0 &&
                    coef_block[coefBlockIndex + JpegConstants.DCTSize * 2] == 0 &&
                    coef_block[coefBlockIndex + JpegConstants.DCTSize * 3] == 0 &&
                    coef_block[coefBlockIndex + JpegConstants.DCTSize * 4] == 0 &&
                    coef_block[coefBlockIndex + JpegConstants.DCTSize * 5] == 0 &&
                    coef_block[coefBlockIndex + JpegConstants.DCTSize * 6] == 0 &&
                    coef_block[coefBlockIndex + JpegConstants.DCTSize * 7] == 0)
                {
                    /* AC terms all zero */
                    int dcval = SLOW_INTEGER_DEQUANTIZE(coef_block[coefBlockIndex + JpegConstants.DCTSize * 0],
                        quantTable[quantTableIndex + JpegConstants.DCTSize * 0]) << SLOW_INTEGER_PASS1_BITS;

                    workspace[workspaceIndex + JpegConstants.DCTSize * 0] = dcval;
                    workspace[workspaceIndex + JpegConstants.DCTSize * 1] = dcval;
                    workspace[workspaceIndex + JpegConstants.DCTSize * 2] = dcval;
                    workspace[workspaceIndex + JpegConstants.DCTSize * 3] = dcval;
                    workspace[workspaceIndex + JpegConstants.DCTSize * 4] = dcval;
                    workspace[workspaceIndex + JpegConstants.DCTSize * 5] = dcval;
                    workspace[workspaceIndex + JpegConstants.DCTSize * 6] = dcval;
                    workspace[workspaceIndex + JpegConstants.DCTSize * 7] = dcval;

                    /* advance pointers to next column */
                    coefBlockIndex++;
                    quantTableIndex++;
                    workspaceIndex++;
                    continue;
                }

                /* Even part: reverse the even part of the forward DCT. */
                /* The rotator is sqrt(2)*c(-6). */

                int z2 = SLOW_INTEGER_DEQUANTIZE(coef_block[coefBlockIndex + JpegConstants.DCTSize * 2],
                    quantTable[quantTableIndex + JpegConstants.DCTSize * 2]);
                int z3 = SLOW_INTEGER_DEQUANTIZE(coef_block[coefBlockIndex + JpegConstants.DCTSize * 6],
                    quantTable[quantTableIndex + JpegConstants.DCTSize * 6]);

                int z1 = (z2 + z3) * SLOW_INTEGER_FIX_0_541196100;
                int tmp2 = z1 + z3 * (-SLOW_INTEGER_FIX_1_847759065);
                int tmp3 = z1 + z2 * SLOW_INTEGER_FIX_0_765366865;

                z2 = SLOW_INTEGER_DEQUANTIZE(coef_block[coefBlockIndex + JpegConstants.DCTSize * 0],
                    quantTable[quantTableIndex + JpegConstants.DCTSize * 0]);
                z3 = SLOW_INTEGER_DEQUANTIZE(coef_block[coefBlockIndex + JpegConstants.DCTSize * 4],
                    quantTable[quantTableIndex + JpegConstants.DCTSize * 4]);

                int tmp0 = (z2 + z3) << SLOW_INTEGER_CONST_BITS;
                int tmp1 = (z2 - z3) << SLOW_INTEGER_CONST_BITS;

                int tmp10 = tmp0 + tmp3;
                int tmp13 = tmp0 - tmp3;
                int tmp11 = tmp1 + tmp2;
                int tmp12 = tmp1 - tmp2;

                /* Odd part per figure 8; the matrix is unitary and hence its
                * transpose is its inverse.  i0..i3 are y7,y5,y3,y1 respectively.
                */

                tmp0 = SLOW_INTEGER_DEQUANTIZE(coef_block[coefBlockIndex + JpegConstants.DCTSize * 7],
                    quantTable[quantTableIndex + JpegConstants.DCTSize * 7]);
                tmp1 = SLOW_INTEGER_DEQUANTIZE(coef_block[coefBlockIndex + JpegConstants.DCTSize * 5],
                    quantTable[quantTableIndex + JpegConstants.DCTSize * 5]);
                tmp2 = SLOW_INTEGER_DEQUANTIZE(coef_block[coefBlockIndex + JpegConstants.DCTSize * 3],
                    quantTable[quantTableIndex + JpegConstants.DCTSize * 3]);
                tmp3 = SLOW_INTEGER_DEQUANTIZE(coef_block[coefBlockIndex + JpegConstants.DCTSize * 1],
                    quantTable[quantTableIndex + JpegConstants.DCTSize * 1]);

                z1 = tmp0 + tmp3;
                z2 = tmp1 + tmp2;
                z3 = tmp0 + tmp2;
                int z4 = tmp1 + tmp3;
                int z5 = (z3 + z4) * SLOW_INTEGER_FIX_1_175875602; /* sqrt(2) * c3 */

                tmp0 = tmp0 * SLOW_INTEGER_FIX_0_298631336; /* sqrt(2) * (-c1+c3+c5-c7) */
                tmp1 = tmp1 * SLOW_INTEGER_FIX_2_053119869; /* sqrt(2) * ( c1+c3-c5+c7) */
                tmp2 = tmp2 * SLOW_INTEGER_FIX_3_072711026; /* sqrt(2) * ( c1+c3+c5-c7) */
                tmp3 = tmp3 * SLOW_INTEGER_FIX_1_501321110; /* sqrt(2) * ( c1+c3-c5-c7) */
                z1 = z1 * (-SLOW_INTEGER_FIX_0_899976223); /* sqrt(2) * (c7-c3) */
                z2 = z2 * (-SLOW_INTEGER_FIX_2_562915447); /* sqrt(2) * (-c1-c3) */
                z3 = z3 * (-SLOW_INTEGER_FIX_1_961570560); /* sqrt(2) * (-c3-c5) */
                z4 = z4 * (-SLOW_INTEGER_FIX_0_390180644); /* sqrt(2) * (c5-c3) */

                z3 += z5;
                z4 += z5;

                tmp0 += z1 + z3;
                tmp1 += z2 + z4;
                tmp2 += z2 + z3;
                tmp3 += z1 + z4;

                /* Final output stage: inputs are tmp10..tmp13, tmp0..tmp3 */

                workspace[workspaceIndex + JpegConstants.DCTSize * 0] = JpegUtils.DESCALE(tmp10 + tmp3, SLOW_INTEGER_CONST_BITS - SLOW_INTEGER_PASS1_BITS);
                workspace[workspaceIndex + JpegConstants.DCTSize * 7] = JpegUtils.DESCALE(tmp10 - tmp3, SLOW_INTEGER_CONST_BITS - SLOW_INTEGER_PASS1_BITS);
                workspace[workspaceIndex + JpegConstants.DCTSize * 1] = JpegUtils.DESCALE(tmp11 + tmp2, SLOW_INTEGER_CONST_BITS - SLOW_INTEGER_PASS1_BITS);
                workspace[workspaceIndex + JpegConstants.DCTSize * 6] = JpegUtils.DESCALE(tmp11 - tmp2, SLOW_INTEGER_CONST_BITS - SLOW_INTEGER_PASS1_BITS);
                workspace[workspaceIndex + JpegConstants.DCTSize * 2] = JpegUtils.DESCALE(tmp12 + tmp1, SLOW_INTEGER_CONST_BITS - SLOW_INTEGER_PASS1_BITS);
                workspace[workspaceIndex + JpegConstants.DCTSize * 5] = JpegUtils.DESCALE(tmp12 - tmp1, SLOW_INTEGER_CONST_BITS - SLOW_INTEGER_PASS1_BITS);
                workspace[workspaceIndex + JpegConstants.DCTSize * 3] = JpegUtils.DESCALE(tmp13 + tmp0, SLOW_INTEGER_CONST_BITS - SLOW_INTEGER_PASS1_BITS);
                workspace[workspaceIndex + JpegConstants.DCTSize * 4] = JpegUtils.DESCALE(tmp13 - tmp0, SLOW_INTEGER_CONST_BITS - SLOW_INTEGER_PASS1_BITS);

                /* advance pointers to next column */
                coefBlockIndex++;
                quantTableIndex++;
                workspaceIndex++;
            }

            /* Pass 2: process rows from work array, store into output array. */
            /* Note that we must descale the results by a factor of 8 == 2**3, */
            /* and also undo the SLOW_INTEGER_PASS1_BITS scaling. */

            workspaceIndex = 0;
            byte[] limit = m_cinfo.m_sample_range_limit;
            int limitOffset = m_cinfo.m_sampleRangeLimitOffset + JpegConstants.MediumSampleValue;

            for (int ctr = 0; ctr < JpegConstants.DCTSize; ctr++)
            {
                /* Rows of zeroes can be exploited in the same way as we did with columns.
                * However, the column calculation has created many nonzero AC terms, so
                * the simplification applies less often (typically 5% to 10% of the time).
                * On machines with very fast multiplication, it's possible that the
                * test takes more time than it's worth.  In that case this section
                * may be commented out.
                */
                int currentOutRow = output_row + ctr;
                if (workspace[workspaceIndex + 1] == 0 &&
                    workspace[workspaceIndex + 2] == 0 &&
                    workspace[workspaceIndex + 3] == 0 &&
                    workspace[workspaceIndex + 4] == 0 &&
                    workspace[workspaceIndex + 5] == 0 &&
                    workspace[workspaceIndex + 6] == 0 &&
                    workspace[workspaceIndex + 7] == 0)
                {
                    /* AC terms all zero */
                    byte dcval = limit[limitOffset + JpegUtils.DESCALE(workspace[workspaceIndex + 0], SLOW_INTEGER_PASS1_BITS + 3) & RANGE_MASK];

                    m_componentBuffer[currentOutRow][output_col + 0] = dcval;
                    m_componentBuffer[currentOutRow][output_col + 1] = dcval;
                    m_componentBuffer[currentOutRow][output_col + 2] = dcval;
                    m_componentBuffer[currentOutRow][output_col + 3] = dcval;
                    m_componentBuffer[currentOutRow][output_col + 4] = dcval;
                    m_componentBuffer[currentOutRow][output_col + 5] = dcval;
                    m_componentBuffer[currentOutRow][output_col + 6] = dcval;
                    m_componentBuffer[currentOutRow][output_col + 7] = dcval;

                    workspaceIndex += JpegConstants.DCTSize;       /* advance pointer to next row */
                    continue;
                }

                /* Even part: reverse the even part of the forward DCT. */
                /* The rotator is sqrt(2)*c(-6). */

                int z2 = workspace[workspaceIndex + 2];
                int z3 = workspace[workspaceIndex + 6];

                int z1 = (z2 + z3) * SLOW_INTEGER_FIX_0_541196100;
                int tmp2 = z1 + z3 * (-SLOW_INTEGER_FIX_1_847759065);
                int tmp3 = z1 + z2 * SLOW_INTEGER_FIX_0_765366865;

                int tmp0 = (workspace[workspaceIndex + 0] + workspace[workspaceIndex + 4]) << SLOW_INTEGER_CONST_BITS;
                int tmp1 = (workspace[workspaceIndex + 0] - workspace[workspaceIndex + 4]) << SLOW_INTEGER_CONST_BITS;

                int tmp10 = tmp0 + tmp3;
                int tmp13 = tmp0 - tmp3;
                int tmp11 = tmp1 + tmp2;
                int tmp12 = tmp1 - tmp2;

                /* Odd part per figure 8; the matrix is unitary and hence its
                * transpose is its inverse.  i0..i3 are y7,y5,y3,y1 respectively.
                */

                tmp0 = workspace[workspaceIndex + 7];
                tmp1 = workspace[workspaceIndex + 5];
                tmp2 = workspace[workspaceIndex + 3];
                tmp3 = workspace[workspaceIndex + 1];

                z1 = tmp0 + tmp3;
                z2 = tmp1 + tmp2;
                z3 = tmp0 + tmp2;
                int z4 = tmp1 + tmp3;
                int z5 = (z3 + z4) * SLOW_INTEGER_FIX_1_175875602; /* sqrt(2) * c3 */

                tmp0 = tmp0 * SLOW_INTEGER_FIX_0_298631336; /* sqrt(2) * (-c1+c3+c5-c7) */
                tmp1 = tmp1 * SLOW_INTEGER_FIX_2_053119869; /* sqrt(2) * ( c1+c3-c5+c7) */
                tmp2 = tmp2 * SLOW_INTEGER_FIX_3_072711026; /* sqrt(2) * ( c1+c3+c5-c7) */
                tmp3 = tmp3 * SLOW_INTEGER_FIX_1_501321110; /* sqrt(2) * ( c1+c3-c5-c7) */
                z1 = z1 * (-SLOW_INTEGER_FIX_0_899976223); /* sqrt(2) * (c7-c3) */
                z2 = z2 * (-SLOW_INTEGER_FIX_2_562915447); /* sqrt(2) * (-c1-c3) */
                z3 = z3 * (-SLOW_INTEGER_FIX_1_961570560); /* sqrt(2) * (-c3-c5) */
                z4 = z4 * (-SLOW_INTEGER_FIX_0_390180644); /* sqrt(2) * (c5-c3) */

                z3 += z5;
                z4 += z5;

                tmp0 += z1 + z3;
                tmp1 += z2 + z4;
                tmp2 += z2 + z3;
                tmp3 += z1 + z4;

                /* Final output stage: inputs are tmp10..tmp13, tmp0..tmp3 */

                m_componentBuffer[currentOutRow][output_col + 0] = limit[limitOffset + JpegUtils.DESCALE(tmp10 + tmp3, SLOW_INTEGER_CONST_BITS + SLOW_INTEGER_PASS1_BITS + 3) & RANGE_MASK];
                m_componentBuffer[currentOutRow][output_col + 7] = limit[limitOffset + JpegUtils.DESCALE(tmp10 - tmp3, SLOW_INTEGER_CONST_BITS + SLOW_INTEGER_PASS1_BITS + 3) & RANGE_MASK];
                m_componentBuffer[currentOutRow][output_col + 1] = limit[limitOffset + JpegUtils.DESCALE(tmp11 + tmp2, SLOW_INTEGER_CONST_BITS + SLOW_INTEGER_PASS1_BITS + 3) & RANGE_MASK];
                m_componentBuffer[currentOutRow][output_col + 6] = limit[limitOffset + JpegUtils.DESCALE(tmp11 - tmp2, SLOW_INTEGER_CONST_BITS + SLOW_INTEGER_PASS1_BITS + 3) & RANGE_MASK];
                m_componentBuffer[currentOutRow][output_col + 2] = limit[limitOffset + JpegUtils.DESCALE(tmp12 + tmp1, SLOW_INTEGER_CONST_BITS + SLOW_INTEGER_PASS1_BITS + 3) & RANGE_MASK];
                m_componentBuffer[currentOutRow][output_col + 5] = limit[limitOffset + JpegUtils.DESCALE(tmp12 - tmp1, SLOW_INTEGER_CONST_BITS + SLOW_INTEGER_PASS1_BITS + 3) & RANGE_MASK];
                m_componentBuffer[currentOutRow][output_col + 3] = limit[limitOffset + JpegUtils.DESCALE(tmp13 + tmp0, SLOW_INTEGER_CONST_BITS + SLOW_INTEGER_PASS1_BITS + 3) & RANGE_MASK];
                m_componentBuffer[currentOutRow][output_col + 4] = limit[limitOffset + JpegUtils.DESCALE(tmp13 - tmp0, SLOW_INTEGER_CONST_BITS + SLOW_INTEGER_PASS1_BITS + 3) & RANGE_MASK];

                /* advance pointer to next row */
                workspaceIndex += JpegConstants.DCTSize;
            }
        }

        /// <summary>
        /// Dequantize a coefficient by multiplying it by the multiplier-table
        /// entry; produce an int result.  In this module, both inputs and result
        /// are 16 bits or less, so either int or short multiply will work.
        /// </summary>
        private static int SLOW_INTEGER_DEQUANTIZE(int coef, int quantval)
        {
            return (coef * quantval);
        }

        /// <summary>
        /// Perform dequantization and inverse DCT on one block of coefficients.
        /// NOTE: this code only copes with 8x8 DCTs.
        /// 
        /// A fast, not so accurate integer implementation of the
        /// inverse DCT (Discrete Cosine Transform).  In the IJG code, this routine
        /// must also perform dequantization of the input coefficients.
        /// 
        /// A 2-D IDCT can be done by 1-D IDCT on each column followed by 1-D IDCT
        /// on each row (or vice versa, but it's more convenient to emit a row at
        /// a time).  Direct algorithms are also available, but they are much more
        /// complex and seem not to be any faster when reduced to code.
        /// 
        /// This implementation is based on Arai, Agui, and Nakajima's algorithm for
        /// scaled DCT.  Their original paper (Trans. IEICE E-71(11):1095) is in
        /// Japanese, but the algorithm is described in the Pennebaker &amp; Mitchell
        /// JPEG textbook (see REFERENCES section in file README).  The following code
        /// is based directly on figure 4-8 in P&amp;M.
        /// While an 8-point DCT cannot be done in less than 11 multiplies, it is
        /// possible to arrange the computation so that many of the multiplies are
        /// simple scalings of the final outputs.  These multiplies can then be
        /// folded into the multiplications or divisions by the JPEG quantization
        /// table entries.  The AA&amp;N method leaves only 5 multiplies and 29 adds
        /// to be done in the DCT itself.
        /// The primary disadvantage of this method is that with fixed-point math,
        /// accuracy is lost due to imprecise representation of the scaled
        /// quantization values.  The smaller the quantization table entry, the less
        /// precise the scaled value, so this implementation does worse with high-
        /// quality-setting files than with low-quality ones.
        /// 
        /// Scaling decisions are generally the same as in the LL&amp;M algorithm;
        /// However, we choose to descale
        /// (right shift) multiplication products as soon as they are formed,
        /// rather than carrying additional fractional bits into subsequent additions.
        /// This compromises accuracy slightly, but it lets us save a few shifts.
        /// More importantly, 16-bit arithmetic is then adequate (for 8-bit samples)
        /// everywhere except in the multiplications proper; this saves a good deal
        /// of work on 16-bit-int machines.
        /// 
        /// The dequantized coefficients are not integers because the AA&amp;N scaling
        /// factors have been incorporated.  We represent them scaled up by FAST_INTEGER_PASS1_BITS,
        /// so that the first and second IDCT rounds have the same input scaling.
        /// For 8-bit JSAMPLEs, we choose IFAST_SCALE_BITS = FAST_INTEGER_PASS1_BITS so as to
        /// avoid a descaling shift; this compromises accuracy rather drastically
        /// for small quantization table entries, but it saves a lot of shifts.
        /// For 12-bit JSAMPLEs, there's no hope of using 16x16 multiplies anyway,
        /// so we use a much larger scaling factor to preserve accuracy.
        /// 
        /// A final compromise is to represent the multiplicative constants to only
        /// 8 fractional bits, rather than 13.  This saves some shifting work on some
        /// machines, and may also reduce the cost of multiplication (since there
        /// are fewer one-bits in the constants).
        /// </summary>
        private void jpeg_idct_ifast(int component_index, short[] coef_block, int output_row, int output_col)
        {
            /* buffers data between passes */
            int[] workspace = new int[JpegConstants.DCTSize2];

            /* Pass 1: process columns from input, store into work array. */

            int coefBlockIndex = 0;
            int workspaceIndex = 0;

            int[] quantTable = m_dctTables[component_index].int_array;
            int quantTableIndex = 0;

            for (int ctr = JpegConstants.DCTSize; ctr > 0; ctr--)
            {
                /* Due to quantization, we will usually find that many of the input
                * coefficients are zero, especially the AC terms.  We can exploit this
                * by short-circuiting the IDCT calculation for any column in which all
                * the AC terms are zero.  In that case each output is equal to the
                * DC coefficient (with scale factor as needed).
                * With typical images and quantization tables, half or more of the
                * column DCT calculations can be simplified this way.
                */

                if (coef_block[coefBlockIndex + JpegConstants.DCTSize * 1] == 0 &&
                    coef_block[coefBlockIndex + JpegConstants.DCTSize * 2] == 0 &&
                    coef_block[coefBlockIndex + JpegConstants.DCTSize * 3] == 0 &&
                    coef_block[coefBlockIndex + JpegConstants.DCTSize * 4] == 0 &&
                    coef_block[coefBlockIndex + JpegConstants.DCTSize * 5] == 0 &&
                    coef_block[coefBlockIndex + JpegConstants.DCTSize * 6] == 0 &&
                    coef_block[coefBlockIndex + JpegConstants.DCTSize * 7] == 0)
                {
                    /* AC terms all zero */
                    int dcval = FAST_INTEGER_DEQUANTIZE(coef_block[coefBlockIndex + JpegConstants.DCTSize * 0],
                        quantTable[quantTableIndex + JpegConstants.DCTSize * 0]);

                    workspace[workspaceIndex + JpegConstants.DCTSize * 0] = dcval;
                    workspace[workspaceIndex + JpegConstants.DCTSize * 1] = dcval;
                    workspace[workspaceIndex + JpegConstants.DCTSize * 2] = dcval;
                    workspace[workspaceIndex + JpegConstants.DCTSize * 3] = dcval;
                    workspace[workspaceIndex + JpegConstants.DCTSize * 4] = dcval;
                    workspace[workspaceIndex + JpegConstants.DCTSize * 5] = dcval;
                    workspace[workspaceIndex + JpegConstants.DCTSize * 6] = dcval;
                    workspace[workspaceIndex + JpegConstants.DCTSize * 7] = dcval;

                    /* advance pointers to next column */
                    coefBlockIndex++;
                    quantTableIndex++;
                    workspaceIndex++;
                    continue;
                }

                /* Even part */

                int tmp0 = FAST_INTEGER_DEQUANTIZE(coef_block[coefBlockIndex + JpegConstants.DCTSize * 0],
                    quantTable[quantTableIndex + JpegConstants.DCTSize * 0]);
                int tmp1 = FAST_INTEGER_DEQUANTIZE(coef_block[coefBlockIndex + JpegConstants.DCTSize * 2],
                    quantTable[quantTableIndex + JpegConstants.DCTSize * 2]);
                int tmp2 = FAST_INTEGER_DEQUANTIZE(coef_block[coefBlockIndex + JpegConstants.DCTSize * 4],
                    quantTable[quantTableIndex + JpegConstants.DCTSize * 4]);
                int tmp3 = FAST_INTEGER_DEQUANTIZE(coef_block[coefBlockIndex + JpegConstants.DCTSize * 6],
                    quantTable[quantTableIndex + JpegConstants.DCTSize * 6]);

                int tmp10 = tmp0 + tmp2;    /* phase 3 */
                int tmp11 = tmp0 - tmp2;

                int tmp13 = tmp1 + tmp3;    /* phases 5-3 */
                int tmp12 = FAST_INTEGER_MULTIPLY(tmp1 - tmp3, FAST_INTEGER_FIX_1_414213562) - tmp13; /* 2*c4 */

                tmp0 = tmp10 + tmp13;   /* phase 2 */
                tmp3 = tmp10 - tmp13;
                tmp1 = tmp11 + tmp12;
                tmp2 = tmp11 - tmp12;

                /* Odd part */

                int tmp4 = FAST_INTEGER_DEQUANTIZE(coef_block[coefBlockIndex + JpegConstants.DCTSize * 1],
                    quantTable[quantTableIndex + JpegConstants.DCTSize * 1]);
                int tmp5 = FAST_INTEGER_DEQUANTIZE(coef_block[coefBlockIndex + JpegConstants.DCTSize * 3],
                    quantTable[quantTableIndex + JpegConstants.DCTSize * 3]);
                int tmp6 = FAST_INTEGER_DEQUANTIZE(coef_block[coefBlockIndex + JpegConstants.DCTSize * 5],
                    quantTable[quantTableIndex + JpegConstants.DCTSize * 5]);
                int tmp7 = FAST_INTEGER_DEQUANTIZE(coef_block[coefBlockIndex + JpegConstants.DCTSize * 7],
                    quantTable[quantTableIndex + JpegConstants.DCTSize * 7]);

                int z13 = tmp6 + tmp5;      /* phase 6 */
                int z10 = tmp6 - tmp5;
                int z11 = tmp4 + tmp7;
                int z12 = tmp4 - tmp7;

                tmp7 = z11 + z13;       /* phase 5 */
                tmp11 = FAST_INTEGER_MULTIPLY(z11 - z13, FAST_INTEGER_FIX_1_414213562); /* 2*c4 */

                int z5 = FAST_INTEGER_MULTIPLY(z10 + z12, FAST_INTEGER_FIX_1_847759065); /* 2*c2 */
                tmp10 = FAST_INTEGER_MULTIPLY(z12, FAST_INTEGER_FIX_1_082392200) - z5; /* 2*(c2-c6) */
                tmp12 = FAST_INTEGER_MULTIPLY(z10, -FAST_INTEGER_FIX_2_613125930) + z5; /* -2*(c2+c6) */

                tmp6 = tmp12 - tmp7;    /* phase 2 */
                tmp5 = tmp11 - tmp6;
                tmp4 = tmp10 + tmp5;

                workspace[workspaceIndex + JpegConstants.DCTSize * 0] = tmp0 + tmp7;
                workspace[workspaceIndex + JpegConstants.DCTSize * 7] = tmp0 - tmp7;
                workspace[workspaceIndex + JpegConstants.DCTSize * 1] = tmp1 + tmp6;
                workspace[workspaceIndex + JpegConstants.DCTSize * 6] = tmp1 - tmp6;
                workspace[workspaceIndex + JpegConstants.DCTSize * 2] = tmp2 + tmp5;
                workspace[workspaceIndex + JpegConstants.DCTSize * 5] = tmp2 - tmp5;
                workspace[workspaceIndex + JpegConstants.DCTSize * 4] = tmp3 + tmp4;
                workspace[workspaceIndex + JpegConstants.DCTSize * 3] = tmp3 - tmp4;

                /* advance pointers to next column */
                coefBlockIndex++;
                quantTableIndex++;
                workspaceIndex++;
            }

            /* Pass 2: process rows from work array, store into output array. */
            /* Note that we must descale the results by a factor of 8 == 2**3, */
            /* and also undo the FAST_INTEGER_PASS1_BITS scaling. */

            workspaceIndex = 0;
            byte[] limit = m_cinfo.m_sample_range_limit;
            int limitOffset = m_cinfo.m_sampleRangeLimitOffset + JpegConstants.MediumSampleValue;

            for (int ctr = 0; ctr < JpegConstants.DCTSize; ctr++)
            {
                int currentOutRow = output_row + ctr;

                /* Rows of zeroes can be exploited in the same way as we did with columns.
                * However, the column calculation has created many nonzero AC terms, so
                * the simplification applies less often (typically 5% to 10% of the time).
                * On machines with very fast multiplication, it's possible that the
                * test takes more time than it's worth.  In that case this section
                * may be commented out.
                */

                if (workspace[workspaceIndex + 1] == 0 &&
                    workspace[workspaceIndex + 2] == 0 &&
                    workspace[workspaceIndex + 3] == 0 &&
                    workspace[workspaceIndex + 4] == 0 &&
                    workspace[workspaceIndex + 5] == 0 &&
                    workspace[workspaceIndex + 6] == 0 &&
                    workspace[workspaceIndex + 7] == 0)
                {
                    /* AC terms all zero */
                    byte dcval = limit[limitOffset + FAST_INTEGER_IDESCALE(workspace[workspaceIndex + 0], FAST_INTEGER_PASS1_BITS + 3) & RANGE_MASK];

                    m_componentBuffer[currentOutRow][output_col + 0] = dcval;
                    m_componentBuffer[currentOutRow][output_col + 1] = dcval;
                    m_componentBuffer[currentOutRow][output_col + 2] = dcval;
                    m_componentBuffer[currentOutRow][output_col + 3] = dcval;
                    m_componentBuffer[currentOutRow][output_col + 4] = dcval;
                    m_componentBuffer[currentOutRow][output_col + 5] = dcval;
                    m_componentBuffer[currentOutRow][output_col + 6] = dcval;
                    m_componentBuffer[currentOutRow][output_col + 7] = dcval;

                    /* advance pointer to next row */
                    workspaceIndex += JpegConstants.DCTSize;
                    continue;
                }

                /* Even part */

                int tmp10 = workspace[workspaceIndex + 0] + workspace[workspaceIndex + 4];
                int tmp11 = workspace[workspaceIndex + 0] - workspace[workspaceIndex + 4];

                int tmp13 = workspace[workspaceIndex + 2] + workspace[workspaceIndex + 6];
                int tmp12 = FAST_INTEGER_MULTIPLY(workspace[workspaceIndex + 2] - workspace[workspaceIndex + 6], FAST_INTEGER_FIX_1_414213562) - tmp13;

                int tmp0 = tmp10 + tmp13;
                int tmp3 = tmp10 - tmp13;
                int tmp1 = tmp11 + tmp12;
                int tmp2 = tmp11 - tmp12;

                /* Odd part */

                int z13 = workspace[workspaceIndex + 5] + workspace[workspaceIndex + 3];
                int z10 = workspace[workspaceIndex + 5] - workspace[workspaceIndex + 3];
                int z11 = workspace[workspaceIndex + 1] + workspace[workspaceIndex + 7];
                int z12 = workspace[workspaceIndex + 1] - workspace[workspaceIndex + 7];

                int tmp7 = z11 + z13;       /* phase 5 */
                tmp11 = FAST_INTEGER_MULTIPLY(z11 - z13, FAST_INTEGER_FIX_1_414213562); /* 2*c4 */

                int z5 = FAST_INTEGER_MULTIPLY(z10 + z12, FAST_INTEGER_FIX_1_847759065); /* 2*c2 */
                tmp10 = FAST_INTEGER_MULTIPLY(z12, FAST_INTEGER_FIX_1_082392200) - z5; /* 2*(c2-c6) */
                tmp12 = FAST_INTEGER_MULTIPLY(z10, -FAST_INTEGER_FIX_2_613125930) + z5; /* -2*(c2+c6) */

                int tmp6 = tmp12 - tmp7;    /* phase 2 */
                int tmp5 = tmp11 - tmp6;
                int tmp4 = tmp10 + tmp5;

                /* Final output stage: scale down by a factor of 8 and range-limit */

                m_componentBuffer[currentOutRow][output_col + 0] = limit[limitOffset + FAST_INTEGER_IDESCALE(tmp0 + tmp7, FAST_INTEGER_PASS1_BITS + 3) & RANGE_MASK];
                m_componentBuffer[currentOutRow][output_col + 7] = limit[limitOffset + FAST_INTEGER_IDESCALE(tmp0 - tmp7, FAST_INTEGER_PASS1_BITS + 3) & RANGE_MASK];
                m_componentBuffer[currentOutRow][output_col + 1] = limit[limitOffset + FAST_INTEGER_IDESCALE(tmp1 + tmp6, FAST_INTEGER_PASS1_BITS + 3) & RANGE_MASK];
                m_componentBuffer[currentOutRow][output_col + 6] = limit[limitOffset + FAST_INTEGER_IDESCALE(tmp1 - tmp6, FAST_INTEGER_PASS1_BITS + 3) & RANGE_MASK];
                m_componentBuffer[currentOutRow][output_col + 2] = limit[limitOffset + FAST_INTEGER_IDESCALE(tmp2 + tmp5, FAST_INTEGER_PASS1_BITS + 3) & RANGE_MASK];
                m_componentBuffer[currentOutRow][output_col + 5] = limit[limitOffset + FAST_INTEGER_IDESCALE(tmp2 - tmp5, FAST_INTEGER_PASS1_BITS + 3) & RANGE_MASK];
                m_componentBuffer[currentOutRow][output_col + 4] = limit[limitOffset + FAST_INTEGER_IDESCALE(tmp3 + tmp4, FAST_INTEGER_PASS1_BITS + 3) & RANGE_MASK];
                m_componentBuffer[currentOutRow][output_col + 3] = limit[limitOffset + FAST_INTEGER_IDESCALE(tmp3 - tmp4, FAST_INTEGER_PASS1_BITS + 3) & RANGE_MASK];

                /* advance pointer to next row */
                workspaceIndex += JpegConstants.DCTSize;
            }
        }

        /// <summary>
        /// Multiply a DCTELEM variable by an int constant, and immediately
        /// descale to yield a DCTELEM result.
        /// </summary>
        private static int FAST_INTEGER_MULTIPLY(int var, int c)
        {
            return (JpegUtils.DESCALE(var * c, FAST_INTEGER_CONST_BITS));
        }

        /// <summary>
        /// Dequantize a coefficient by multiplying it by the multiplier-table
        /// entry; produce a DCTELEM result.  For 8-bit data a 16x16->16
        /// multiplication will do.  For 12-bit data, the multiplier table is
        /// declared int, so a 32-bit multiply will be used.
        /// </summary>
        private static int FAST_INTEGER_DEQUANTIZE(short coef, int quantval)
        {
            return ((int)coef * quantval);
        }

        /// <summary>
        /// Like DESCALE, but applies to a DCTELEM and produces an int.
        /// We assume that int right shift is unsigned if int right shift is.
        /// </summary>
        private static int FAST_INTEGER_IRIGHT_SHIFT(int x, int shft)
        {
            return (x >> shft);
        }

        private static int FAST_INTEGER_IDESCALE(int x, int n)
        {
            return (FAST_INTEGER_IRIGHT_SHIFT((x) + (1 << ((n) - 1)), n));
        }

        /// <summary>
        /// Perform dequantization and inverse DCT on one block of coefficients.
        /// NOTE: this code only copes with 8x8 DCTs.
        /// 
        /// A floating-point implementation of the
        /// inverse DCT (Discrete Cosine Transform).  In the IJG code, this routine
        /// must also perform dequantization of the input coefficients.
        /// 
        /// This implementation should be more accurate than either of the integer
        /// IDCT implementations.  However, it may not give the same results on all
        /// machines because of differences in roundoff behavior.  Speed will depend
        /// on the hardware's floating point capacity.
        /// 
        /// A 2-D IDCT can be done by 1-D IDCT on each column followed by 1-D IDCT
        /// on each row (or vice versa, but it's more convenient to emit a row at
        /// a time).  Direct algorithms are also available, but they are much more
        /// complex and seem not to be any faster when reduced to code.
        /// 
        /// This implementation is based on Arai, Agui, and Nakajima's algorithm for
        /// scaled DCT.  Their original paper (Trans. IEICE E-71(11):1095) is in
        /// Japanese, but the algorithm is described in the Pennebaker &amp; Mitchell
        /// JPEG textbook (see REFERENCES section in file README).  The following code
        /// is based directly on figure 4-8 in P&amp;M.
        /// While an 8-point DCT cannot be done in less than 11 multiplies, it is
        /// possible to arrange the computation so that many of the multiplies are
        /// simple scalings of the final outputs.  These multiplies can then be
        /// folded into the multiplications or divisions by the JPEG quantization
        /// table entries.  The AA&amp;N method leaves only 5 multiplies and 29 adds
        /// to be done in the DCT itself.
        /// The primary disadvantage of this method is that with a fixed-point
        /// implementation, accuracy is lost due to imprecise representation of the
        /// scaled quantization values.  However, that problem does not arise if
        /// we use floating point arithmetic.
        /// </summary>
        private void jpeg_idct_float(int component_index, short[] coef_block, int output_row, int output_col)
        {
            /* buffers data between passes */
            float[] workspace = new float[JpegConstants.DCTSize2];

            /* Pass 1: process columns from input, store into work array. */

            int coefBlockIndex = 0;
            int workspaceIndex = 0;

            float[] quantTable = m_dctTables[component_index].float_array;
            int quantTableIndex = 0;

            for (int ctr = JpegConstants.DCTSize; ctr > 0; ctr--)
            {
                /* Due to quantization, we will usually find that many of the input
                * coefficients are zero, especially the AC terms.  We can exploit this
                * by short-circuiting the IDCT calculation for any column in which all
                * the AC terms are zero.  In that case each output is equal to the
                * DC coefficient (with scale factor as needed).
                * With typical images and quantization tables, half or more of the
                * column DCT calculations can be simplified this way.
                */

                if (coef_block[coefBlockIndex + JpegConstants.DCTSize * 1] == 0 &&
                    coef_block[coefBlockIndex + JpegConstants.DCTSize * 2] == 0 &&
                    coef_block[coefBlockIndex + JpegConstants.DCTSize * 3] == 0 &&
                    coef_block[coefBlockIndex + JpegConstants.DCTSize * 4] == 0 &&
                    coef_block[coefBlockIndex + JpegConstants.DCTSize * 5] == 0 &&
                    coef_block[coefBlockIndex + JpegConstants.DCTSize * 6] == 0 &&
                    coef_block[coefBlockIndex + JpegConstants.DCTSize * 7] == 0)
                {
                    /* AC terms all zero */
                    float dcval = FLOAT_DEQUANTIZE(coef_block[coefBlockIndex + JpegConstants.DCTSize * 0],
                        quantTable[quantTableIndex + JpegConstants.DCTSize * 0]);

                    workspace[workspaceIndex + JpegConstants.DCTSize * 0] = dcval;
                    workspace[workspaceIndex + JpegConstants.DCTSize * 1] = dcval;
                    workspace[workspaceIndex + JpegConstants.DCTSize * 2] = dcval;
                    workspace[workspaceIndex + JpegConstants.DCTSize * 3] = dcval;
                    workspace[workspaceIndex + JpegConstants.DCTSize * 4] = dcval;
                    workspace[workspaceIndex + JpegConstants.DCTSize * 5] = dcval;
                    workspace[workspaceIndex + JpegConstants.DCTSize * 6] = dcval;
                    workspace[workspaceIndex + JpegConstants.DCTSize * 7] = dcval;

                    coefBlockIndex++;            /* advance pointers to next column */
                    quantTableIndex++;
                    workspaceIndex++;
                    continue;
                }

                /* Even part */

                float tmp0 = FLOAT_DEQUANTIZE(coef_block[coefBlockIndex + JpegConstants.DCTSize * 0],
                    quantTable[quantTableIndex + JpegConstants.DCTSize * 0]);
                float tmp1 = FLOAT_DEQUANTIZE(coef_block[coefBlockIndex + JpegConstants.DCTSize * 2],
                    quantTable[quantTableIndex + JpegConstants.DCTSize * 2]);
                float tmp2 = FLOAT_DEQUANTIZE(coef_block[coefBlockIndex + JpegConstants.DCTSize * 4],
                    quantTable[quantTableIndex + JpegConstants.DCTSize * 4]);
                float tmp3 = FLOAT_DEQUANTIZE(coef_block[coefBlockIndex + JpegConstants.DCTSize * 6],
                    quantTable[quantTableIndex + JpegConstants.DCTSize * 6]);

                float tmp10 = tmp0 + tmp2;    /* phase 3 */
                float tmp11 = tmp0 - tmp2;

                float tmp13 = tmp1 + tmp3;    /* phases 5-3 */
                float tmp12 = (tmp1 - tmp3) * 1.414213562f - tmp13; /* 2*c4 */

                tmp0 = tmp10 + tmp13;   /* phase 2 */
                tmp3 = tmp10 - tmp13;
                tmp1 = tmp11 + tmp12;
                tmp2 = tmp11 - tmp12;

                /* Odd part */

                float tmp4 = FLOAT_DEQUANTIZE(coef_block[coefBlockIndex + JpegConstants.DCTSize * 1],
                    quantTable[quantTableIndex + JpegConstants.DCTSize * 1]);
                float tmp5 = FLOAT_DEQUANTIZE(coef_block[coefBlockIndex + JpegConstants.DCTSize * 3],
                    quantTable[quantTableIndex + JpegConstants.DCTSize * 3]);
                float tmp6 = FLOAT_DEQUANTIZE(coef_block[coefBlockIndex + JpegConstants.DCTSize * 5],
                    quantTable[quantTableIndex + JpegConstants.DCTSize * 5]);
                float tmp7 = FLOAT_DEQUANTIZE(coef_block[coefBlockIndex + JpegConstants.DCTSize * 7],
                    quantTable[quantTableIndex + JpegConstants.DCTSize * 7]);

                float z13 = tmp6 + tmp5;      /* phase 6 */
                float z10 = tmp6 - tmp5;
                float z11 = tmp4 + tmp7;
                float z12 = tmp4 - tmp7;

                tmp7 = z11 + z13;       /* phase 5 */
                tmp11 = (z11 - z13) * 1.414213562f; /* 2*c4 */

                float z5 = (z10 + z12) * 1.847759065f; /* 2*c2 */
                tmp10 = 1.082392200f * z12 - z5; /* 2*(c2-c6) */
                tmp12 = -2.613125930f * z10 + z5; /* -2*(c2+c6) */

                tmp6 = tmp12 - tmp7;    /* phase 2 */
                tmp5 = tmp11 - tmp6;
                tmp4 = tmp10 + tmp5;

                workspace[workspaceIndex + JpegConstants.DCTSize * 0] = tmp0 + tmp7;
                workspace[workspaceIndex + JpegConstants.DCTSize * 7] = tmp0 - tmp7;
                workspace[workspaceIndex + JpegConstants.DCTSize * 1] = tmp1 + tmp6;
                workspace[workspaceIndex + JpegConstants.DCTSize * 6] = tmp1 - tmp6;
                workspace[workspaceIndex + JpegConstants.DCTSize * 2] = tmp2 + tmp5;
                workspace[workspaceIndex + JpegConstants.DCTSize * 5] = tmp2 - tmp5;
                workspace[workspaceIndex + JpegConstants.DCTSize * 4] = tmp3 + tmp4;
                workspace[workspaceIndex + JpegConstants.DCTSize * 3] = tmp3 - tmp4;

                coefBlockIndex++;            /* advance pointers to next column */
                quantTableIndex++;
                workspaceIndex++;
            }

            /* Pass 2: process rows from work array, store into output array. */
            /* Note that we must descale the results by a factor of 8 == 2**3. */
            workspaceIndex = 0;
            byte[] limit = m_cinfo.m_sample_range_limit;
            int limitOffset = m_cinfo.m_sampleRangeLimitOffset + JpegConstants.MediumSampleValue;

            for (int ctr = 0; ctr < JpegConstants.DCTSize; ctr++)
            {
                /* Rows of zeroes can be exploited in the same way as we did with columns.
                * However, the column calculation has created many nonzero AC terms, so
                * the simplification applies less often (typically 5% to 10% of the time).
                * And testing floats for zero is relatively expensive, so we don't bother.
                */

                /* Even part */

                float tmp10 = workspace[workspaceIndex + 0] + workspace[workspaceIndex + 4];
                float tmp11 = workspace[workspaceIndex + 0] - workspace[workspaceIndex + 4];

                float tmp13 = workspace[workspaceIndex + 2] + workspace[workspaceIndex + 6];
                float tmp12 = (workspace[workspaceIndex + 2] - workspace[workspaceIndex + 6]) * 1.414213562f - tmp13;

                float tmp0 = tmp10 + tmp13;
                float tmp3 = tmp10 - tmp13;
                float tmp1 = tmp11 + tmp12;
                float tmp2 = tmp11 - tmp12;

                /* Odd part */

                float z13 = workspace[workspaceIndex + 5] + workspace[workspaceIndex + 3];
                float z10 = workspace[workspaceIndex + 5] - workspace[workspaceIndex + 3];
                float z11 = workspace[workspaceIndex + 1] + workspace[workspaceIndex + 7];
                float z12 = workspace[workspaceIndex + 1] - workspace[workspaceIndex + 7];

                float tmp7 = z11 + z13;
                tmp11 = (z11 - z13) * 1.414213562f;

                float z5 = (z10 + z12) * 1.847759065f; /* 2*c2 */
                tmp10 = 1.082392200f * z12 - z5; /* 2*(c2-c6) */
                tmp12 = -2.613125930f * z10 + z5; /* -2*(c2+c6) */

                float tmp6 = tmp12 - tmp7;
                float tmp5 = tmp11 - tmp6;
                float tmp4 = tmp10 + tmp5;

                /* Final output stage: scale down by a factor of 8 and range-limit */
                int currentOutRow = output_row + ctr;
                m_componentBuffer[currentOutRow][output_col + 0] = limit[limitOffset + JpegUtils.DESCALE((int)(tmp0 + tmp7), 3) & RANGE_MASK];
                m_componentBuffer[currentOutRow][output_col + 7] = limit[limitOffset + JpegUtils.DESCALE((int)(tmp0 - tmp7), 3) & RANGE_MASK];
                m_componentBuffer[currentOutRow][output_col + 1] = limit[limitOffset + JpegUtils.DESCALE((int)(tmp1 + tmp6), 3) & RANGE_MASK];
                m_componentBuffer[currentOutRow][output_col + 6] = limit[limitOffset + JpegUtils.DESCALE((int)(tmp1 - tmp6), 3) & RANGE_MASK];
                m_componentBuffer[currentOutRow][output_col + 2] = limit[limitOffset + JpegUtils.DESCALE((int)(tmp2 + tmp5), 3) & RANGE_MASK];
                m_componentBuffer[currentOutRow][output_col + 5] = limit[limitOffset + JpegUtils.DESCALE((int)(tmp2 - tmp5), 3) & RANGE_MASK];
                m_componentBuffer[currentOutRow][output_col + 4] = limit[limitOffset + JpegUtils.DESCALE((int)(tmp3 + tmp4), 3) & RANGE_MASK];
                m_componentBuffer[currentOutRow][output_col + 3] = limit[limitOffset + JpegUtils.DESCALE((int)(tmp3 - tmp4), 3) & RANGE_MASK];

                workspaceIndex += JpegConstants.DCTSize;       /* advance pointer to next row */
            }
        }

        /// <summary>
        /// Dequantize a coefficient by multiplying it by the multiplier-table
        /// entry; produce a float result.
        /// </summary>
        private static float FLOAT_DEQUANTIZE(short coef, float quantval)
        {
            return (((float)(coef)) * (quantval));
        }

        /// <summary>
        /// Inverse-DCT routines that produce reduced-size output:
        /// either 4x4, 2x2, or 1x1 pixels from an 8x8 DCT block.
        /// 
        /// NOTE: this code only copes with 8x8 DCTs.
        /// 
        /// The implementation is based on the Loeffler, Ligtenberg and Moschytz (LL&amp;M)
        /// algorithm. We simply replace each 8-to-8 1-D IDCT step
        /// with an 8-to-4 step that produces the four averages of two adjacent outputs
        /// (or an 8-to-2 step producing two averages of four outputs, for 2x2 output).
        /// These steps were derived by computing the corresponding values at the end
        /// of the normal LL&amp;M code, then simplifying as much as possible.
        /// 
        /// 1x1 is trivial: just take the DC coefficient divided by 8.
        /// 
        /// Perform dequantization and inverse DCT on one block of coefficients,
        /// producing a reduced-size 4x4 output block.
        /// </summary>
        private void jpeg_idct_4x4(int component_index, short[] coef_block, int output_row, int output_col)
        {
            /* buffers data between passes */
            int[] workspace = new int[JpegConstants.DCTSize * 4];

            /* Pass 1: process columns from input, store into work array. */
            int coefBlockIndex = 0;
            int workspaceIndex = 0;

            int[] quantTable = m_dctTables[component_index].int_array;
            int quantTableIndex = 0;

            for (int ctr = JpegConstants.DCTSize; ctr > 0; coefBlockIndex++, quantTableIndex++, workspaceIndex++, ctr--)
            {
                /* Don't bother to process column 4, because second pass won't use it */
                if (ctr == JpegConstants.DCTSize - 4)
                    continue;

                if (coef_block[coefBlockIndex + JpegConstants.DCTSize * 1] == 0 &&
                    coef_block[coefBlockIndex + JpegConstants.DCTSize * 2] == 0 &&
                    coef_block[coefBlockIndex + JpegConstants.DCTSize * 3] == 0 &&
                    coef_block[coefBlockIndex + JpegConstants.DCTSize * 5] == 0 &&
                    coef_block[coefBlockIndex + JpegConstants.DCTSize * 6] == 0 &&
                    coef_block[coefBlockIndex + JpegConstants.DCTSize * 7] == 0)
                {
                    /* AC terms all zero; we need not examine term 4 for 4x4 output */
                    int dcval = REDUCED_DEQUANTIZE(coef_block[coefBlockIndex + JpegConstants.DCTSize * 0],
                        quantTable[quantTableIndex + JpegConstants.DCTSize * 0]) << REDUCED_PASS1_BITS;

                    workspace[workspaceIndex + JpegConstants.DCTSize * 0] = dcval;
                    workspace[workspaceIndex + JpegConstants.DCTSize * 1] = dcval;
                    workspace[workspaceIndex + JpegConstants.DCTSize * 2] = dcval;
                    workspace[workspaceIndex + JpegConstants.DCTSize * 3] = dcval;

                    continue;
                }

                /* Even part */

                int tmp0 = REDUCED_DEQUANTIZE(coef_block[coefBlockIndex + JpegConstants.DCTSize * 0],
                    quantTable[quantTableIndex + JpegConstants.DCTSize * 0]);
                tmp0 <<= (REDUCED_CONST_BITS + 1);

                int z2 = REDUCED_DEQUANTIZE(coef_block[coefBlockIndex + JpegConstants.DCTSize * 2],
                    quantTable[quantTableIndex + JpegConstants.DCTSize * 2]);
                int z3 = REDUCED_DEQUANTIZE(coef_block[coefBlockIndex + JpegConstants.DCTSize * 6],
                    quantTable[quantTableIndex + JpegConstants.DCTSize * 6]);

                int tmp2 = z2 * REDUCED_FIX_1_847759065 + z3 * (-REDUCED_FIX_0_765366865);

                int tmp10 = tmp0 + tmp2;
                int tmp12 = tmp0 - tmp2;

                /* Odd part */

                int z1 = REDUCED_DEQUANTIZE(coef_block[coefBlockIndex + JpegConstants.DCTSize * 7],
                    quantTable[quantTableIndex + JpegConstants.DCTSize * 7]);
                z2 = REDUCED_DEQUANTIZE(coef_block[coefBlockIndex + JpegConstants.DCTSize * 5],
                    quantTable[quantTableIndex + JpegConstants.DCTSize * 5]);
                z3 = REDUCED_DEQUANTIZE(coef_block[coefBlockIndex + JpegConstants.DCTSize * 3],
                    quantTable[quantTableIndex + JpegConstants.DCTSize * 3]);
                int z4 = REDUCED_DEQUANTIZE(coef_block[coefBlockIndex + JpegConstants.DCTSize * 1],
                    quantTable[quantTableIndex + JpegConstants.DCTSize * 1]);

                tmp0 = z1 * (-REDUCED_FIX_0_211164243) /* sqrt(2) * (c3-c1) */ +
                       z2 * REDUCED_FIX_1_451774981 /* sqrt(2) * (c3+c7) */ +
                       z3 * (-REDUCED_FIX_2_172734803) /* sqrt(2) * (-c1-c5) */ +
                       z4 * REDUCED_FIX_1_061594337; /* sqrt(2) * (c5+c7) */

                tmp2 = z1 * (-REDUCED_FIX_0_509795579) /* sqrt(2) * (c7-c5) */ +
                       z2 * (-REDUCED_FIX_0_601344887) /* sqrt(2) * (c5-c1) */ +
                       z3 * REDUCED_FIX_0_899976223 /* sqrt(2) * (c3-c7) */ +
                       z4 * REDUCED_FIX_2_562915447; /* sqrt(2) * (c1+c3) */

                /* Final output stage */

                workspace[workspaceIndex + JpegConstants.DCTSize * 0] = JpegUtils.DESCALE(tmp10 + tmp2, REDUCED_CONST_BITS - REDUCED_PASS1_BITS + 1);
                workspace[workspaceIndex + JpegConstants.DCTSize * 3] = JpegUtils.DESCALE(tmp10 - tmp2, REDUCED_CONST_BITS - REDUCED_PASS1_BITS + 1);
                workspace[workspaceIndex + JpegConstants.DCTSize * 1] = JpegUtils.DESCALE(tmp12 + tmp0, REDUCED_CONST_BITS - REDUCED_PASS1_BITS + 1);
                workspace[workspaceIndex + JpegConstants.DCTSize * 2] = JpegUtils.DESCALE(tmp12 - tmp0, REDUCED_CONST_BITS - REDUCED_PASS1_BITS + 1);
            }

            /* Pass 2: process 4 rows from work array, store into output array. */
            byte[] limit = m_cinfo.m_sample_range_limit;
            int limitOffset = m_cinfo.m_sampleRangeLimitOffset + JpegConstants.MediumSampleValue;

            workspaceIndex = 0;
            for (int ctr = 0; ctr < 4; ctr++)
            {
                int currentOutRow = output_row + ctr;
                /* It's not clear whether a zero row test is worthwhile here ... */

                if (workspace[workspaceIndex + 1] == 0 &&
                    workspace[workspaceIndex + 2] == 0 &&
                    workspace[workspaceIndex + 3] == 0 &&
                    workspace[workspaceIndex + 5] == 0 &&
                    workspace[workspaceIndex + 6] == 0 &&
                    workspace[workspaceIndex + 7] == 0)
                {
                    /* AC terms all zero */
                    byte dcval = limit[limitOffset + JpegUtils.DESCALE(workspace[workspaceIndex + 0], REDUCED_PASS1_BITS + 3) & RANGE_MASK];

                    m_componentBuffer[currentOutRow][output_col + 0] = dcval;
                    m_componentBuffer[currentOutRow][output_col + 1] = dcval;
                    m_componentBuffer[currentOutRow][output_col + 2] = dcval;
                    m_componentBuffer[currentOutRow][output_col + 3] = dcval;

                    workspaceIndex += JpegConstants.DCTSize;       /* advance pointer to next row */
                    continue;
                }

                /* Even part */

                int tmp0 = (workspace[workspaceIndex + 0]) << (REDUCED_CONST_BITS + 1);

                int tmp2 = workspace[workspaceIndex + 2] * REDUCED_FIX_1_847759065 + workspace[workspaceIndex + 6] * (-REDUCED_FIX_0_765366865);

                int tmp10 = tmp0 + tmp2;
                int tmp12 = tmp0 - tmp2;

                /* Odd part */

                int z1 = workspace[workspaceIndex + 7];
                int z2 = workspace[workspaceIndex + 5];
                int z3 = workspace[workspaceIndex + 3];
                int z4 = workspace[workspaceIndex + 1];

                tmp0 = z1 * (-REDUCED_FIX_0_211164243) /* sqrt(2) * (c3-c1) */ +
                       z2 * REDUCED_FIX_1_451774981 /* sqrt(2) * (c3+c7) */ +
                       z3 * (-REDUCED_FIX_2_172734803) /* sqrt(2) * (-c1-c5) */ +
                       z4 * REDUCED_FIX_1_061594337; /* sqrt(2) * (c5+c7) */

                tmp2 = z1 * (-REDUCED_FIX_0_509795579) /* sqrt(2) * (c7-c5) */ +
                       z2 * (-REDUCED_FIX_0_601344887) /* sqrt(2) * (c5-c1) */ +
                       z3 * REDUCED_FIX_0_899976223 /* sqrt(2) * (c3-c7) */ +
                       z4 * REDUCED_FIX_2_562915447; /* sqrt(2) * (c1+c3) */

                /* Final output stage */

                m_componentBuffer[currentOutRow][output_col + 0] = limit[limitOffset + JpegUtils.DESCALE(tmp10 + tmp2, REDUCED_CONST_BITS + REDUCED_PASS1_BITS + 3 + 1) & RANGE_MASK];
                m_componentBuffer[currentOutRow][output_col + 3] = limit[limitOffset + JpegUtils.DESCALE(tmp10 - tmp2, REDUCED_CONST_BITS + REDUCED_PASS1_BITS + 3 + 1) & RANGE_MASK];
                m_componentBuffer[currentOutRow][output_col + 1] = limit[limitOffset + JpegUtils.DESCALE(tmp12 + tmp0, REDUCED_CONST_BITS + REDUCED_PASS1_BITS + 3 + 1) & RANGE_MASK];
                m_componentBuffer[currentOutRow][output_col + 2] = limit[limitOffset + JpegUtils.DESCALE(tmp12 - tmp0, REDUCED_CONST_BITS + REDUCED_PASS1_BITS + 3 + 1) & RANGE_MASK];

                workspaceIndex += JpegConstants.DCTSize;       /* advance pointer to next row */
            }
        }

        /// <summary>
        /// Perform dequantization and inverse DCT on one block of coefficients,
        /// producing a reduced-size 2x2 output block.
        /// </summary>
        private void jpeg_idct_2x2(int component_index, short[] coef_block, int output_row, int output_col)
        {
            /* buffers data between passes */
            int[] workspace = new int[JpegConstants.DCTSize * 2];

            /* Pass 1: process columns from input, store into work array. */
            int coefBlockIndex = 0;
            int workspaceIndex = 0;

            int[] quantTable = m_dctTables[component_index].int_array;
            int quantTableIndex = 0;

            for (int ctr = JpegConstants.DCTSize; ctr > 0; coefBlockIndex++, quantTableIndex++, workspaceIndex++, ctr--)
            {
                /* Don't bother to process columns 2,4,6 */
                if (ctr == JpegConstants.DCTSize - 2 || ctr == JpegConstants.DCTSize - 4 || ctr == JpegConstants.DCTSize - 6)
                    continue;

                if (coef_block[coefBlockIndex + JpegConstants.DCTSize * 1] == 0 &&
                    coef_block[coefBlockIndex + JpegConstants.DCTSize * 3] == 0 &&
                    coef_block[coefBlockIndex + JpegConstants.DCTSize * 5] == 0 &&
                    coef_block[coefBlockIndex + JpegConstants.DCTSize * 7] == 0)
                {
                    /* AC terms all zero; we need not examine terms 2,4,6 for 2x2 output */
                    int dcval = REDUCED_DEQUANTIZE(coef_block[coefBlockIndex + JpegConstants.DCTSize * 0],
                        quantTable[quantTableIndex + JpegConstants.DCTSize * 0]) << REDUCED_PASS1_BITS;

                    workspace[workspaceIndex + JpegConstants.DCTSize * 0] = dcval;
                    workspace[workspaceIndex + JpegConstants.DCTSize * 1] = dcval;

                    continue;
                }

                /* Even part */

                int z1 = REDUCED_DEQUANTIZE(coef_block[coefBlockIndex + JpegConstants.DCTSize * 0],
                    quantTable[quantTableIndex + JpegConstants.DCTSize * 0]);
                int tmp10 = z1 << (REDUCED_CONST_BITS + 2);

                /* Odd part */

                z1 = REDUCED_DEQUANTIZE(coef_block[coefBlockIndex + JpegConstants.DCTSize * 7],
                    quantTable[quantTableIndex + JpegConstants.DCTSize * 7]);
                int tmp0 = z1 * -REDUCED_FIX_0_720959822; /* sqrt(2) * (c7-c5+c3-c1) */
                z1 = REDUCED_DEQUANTIZE(coef_block[coefBlockIndex + JpegConstants.DCTSize * 5],
                    quantTable[quantTableIndex + JpegConstants.DCTSize * 5]);
                tmp0 += z1 * REDUCED_FIX_0_850430095; /* sqrt(2) * (-c1+c3+c5+c7) */
                z1 = REDUCED_DEQUANTIZE(coef_block[coefBlockIndex + JpegConstants.DCTSize * 3],
                    quantTable[quantTableIndex + JpegConstants.DCTSize * 3]);
                tmp0 += z1 * (-REDUCED_FIX_1_272758580); /* sqrt(2) * (-c1+c3-c5-c7) */
                z1 = REDUCED_DEQUANTIZE(coef_block[coefBlockIndex + JpegConstants.DCTSize * 1],
                    quantTable[quantTableIndex + JpegConstants.DCTSize * 1]);
                tmp0 += z1 * REDUCED_FIX_3_624509785; /* sqrt(2) * (c1+c3+c5+c7) */

                /* Final output stage */

                workspace[workspaceIndex + JpegConstants.DCTSize * 0] = JpegUtils.DESCALE(tmp10 + tmp0, REDUCED_CONST_BITS - REDUCED_PASS1_BITS + 2);
                workspace[workspaceIndex + JpegConstants.DCTSize * 1] = JpegUtils.DESCALE(tmp10 - tmp0, REDUCED_CONST_BITS - REDUCED_PASS1_BITS + 2);
            }

            /* Pass 2: process 2 rows from work array, store into output array. */
            workspaceIndex = 0;
            byte[] limit = m_cinfo.m_sample_range_limit;
            int limitOffset = m_cinfo.m_sampleRangeLimitOffset + JpegConstants.MediumSampleValue;

            for (int ctr = 0; ctr < 2; ctr++)
            {
                int currentOutRow = output_row + ctr;
                /* It's not clear whether a zero row test is worthwhile here ... */

                if (workspace[workspaceIndex + 1] == 0 &&
                    workspace[workspaceIndex + 3] == 0 &&
                    workspace[workspaceIndex + 5] == 0 &&
                    workspace[workspaceIndex + 7] == 0)
                {
                    /* AC terms all zero */
                    byte dcval = limit[limitOffset + JpegUtils.DESCALE(workspace[workspaceIndex + 0], REDUCED_PASS1_BITS + 3) & RANGE_MASK];

                    m_componentBuffer[currentOutRow][output_col + 0] = dcval;
                    m_componentBuffer[currentOutRow][output_col + 1] = dcval;

                    workspaceIndex += JpegConstants.DCTSize;       /* advance pointer to next row */
                    continue;
                }

                /* Even part */

                int tmp10 = (workspace[workspaceIndex + 0]) << (REDUCED_CONST_BITS + 2);

                /* Odd part */

                int tmp0 = workspace[workspaceIndex + 7] * (-REDUCED_FIX_0_720959822) /* sqrt(2) * (c7-c5+c3-c1) */ +
                       workspace[workspaceIndex + 5] * REDUCED_FIX_0_850430095 /* sqrt(2) * (-c1+c3+c5+c7) */ +
                       workspace[workspaceIndex + 3] * (-REDUCED_FIX_1_272758580) /* sqrt(2) * (-c1+c3-c5-c7) */ +
                       workspace[workspaceIndex + 1] * REDUCED_FIX_3_624509785; /* sqrt(2) * (c1+c3+c5+c7) */

                /* Final output stage */

                m_componentBuffer[currentOutRow][output_col + 0] = limit[limitOffset + JpegUtils.DESCALE(tmp10 + tmp0, REDUCED_CONST_BITS + REDUCED_PASS1_BITS + 3 + 2) & RANGE_MASK];
                m_componentBuffer[currentOutRow][output_col + 1] = limit[limitOffset + JpegUtils.DESCALE(tmp10 - tmp0, REDUCED_CONST_BITS + REDUCED_PASS1_BITS + 3 + 2) & RANGE_MASK];

                workspaceIndex += JpegConstants.DCTSize;       /* advance pointer to next row */
            }
        }

        /// <summary>
        /// Perform dequantization and inverse DCT on one block of coefficients,
        /// producing a reduced-size 1x1 output block.
        /// </summary>
        private void jpeg_idct_1x1(int component_index, short[] coef_block, int output_row, int output_col)
        {
            /* We hardly need an inverse DCT routine for this: just take the
            * average pixel value, which is one-eighth of the DC coefficient.
            */
            int[] quantptr = m_dctTables[component_index].int_array;
            int dcval = REDUCED_DEQUANTIZE(coef_block[0], quantptr[0]);
            dcval = JpegUtils.DESCALE(dcval, 3);

            byte[] limit = m_cinfo.m_sample_range_limit;
            int limitOffset = m_cinfo.m_sampleRangeLimitOffset + JpegConstants.MediumSampleValue;

            m_componentBuffer[output_row + 0][output_col] = limit[limitOffset + dcval & RANGE_MASK];
        }

        /// <summary>
        /// Dequantize a coefficient by multiplying it by the multiplier-table
        /// entry; produce an int result.  In this module, both inputs and result
        /// are 16 bits or less, so either int or short multiply will work.
        /// </summary>
        private static int REDUCED_DEQUANTIZE(short coef, int quantval)
        {
            return ((int)coef * quantval);
        }
    }
    #endregion

    #region JpegMarker
    /// <summary>
    /// Representation of special JPEG marker.
    /// </summary>
    /// <remarks>You can't create instance of this class manually.
    /// Concrete objects are instantiated by library and you can get them
    /// through <see cref="JpegDecompressor.Marker_list">Marker_list</see> property.
    /// </remarks>
    /// <seealso cref="JpegDecompressor.Marker_list"/>
    /// <seealso href="81c88818-a5d7-4550-9ce5-024a768f7b1e.htm" target="_self">Special markers</seealso>
    public class JpegMarker
    {
        /// <summary>
        /// marker code: JPEG_COM, or JPEG_APP0+n
        /// </summary>
        private byte m_marker;
        /// <summary>
        /// # bytes of data in the file
        /// </summary>
        private int m_originalLength;
        /// <summary>
        /// the data contained in the marker
        /// </summary>
        private byte[] m_data;

        internal JpegMarker(byte marker, int originalDataLength, int lengthLimit)
        {
            m_marker = marker;
            m_originalLength = originalDataLength;
            m_data = new byte[lengthLimit];
        }

        /// <summary>
        /// Gets the special marker.
        /// </summary>
        /// <value>The marker value.</value>
        public byte Marker
        {
            get
            {
                return m_marker;
            }
        }

        /// <summary>
        /// Gets the full length of original data associated with the marker.
        /// </summary>
        /// <value>The length of original data associated with the marker.</value>
        /// <remarks>This length excludes the marker length word, whereas the stored representation 
        /// within the JPEG file includes it. (Hence the maximum data length is really only 65533.)
        /// </remarks>
        public int OriginalLength
        {
            get
            {
                return m_originalLength;
            }
        }

        /// <summary>
        /// Gets the data associated with the marker.
        /// </summary>
        /// <value>The data associated with the marker.</value>
        /// <remarks>The length of this array doesn't exceed <c>length_limit</c> for the particular marker type.
        /// Note that this length excludes the marker length word, whereas the stored representation 
        /// within the JPEG file includes it. (Hence the maximum data length is really only 65533.)
        /// </remarks>
        public byte[] Data
        {
            get
            {
                return m_data;
            }
        }
    }
    #endregion

    #region JpegMarkerReader
    /// <summary>
    /// Marker reading and parsing
    /// </summary>
    class JpegMarkerReader
    {
        private const int APP0_DATA_LEN = 14;  /* Length of interesting data in APP0 */
        private const int APP14_DATA_LEN = 12;  /* Length of interesting data in APP14 */
        private const int APPN_DATA_LEN = 14;  /* Must be the largest of the above!! */

        private JpegDecompressor m_cinfo;

        /* Application-overridable marker processing methods */
        private JpegDecompressor.jpeg_marker_parser_method m_process_COM;
        private JpegDecompressor.jpeg_marker_parser_method[] m_process_APPn = new JpegDecompressor.jpeg_marker_parser_method[16];

        /* Limit on marker data length to save for each marker type */
        private int m_length_limit_COM;
        private int[] m_length_limit_APPn = new int[16];

        private bool m_saw_SOI;       /* found SOI? */
        private bool m_saw_SOF;       /* found SOF? */
        private int m_next_restart_num;       /* next restart number expected (0-7) */
        private int m_discarded_bytes;   /* # of bytes skipped looking for a marker */

        /* Status of COM/APPn marker saving */
        private JpegMarker m_cur_marker; /* null if not processing a marker */
        private int m_bytes_read;        /* data bytes read so far in marker */
        /* Note: cur_marker is not linked into marker_list until it's all read. */

        /// <summary>
        /// Initialize the marker reader module.
        /// This is called only once, when the decompression object is created.
        /// </summary>
        public JpegMarkerReader(JpegDecompressor cinfo)
        {
            m_cinfo = cinfo;

            /* Initialize COM/APPn processing.
            * By default, we examine and then discard APP0 and APP14,
            * but simply discard COM and all other APPn.
            */
            m_process_COM = skip_variable;

            for (int i = 0; i < 16; i++)
            {
                m_process_APPn[i] = skip_variable;
                m_length_limit_APPn[i] = 0;
            }

            m_process_APPn[0] = get_interesting_appn;
            m_process_APPn[14] = get_interesting_appn;

            /* Reset marker processing state */
            reset_marker_reader();
        }

        /// <summary>
        /// Reset marker processing state to begin a fresh datastream.
        /// </summary>
        public void reset_marker_reader()
        {
            m_cinfo.Comp_info = null;        /* until allocated by get_sof */
            m_cinfo.m_input_scan_number = 0;       /* no SOS seen yet */
            m_cinfo.m_unread_marker = 0;       /* no pending marker */
            m_saw_SOI = false;        /* set internal state too */
            m_saw_SOF = false;
            m_discarded_bytes = 0;
            m_cur_marker = null;
        }

        /// <summary>
        /// Read markers until SOS or EOI.
        /// 
        /// Returns same codes as are defined for jpeg_consume_input:
        /// JPEG_SUSPENDED, JPEG_REACHED_SOS, or JPEG_REACHED_EOI.
        /// </summary>
        public ReadResult read_markers()
        {
            /* Outer loop repeats once for each marker. */
            for (; ; )
            {
                /* Collect the marker proper, unless we already did. */
                /* NB: first_marker() enforces the requirement that SOI appear first. */
                if (m_cinfo.m_unread_marker == 0)
                {
                    if (!m_cinfo.m_marker.m_saw_SOI)
                    {
                        if (!first_marker())
                            return ReadResult.Suspended;
                    }
                    else
                    {
                        if (!next_marker())
                            return ReadResult.Suspended;
                    }
                }

                /* At this point m_cinfo.unread_marker contains the marker code and the
                 * input point is just past the marker proper, but before any parameters.
                 * A suspension will cause us to return with this state still true.
                 */
                switch ((JpegMarkerType)m_cinfo.m_unread_marker)
                {
                    case JpegMarkerType.SOI:
                        if (!get_soi())
                            return ReadResult.Suspended;
                        break;

                    case JpegMarkerType.SOF0:
                    /* Baseline */
                    case JpegMarkerType.SOF1:
                        /* Extended sequential, Huffman */
                        if (!get_sof(false))
                            return ReadResult.Suspended;
                        break;

                    case JpegMarkerType.SOF2:
                        /* Progressive, Huffman */
                        if (!get_sof(true))
                            return ReadResult.Suspended;
                        break;

                    /* Currently unsupported SOFn types */
                    case JpegMarkerType.SOF3:
                    /* Lossless, Huffman */
                    case JpegMarkerType.SOF5:
                    /* Differential sequential, Huffman */
                    case JpegMarkerType.SOF6:
                    /* Differential progressive, Huffman */
                    case JpegMarkerType.SOF7:
                    /* Differential lossless, Huffman */
                    case JpegMarkerType.SOF9:
                    /* Extended sequential, arithmetic */
                    case JpegMarkerType.SOF10:
                    /* Progressive, arithmetic */
                    case JpegMarkerType.JPG:
                    /* Reserved for JPEG extensions */
                    case JpegMarkerType.SOF11:
                    /* Lossless, arithmetic */
                    case JpegMarkerType.SOF13:
                    /* Differential sequential, arithmetic */
                    case JpegMarkerType.SOF14:
                    /* Differential progressive, arithmetic */
                    case JpegMarkerType.SOF15:
                        /* Differential lossless, arithmetic */
                        throw new Exception(String.Format("Unsupported JPEG process: SOF type 0x{0:X2}", m_cinfo.m_unread_marker));

                    case JpegMarkerType.SOS:
                        if (!get_sos())
                            return ReadResult.Suspended;
                        m_cinfo.m_unread_marker = 0;   /* processed the marker */
                        return ReadResult.Reached_SOS;

                    case JpegMarkerType.EOI:
                        m_cinfo.m_unread_marker = 0;   /* processed the marker */
                        return ReadResult.Reached_EOI;

                    case JpegMarkerType.DAC:
                        if (!skip_variable(m_cinfo))
                            return ReadResult.Suspended;
                        break;

                    case JpegMarkerType.DHT:
                        if (!get_dht())
                            return ReadResult.Suspended;
                        break;

                    case JpegMarkerType.DQT:
                        if (!get_dqt())
                            return ReadResult.Suspended;
                        break;

                    case JpegMarkerType.DRI:
                        if (!get_dri())
                            return ReadResult.Suspended;
                        break;

                    case JpegMarkerType.APP0:
                    case JpegMarkerType.APP1:
                    case JpegMarkerType.APP2:
                    case JpegMarkerType.APP3:
                    case JpegMarkerType.APP4:
                    case JpegMarkerType.APP5:
                    case JpegMarkerType.APP6:
                    case JpegMarkerType.APP7:
                    case JpegMarkerType.APP8:
                    case JpegMarkerType.APP9:
                    case JpegMarkerType.APP10:
                    case JpegMarkerType.APP11:
                    case JpegMarkerType.APP12:
                    case JpegMarkerType.APP13:
                    case JpegMarkerType.APP14:
                    case JpegMarkerType.APP15:
                        if (!m_cinfo.m_marker.m_process_APPn[m_cinfo.m_unread_marker - (int)JpegMarkerType.APP0](m_cinfo))
                            return ReadResult.Suspended;
                        break;

                    case JpegMarkerType.COM:
                        if (!m_cinfo.m_marker.m_process_COM(m_cinfo))
                            return ReadResult.Suspended;
                        break;

                    /* these are all parameter-less */
                    case JpegMarkerType.RST0:
                    case JpegMarkerType.RST1:
                    case JpegMarkerType.RST2:
                    case JpegMarkerType.RST3:
                    case JpegMarkerType.RST4:
                    case JpegMarkerType.RST5:
                    case JpegMarkerType.RST6:
                    case JpegMarkerType.RST7:
                    case JpegMarkerType.TEM:
                        break;

                    case JpegMarkerType.DNL:
                        /* Ignore DNL ... perhaps the wrong thing */
                        if (!skip_variable(m_cinfo))
                            return ReadResult.Suspended;
                        break;

                    default:
                        /* must be DHP, EXP, JPGn, or RESn */
                        /* For now, we treat the reserved markers as fatal errors since they are
                         * likely to be used to signal incompatible JPEG Part 3 extensions.
                         * Once the JPEG 3 version-number marker is well defined, this code
                         * ought to change!
                         */
                        throw new Exception(String.Format("Unsupported marker type 0x{0:X2}", m_cinfo.m_unread_marker));
                }

                /* Successfully processed marker, so reset state variable */
                m_cinfo.m_unread_marker = 0;
            } /* end loop */
        }

        /// <summary>
        /// Read a restart marker, which is expected to appear next in the data-stream;
        /// if the marker is not there, take appropriate recovery action.
        /// Returns false if suspension is required.
        /// 
        /// Made public for use by entropy decoder only
        /// 
        /// This is called by the entropy decoder after it has read an appropriate
        /// number of MCUs.  cinfo.unread_marker may be nonzero if the entropy decoder
        /// has already read a marker from the data source.  Under normal conditions
        /// cinfo.unread_marker will be reset to 0 before returning; if not reset,
        /// it holds a marker which the decoder will be unable to read past.
        /// </summary>
        public bool read_restart_marker()
        {
            /* Obtain a marker unless we already did. */
            /* Note that next_marker will complain if it skips any data. */
            if (m_cinfo.m_unread_marker == 0)
            {
                if (!next_marker())
                    return false;
            }

            if (m_cinfo.m_unread_marker == ((int)JpegMarkerType.RST0 + m_cinfo.m_marker.m_next_restart_num))
            {
                /* Normal case --- swallow the marker and let entropy decoder continue */
                m_cinfo.m_unread_marker = 0;
            }
            else
            {
                /* Uh-oh, the restart markers have been messed up. */
                /* Let the data source manager determine how to re-sync. */
                if (!m_cinfo.m_src.resync_to_restart(m_cinfo, m_cinfo.m_marker.m_next_restart_num))
                    return false;
            }

            /* Update next-restart state */
            m_cinfo.m_marker.m_next_restart_num = (m_cinfo.m_marker.m_next_restart_num + 1) & 7;

            return true;
        }

        /// <summary>
        /// Find the next JPEG marker, save it in cinfo.unread_marker.
        /// Returns false if had to suspend before reaching a marker;
        /// in that case cinfo.unread_marker is unchanged.
        /// 
        /// Note that the result might not be a valid marker code,
        /// but it will never be 0 or FF.
        /// </summary>
        public bool next_marker()
        {
            int c;
            for (; ; )
            {
                if (!m_cinfo.m_src.GetByte(out c))
                    return false;

                /* Skip any non-FF bytes.
                 * This may look a bit inefficient, but it will not occur in a valid file.
                 * We sync after each discarded byte so that a suspending data source
                 * can discard the byte from its buffer.
                 */
                while (c != 0xFF)
                {
                    m_cinfo.m_marker.m_discarded_bytes++;
                    if (!m_cinfo.m_src.GetByte(out c))
                        return false;
                }

                /* This loop swallows any duplicate FF bytes.  Extra FFs are legal as
                 * pad bytes, so don't count them in discarded_bytes.  We assume there
                 * will not be so many consecutive FF bytes as to overflow a suspending
                 * data source's input buffer.
                 */
                do
                {
                    if (!m_cinfo.m_src.GetByte(out c))
                        return false;
                }
                while (c == 0xFF);

                if (c != 0)
                {
                    /* found a valid marker, exit loop */
                    break;
                }

                /* Reach here if we found a stuffed-zero data sequence (FF/00).
                 * Discard it and loop back to try again.
                 */
                m_cinfo.m_marker.m_discarded_bytes += 2;
            }

            if (m_cinfo.m_marker.m_discarded_bytes != 0)
            {
                m_cinfo.m_marker.m_discarded_bytes = 0;
            }

            m_cinfo.m_unread_marker = c;
            return true;
        }

        /// <summary>
        /// Install a special processing method for COM or APPn markers.
        /// </summary>
        public void jpeg_set_marker_processor(int marker_code, JpegDecompressor.jpeg_marker_parser_method routine)
        {
            if (marker_code == (int)JpegMarkerType.COM)
                m_process_COM = routine;
            else if (marker_code >= (int)JpegMarkerType.APP0 && marker_code <= (int)JpegMarkerType.APP15)
                m_process_APPn[marker_code - (int)JpegMarkerType.APP0] = routine;
            else
                throw new Exception(String.Format("Unsupported marker type 0x{0:X2}", marker_code));
        }

        public void jpeg_save_markers(int marker_code, int length_limit)
        {
            /* Choose processor routine to use.
             * APP0/APP14 have special requirements.
             */
            JpegDecompressor.jpeg_marker_parser_method processor;
            if (length_limit != 0)
            {
                processor = save_marker;
                /* If saving APP0/APP14, save at least enough for our internal use. */
                if (marker_code == (int)JpegMarkerType.APP0 && length_limit < APP0_DATA_LEN)
                    length_limit = APP0_DATA_LEN;
                else if (marker_code == (int)JpegMarkerType.APP14 && length_limit < APP14_DATA_LEN)
                    length_limit = APP14_DATA_LEN;
            }
            else
            {
                processor = skip_variable;
                /* If discarding APP0/APP14, use our regular on-the-fly processor. */
                if (marker_code == (int)JpegMarkerType.APP0 || marker_code == (int)JpegMarkerType.APP14)
                    processor = get_interesting_appn;
            }

            if (marker_code == (int)JpegMarkerType.COM)
            {
                m_process_COM = processor;
                m_length_limit_COM = length_limit;
            }
            else if (marker_code >= (int)JpegMarkerType.APP0 && marker_code <= (int)JpegMarkerType.APP15)
            {
                m_process_APPn[marker_code - (int)JpegMarkerType.APP0] = processor;
                m_length_limit_APPn[marker_code - (int)JpegMarkerType.APP0] = length_limit;
            }
            else
                throw new Exception(String.Format("Unsupported marker type 0x{0:X2}", marker_code));
        }

        /* State of marker reader, applications
        * supplying COM or APPn handlers might like to know the state.
        */
        public bool SawSOI()
        {
            return m_saw_SOI;
        }

        public bool SawSOF()
        {
            return m_saw_SOF;
        }

        public int NextRestartNumber()
        {
            return m_next_restart_num;
        }

        public int DiscardedByteCount()
        {
            return m_discarded_bytes;
        }

        public void SkipBytes(int count)
        {
            m_discarded_bytes += count;
        }

        /// <summary>
        /// Save an APPn or COM marker into the marker list
        /// </summary>
        private static bool save_marker(JpegDecompressor cinfo)
        {
            JpegMarker cur_marker = cinfo.m_marker.m_cur_marker;

            byte[] data = null;
            int length = 0;
            int bytes_read;
            int data_length;
            int dataOffset = 0;

            if (cur_marker == null)
            {
                /* begin reading a marker */
                if (!cinfo.m_src.GetTwoBytes(out length))
                    return false;

                length -= 2;
                if (length >= 0)
                {
                    /* watch out for bogus length word */
                    /* figure out how much we want to save */
                    int limit;
                    if (cinfo.m_unread_marker == (int)JpegMarkerType.COM)
                        limit = cinfo.m_marker.m_length_limit_COM;
                    else
                        limit = cinfo.m_marker.m_length_limit_APPn[cinfo.m_unread_marker - (int)JpegMarkerType.APP0];

                    if (length < limit)
                        limit = length;

                    /* allocate and initialize the marker item */
                    cur_marker = new JpegMarker((byte)cinfo.m_unread_marker, length, limit);

                    /* data area is just beyond the JpegMarker */
                    data = cur_marker.Data;
                    cinfo.m_marker.m_cur_marker = cur_marker;
                    cinfo.m_marker.m_bytes_read = 0;
                    bytes_read = 0;
                    data_length = limit;
                }
                else
                {
                    /* deal with bogus length word */
                    bytes_read = data_length = 0;
                    data = null;
                }
            }
            else
            {
                /* resume reading a marker */
                bytes_read = cinfo.m_marker.m_bytes_read;
                data_length = cur_marker.Data.Length;
                data = cur_marker.Data;
                dataOffset = bytes_read;
            }

            byte[] tempData = null;
            if (data_length != 0)
                tempData = new byte[data.Length];

            while (bytes_read < data_length)
            {
                /* move the restart point to here */
                cinfo.m_marker.m_bytes_read = bytes_read;

                /* If there's not at least one byte in buffer, suspend */
                if (!cinfo.m_src.MakeByteAvailable())
                    return false;

                /* Copy bytes with reasonable rapidity */
                int read = cinfo.m_src.GetBytes(tempData, data_length - bytes_read);
                Buffer.BlockCopy(tempData, 0, data, dataOffset, data_length - bytes_read);
                bytes_read += read;
            }

            /* Done reading what we want to read */
            if (cur_marker != null)
            {
                /* will be null if bogus length word */
                /* Add new marker to end of list */
                cinfo.m_marker_list.Add(cur_marker);

                /* Reset pointer & calc remaining data length */
                data = cur_marker.Data;
                dataOffset = 0;
                length = cur_marker.OriginalLength - data_length;
            }

            /* Reset to initial state for next marker */
            cinfo.m_marker.m_cur_marker = null;

            JpegMarkerType currentMarker = (JpegMarkerType)cinfo.m_unread_marker;
            if (data_length != 0 && (currentMarker == JpegMarkerType.APP0 || currentMarker == JpegMarkerType.APP14))
            {
                tempData = new byte[data.Length];
                Buffer.BlockCopy(data, dataOffset, tempData, 0, data.Length - dataOffset);
            }

            /* Process the marker if interesting; else just make a generic trace msg */
            switch ((JpegMarkerType)cinfo.m_unread_marker)
            {
                case JpegMarkerType.APP0:
                    examine_app0(cinfo, tempData, data_length, length);
                    break;
                case JpegMarkerType.APP14:
                    examine_app14(cinfo, tempData, data_length, length);
                    break;
                default:
                    break;
            }

            /* skip any remaining data -- could be lots */
            if (length > 0)
                cinfo.m_src.skip_input_data(length);

            return true;
        }

        /// <summary>
        /// Skip over an unknown or uninteresting variable-length marker
        /// </summary>
        private static bool skip_variable(JpegDecompressor cinfo)
        {
            int length;
            if (!cinfo.m_src.GetTwoBytes(out length))
                return false;

            length -= 2;

            if (length > 0)
                cinfo.m_src.skip_input_data(length);

            return true;
        }

        /// <summary>
        /// Process an APP0 or APP14 marker without saving it
        /// </summary>
        private static bool get_interesting_appn(JpegDecompressor cinfo)
        {
            int length;
            if (!cinfo.m_src.GetTwoBytes(out length))
                return false;

            length -= 2;

            /* get the interesting part of the marker data */
            int numtoread = 0;
            if (length >= APPN_DATA_LEN)
                numtoread = APPN_DATA_LEN;
            else if (length > 0)
                numtoread = length;

            byte[] b = new byte[APPN_DATA_LEN];
            for (int i = 0; i < numtoread; i++)
            {
                int temp = 0;
                if (!cinfo.m_src.GetByte(out temp))
                    return false;

                b[i] = (byte)temp;
            }

            length -= numtoread;

            /* process it */
            switch ((JpegMarkerType)cinfo.m_unread_marker)
            {
                case JpegMarkerType.APP0:
                    examine_app0(cinfo, b, numtoread, length);
                    break;
                case JpegMarkerType.APP14:
                    examine_app14(cinfo, b, numtoread, length);
                    break;
                default:
                    /* can't get here unless jpeg_save_markers chooses wrong processor */
                    throw new Exception(String.Format("Unsupported marker type 0x{0:X2}", cinfo.m_unread_marker));
            }

            /* skip any remaining data -- could be lots */
            if (length > 0)
                cinfo.m_src.skip_input_data(length);

            return true;
        }

        /*
         * Routines for processing APPn and COM markers.
         * These are either saved in memory or discarded, per application request.
         * APP0 and APP14 are specially checked to see if they are
         * JFIF and Adobe markers, respectively.
         */

        /// <summary>
        /// Examine first few bytes from an APP0.
        /// Take appropriate action if it is a JFIF marker.
        /// datalen is # of bytes at data[], remaining is length of rest of marker data.
        /// </summary>
        private static void examine_app0(JpegDecompressor cinfo, byte[] data, int datalen, int remaining)
        {
            int totallen = datalen + remaining;

            if (datalen >= APP0_DATA_LEN &&
                data[0] == 0x4A &&
                data[1] == 0x46 &&
                data[2] == 0x49 &&
                data[3] == 0x46 &&
                data[4] == 0)
            {
                /* Found JFIF APP0 marker: save info */
                cinfo.m_saw_JFIF_marker = true;
                cinfo.m_JFIF_major_version = data[5];
                cinfo.m_JFIF_minor_version = data[6];
                cinfo.m_density_unit = (DensityUnit)data[7];
                cinfo.m_X_density = (short)((data[8] << 8) + data[9]);
                cinfo.m_Y_density = (short)((data[10] << 8) + data[11]);
            }
            else if (datalen >= 6 && data[0] == 0x4A && data[1] == 0x46 && data[2] == 0x58 && data[3] == 0x58 && data[4] == 0)
            {
                /* Found JFIF "JFXX" extension APP0 marker */
                /* The library doesn't actually do anything with these.
                 */
            }
            else
            {
                /* Start of APP0 does not match "JFIF" or "JFXX", or too short */
            }
        }

        /// <summary>
        /// Examine first few bytes from an APP14.
        /// Take appropriate action if it is an Adobe marker.
        /// datalen is # of bytes at data[], remaining is length of rest of marker data.
        /// </summary>
        private static void examine_app14(JpegDecompressor cinfo, byte[] data, int datalen, int remaining)
        {
            if (datalen >= APP14_DATA_LEN &&
                data[0] == 0x41 &&
                data[1] == 0x64 &&
                data[2] == 0x6F &&
                data[3] == 0x62 &&
                data[4] == 0x65)
            {
                /* Found Adobe APP14 marker */
                int version = (data[5] << 8) + data[6];
                int flags0 = (data[7] << 8) + data[8];
                int flags1 = (data[9] << 8) + data[10];
                int transform = data[11];
                cinfo.m_saw_Adobe_marker = true;
                cinfo.m_Adobe_transform = (byte)transform;
            }
            else
            {
                /* Start of APP14 does not match "Adobe", or too short */
            }
        }

        /*
         * Routines to process JPEG markers.
         *
         * Entry condition: JPEG marker itself has been read and its code saved
         *   in cinfo.unread_marker; input restart point is just after the marker.
         *
         * Exit: if return true, have read and processed any parameters, and have
         *   updated the restart point to point after the parameters.
         *   If return false, was forced to suspend before reaching end of
         *   marker parameters; restart point has not been moved.  Same routine
         *   will be called again after application supplies more input data.
         *
         * This approach to suspension assumes that all of a marker's parameters
         * can fit into a single input buffer-load.  This should hold for "normal"
         * markers.  Some COM/APPn markers might have large parameter segments
         * that might not fit.  If we are simply dropping such a marker, we use
         * skip_input_data to get past it, and thereby put the problem on the
         * source manager's shoulders.  If we are saving the marker's contents
         * into memory, we use a slightly different convention: when forced to
         * suspend, the marker processor updates the restart point to the end of
         * what it's consumed (ie, the end of the buffer) before returning false.
         * On resumption, cinfo.unread_marker still contains the marker code,
         * but the data source will point to the next chunk of marker data.
         * The marker processor must retain internal state to deal with this.
         *
         * Note that we don't bother to avoid duplicate trace messages if a
         * suspension occurs within marker parameters.  Other side effects
         * require more care.
         */


        /// <summary>
        /// Process an SOI marker
        /// </summary>
        private bool get_soi()
        {
            if (m_cinfo.m_marker.m_saw_SOI)
                throw new Exception("Invalid JPEG file structure: two SOI markers");

            /* Reset all parameters that are defined to be reset by SOI */
            m_cinfo.m_restart_interval = 0;

            /* Set initial assumptions for colorspace etc */

            m_cinfo.m_jpeg_color_space = ColorSpace.Unknown;
            m_cinfo.m_CCIR601_sampling = false; /* Assume non-CCIR sampling??? */

            m_cinfo.m_saw_JFIF_marker = false;
            m_cinfo.m_JFIF_major_version = 1; /* set default JFIF APP0 values */
            m_cinfo.m_JFIF_minor_version = 1;
            m_cinfo.m_density_unit = DensityUnit.Unknown;
            m_cinfo.m_X_density = 1;
            m_cinfo.m_Y_density = 1;
            m_cinfo.m_saw_Adobe_marker = false;
            m_cinfo.m_Adobe_transform = 0;

            m_cinfo.m_marker.m_saw_SOI = true;

            return true;
        }

        /// <summary>
        /// Process a SOFn marker
        /// </summary>
        private bool get_sof(bool is_prog)
        {
            m_cinfo.m_progressive_mode = is_prog;

            int length;
            if (!m_cinfo.m_src.GetTwoBytes(out length))
                return false;

            if (!m_cinfo.m_src.GetByte(out m_cinfo.m_data_precision))
                return false;

            int temp = 0;
            if (!m_cinfo.m_src.GetTwoBytes(out temp))
                return false;
            m_cinfo.m_image_height = temp;

            if (!m_cinfo.m_src.GetTwoBytes(out temp))
                return false;
            m_cinfo.m_image_width = temp;

            if (!m_cinfo.m_src.GetByte(out m_cinfo.m_num_components))
                return false;

            length -= 8;

            if (m_cinfo.m_marker.m_saw_SOF)
                throw new Exception("Invalid JPEG file structure: two SOI markers");

            /* We don't support files in which the image height is initially specified */
            /* as 0 and is later redefined by DNL.  As long as we have to check that,  */
            /* might as well have a general sanity check. */
            if (m_cinfo.m_image_height <= 0 || m_cinfo.m_image_width <= 0 || m_cinfo.m_num_components <= 0)
                throw new Exception("Empty JPEG image (DNL not supported)");

            if (length != (m_cinfo.m_num_components * 3))
                throw new Exception("Bogus marker length");

            if (m_cinfo.Comp_info == null)
            {
                /* do only once, even if suspend */
                m_cinfo.Comp_info = JpegComponent.createArrayOfComponents(m_cinfo.m_num_components);
            }

            for (int ci = 0; ci < m_cinfo.m_num_components; ci++)
            {
                m_cinfo.Comp_info[ci].Component_index = ci;

                int component_id;
                if (!m_cinfo.m_src.GetByte(out component_id))
                    return false;

                m_cinfo.Comp_info[ci].Component_id = component_id;

                int c;
                if (!m_cinfo.m_src.GetByte(out c))
                    return false;

                m_cinfo.Comp_info[ci].H_samp_factor = (c >> 4) & 15;
                m_cinfo.Comp_info[ci].V_samp_factor = (c) & 15;

                int quant_tbl_no;
                if (!m_cinfo.m_src.GetByte(out quant_tbl_no))
                    return false;

                m_cinfo.Comp_info[ci].Quant_tbl_no = quant_tbl_no;
            }

            m_cinfo.m_marker.m_saw_SOF = true;
            return true;
        }

        /// <summary>
        /// Process a SOS marker
        /// </summary>
        private bool get_sos()
        {
            if (!m_cinfo.m_marker.m_saw_SOF)
                throw new Exception("Invalid JPEG file structure: SOS before SOF");

            int length;
            if (!m_cinfo.m_src.GetTwoBytes(out length))
                return false;

            /* Number of components */
            int n;
            if (!m_cinfo.m_src.GetByte(out n))
                return false;

            if (length != (n * 2 + 6) || n < 1 || n > JpegConstants.MaxComponentsInScan)
                throw new Exception("Bogus marker length");

            m_cinfo.m_comps_in_scan = n;

            /* Collect the component-spec parameters */

            for (int i = 0; i < n; i++)
            {
                int cc;
                if (!m_cinfo.m_src.GetByte(out cc))
                    return false;

                int c;
                if (!m_cinfo.m_src.GetByte(out c))
                    return false;

                bool idFound = false;
                int foundIndex = -1;
                for (int ci = 0; ci < m_cinfo.m_num_components; ci++)
                {
                    if (cc == m_cinfo.Comp_info[ci].Component_id)
                    {
                        foundIndex = ci;
                        idFound = true;
                        break;
                    }
                }

                if (!idFound)
                    throw new Exception(String.Format("Invalid component ID {0} in SOS", cc));

                m_cinfo.m_cur_comp_info[i] = foundIndex;
                m_cinfo.Comp_info[foundIndex].Dc_tbl_no = (c >> 4) & 15;
                m_cinfo.Comp_info[foundIndex].Ac_tbl_no = (c) & 15;
            }

            /* Collect the additional scan parameters Ss, Se, Ah/Al. */
            int temp;
            if (!m_cinfo.m_src.GetByte(out temp))
                return false;

            m_cinfo.m_Ss = temp;
            if (!m_cinfo.m_src.GetByte(out temp))
                return false;

            m_cinfo.m_Se = temp;
            if (!m_cinfo.m_src.GetByte(out temp))
                return false;

            m_cinfo.m_Ah = (temp >> 4) & 15;
            m_cinfo.m_Al = (temp) & 15;

            /* Prepare to scan data & restart markers */
            m_cinfo.m_marker.m_next_restart_num = 0;

            /* Count another SOS marker */
            m_cinfo.m_input_scan_number++;
            return true;
        }

        /// <summary>
        /// Process a DHT marker
        /// </summary>
        private bool get_dht()
        {
            int length;
            if (!m_cinfo.m_src.GetTwoBytes(out length))
                return false;

            length -= 2;

            byte[] bits = new byte[17];
            byte[] huffval = new byte[256];
            while (length > 16)
            {
                int index;
                if (!m_cinfo.m_src.GetByte(out index))
                    return false;

                bits[0] = 0;
                int count = 0;
                for (int i = 1; i <= 16; i++)
                {
                    int temp = 0;
                    if (!m_cinfo.m_src.GetByte(out temp))
                        return false;

                    bits[i] = (byte)temp;
                    count += bits[i];
                }

                length -= 1 + 16;

                /* Here we just do minimal validation of the counts to avoid walking
                 * off the end of our table space. HuffEntropyDecoder will check more carefully.
                 */
                if (count > 256 || count > length)
                    throw new Exception("Bogus Huffman table definition");

                for (int i = 0; i < count; i++)
                {
                    int temp = 0;
                    if (!m_cinfo.m_src.GetByte(out temp))
                        return false;

                    huffval[i] = (byte)temp;
                }

                length -= count;

                JpegHuffmanTable htblptr = null;
                if ((index & 0x10) != 0)
                {
                    /* AC table definition */
                    index -= 0x10;
                    if (m_cinfo.m_ac_huff_tbl_ptrs[index] == null)
                        m_cinfo.m_ac_huff_tbl_ptrs[index] = new JpegHuffmanTable();

                    htblptr = m_cinfo.m_ac_huff_tbl_ptrs[index];
                }
                else
                {
                    /* DC table definition */
                    if (m_cinfo.m_dc_huff_tbl_ptrs[index] == null)
                        m_cinfo.m_dc_huff_tbl_ptrs[index] = new JpegHuffmanTable();

                    htblptr = m_cinfo.m_dc_huff_tbl_ptrs[index];
                }

                if (index < 0 || index >= JpegConstants.NumberOfHuffmanTables)
                    throw new Exception(String.Format("Bogus DHT index {0}", index));

                Buffer.BlockCopy(bits, 0, htblptr.Bits, 0, htblptr.Bits.Length);
                Buffer.BlockCopy(huffval, 0, htblptr.Huffval, 0, htblptr.Huffval.Length);
            }

            if (length != 0)
                throw new Exception("Bogus marker length");

            return true;
        }

        /// <summary>
        /// Process a DQT marker
        /// </summary>
        private bool get_dqt()
        {
            int length;
            if (!m_cinfo.m_src.GetTwoBytes(out length))
                return false;

            length -= 2;
            while (length > 0)
            {
                int n;
                if (!m_cinfo.m_src.GetByte(out n))
                    return false;

                int prec = n >> 4;
                n &= 0x0F;

                if (n >= JpegConstants.NumberOfQuantTables)
                    throw new Exception(String.Format("Bogus DQT index {0}", n));

                if (m_cinfo.m_quant_tbl_ptrs[n] == null)
                    m_cinfo.m_quant_tbl_ptrs[n] = new JpegQuantizationTable();

                JpegQuantizationTable quant_ptr = m_cinfo.m_quant_tbl_ptrs[n];

                for (int i = 0; i < JpegConstants.DCTSize2; i++)
                {
                    int tmp;
                    if (prec != 0)
                    {
                        int temp = 0;
                        if (!m_cinfo.m_src.GetTwoBytes(out temp))
                            return false;

                        tmp = temp;
                    }
                    else
                    {
                        int temp = 0;
                        if (!m_cinfo.m_src.GetByte(out temp))
                            return false;

                        tmp = temp;
                    }

                    /* We convert the zigzag-order table to natural array order. */
                    quant_ptr.quantval[JpegUtils.jpeg_natural_order[i]] = (short)tmp;
                }

                length -= JpegConstants.DCTSize2 + 1;
                if (prec != 0)
                    length -= JpegConstants.DCTSize2;
            }

            if (length != 0)
                throw new Exception("Bogus marker length");

            return true;
        }

        /// <summary>
        /// Process a DRI marker
        /// </summary>
        private bool get_dri()
        {
            int length;
            if (!m_cinfo.m_src.GetTwoBytes(out length))
                return false;

            if (length != 4)
                throw new Exception("Bogus marker length");

            int temp = 0;
            if (!m_cinfo.m_src.GetTwoBytes(out temp))
                return false;

            int tmp = temp;
            m_cinfo.m_restart_interval = tmp;

            return true;
        }

        /// <summary>
        /// Like next_marker, but used to obtain the initial SOI marker.
        /// For this marker, we do not allow preceding garbage or fill; otherwise,
        /// we might well scan an entire input file before realizing it's not JPEG.
        /// If an application wants to process non-JFIF files, it must seek to the
        /// SOI before calling the JPEG library.
        /// </summary>
        private bool first_marker()
        {
            int c;
            if (!m_cinfo.m_src.GetByte(out c))
                return false;

            int c2;
            if (!m_cinfo.m_src.GetByte(out c2))
                return false;

            if (c != 0xFF || c2 != (int)JpegMarkerType.SOI)
                throw new Exception(String.Format("Not a JPEG file: starts with 0x{0:X2} 0x{1:X2}", c, c2));

            m_cinfo.m_unread_marker = c2;
            return true;
        }
    }
    #endregion

    #region JpegMarkerType
    /// <summary>
    /// JPEG marker codes.
    /// </summary>
    /// <seealso href="81c88818-a5d7-4550-9ce5-024a768f7b1e.htm" target="_self">Special markers</seealso>
    public enum JpegMarkerType
    {
        /// <summary>
        /// 
        /// </summary>
        SOF0 = 0xc0,
        /// <summary>
        /// 
        /// </summary>
        SOF1 = 0xc1,
        /// <summary>
        /// 
        /// </summary>
        SOF2 = 0xc2,
        /// <summary>
        /// 
        /// </summary>
        SOF3 = 0xc3,
        /// <summary>
        /// 
        /// </summary>
        SOF5 = 0xc5,
        /// <summary>
        /// 
        /// </summary>
        SOF6 = 0xc6,
        /// <summary>
        /// 
        /// </summary>
        SOF7 = 0xc7,
        /// <summary>
        /// 
        /// </summary>
        JPG = 0xc8,
        /// <summary>
        /// 
        /// </summary>
        SOF9 = 0xc9,
        /// <summary>
        /// 
        /// </summary>
        SOF10 = 0xca,
        /// <summary>
        /// 
        /// </summary>
        SOF11 = 0xcb,
        /// <summary>
        /// 
        /// </summary>
        SOF13 = 0xcd,
        /// <summary>
        /// 
        /// </summary>
        SOF14 = 0xce,
        /// <summary>
        /// 
        /// </summary>
        SOF15 = 0xcf,
        /// <summary>
        /// 
        /// </summary>
        DHT = 0xc4,
        /// <summary>
        /// 
        /// </summary>
        DAC = 0xcc,
        /// <summary>
        /// 
        /// </summary>
        RST0 = 0xd0,
        /// <summary>
        /// 
        /// </summary>
        RST1 = 0xd1,
        /// <summary>
        /// 
        /// </summary>
        RST2 = 0xd2,
        /// <summary>
        /// 
        /// </summary>
        RST3 = 0xd3,
        /// <summary>
        /// 
        /// </summary>
        RST4 = 0xd4,
        /// <summary>
        /// 
        /// </summary>
        RST5 = 0xd5,
        /// <summary>
        /// 
        /// </summary>
        RST6 = 0xd6,
        /// <summary>
        /// 
        /// </summary>
        RST7 = 0xd7,
        /// <summary>
        /// 
        /// </summary>
        SOI = 0xd8,
        /// <summary>
        /// 
        /// </summary>
        EOI = 0xd9,
        /// <summary>
        /// 
        /// </summary>
        SOS = 0xda,
        /// <summary>
        /// 
        /// </summary>
        DQT = 0xdb,
        /// <summary>
        /// 
        /// </summary>
        DNL = 0xdc,
        /// <summary>
        /// 
        /// </summary>
        DRI = 0xdd,
        /// <summary>
        /// 
        /// </summary>
        DHP = 0xde,
        /// <summary>
        /// 
        /// </summary>
        EXP = 0xdf,
        /// <summary>
        /// 
        /// </summary>
        APP0 = 0xe0,
        /// <summary>
        /// 
        /// </summary>
        APP1 = 0xe1,
        /// <summary>
        /// 
        /// </summary>
        APP2 = 0xe2,
        /// <summary>
        /// 
        /// </summary>
        APP3 = 0xe3,
        /// <summary>
        /// 
        /// </summary>
        APP4 = 0xe4,
        /// <summary>
        /// 
        /// </summary>
        APP5 = 0xe5,
        /// <summary>
        /// 
        /// </summary>
        APP6 = 0xe6,
        /// <summary>
        /// 
        /// </summary>
        APP7 = 0xe7,
        /// <summary>
        /// 
        /// </summary>
        APP8 = 0xe8,
        /// <summary>
        /// 
        /// </summary>
        APP9 = 0xe9,
        /// <summary>
        /// 
        /// </summary>
        APP10 = 0xea,
        /// <summary>
        /// 
        /// </summary>
        APP11 = 0xeb,
        /// <summary>
        /// 
        /// </summary>
        APP12 = 0xec,
        /// <summary>
        /// 
        /// </summary>
        APP13 = 0xed,
        /// <summary>
        /// 
        /// </summary>
        APP14 = 0xee,
        /// <summary>
        /// 
        /// </summary>
        APP15 = 0xef,
        /// <summary>
        /// 
        /// </summary>
        JPG0 = 0xf0,
        /// <summary>
        /// 
        /// </summary>
        JPG13 = 0xfd,
        /// <summary>
        /// 
        /// </summary>
        COM = 0xfe,
        /// <summary>
        /// 
        /// </summary>
        TEM = 0x01,
        /// <summary>
        /// 
        /// </summary>
        ERROR = 0x100
    }
    #endregion

    #region JpegMarkerWriter
    /// <summary>
    /// Marker writing
    /// </summary>
    class JpegMarkerWriter
    {
        private JpegCompressor m_cinfo;
        private int m_last_restart_interval; /* last DRI value emitted; 0 after SOI */

        public JpegMarkerWriter(JpegCompressor cinfo)
        {
            m_cinfo = cinfo;
        }

        /// <summary>
        /// Write datastream header.
        /// This consists of an SOI and optional APPn markers.
        /// We recommend use of the JFIF marker, but not the Adobe marker,
        /// when using YCbCr or grayscale data.  The JFIF marker should NOT
        /// be used for any other JPEG colorspace.  The Adobe marker is helpful
        /// to distinguish RGB, CMYK, and YCCK colorspaces.
        /// Note that an application can write additional header markers after
        /// jpeg_start_compress returns.
        /// </summary>
        public void write_file_header()
        {
            emit_marker(JpegMarkerType.SOI);  /* first the SOI */

            /* SOI is defined to reset restart interval to 0 */
            m_last_restart_interval = 0;

            if (m_cinfo.m_write_JFIF_header)   /* next an optional JFIF APP0 */
                emit_jfif_app0();
            if (m_cinfo.m_write_Adobe_marker) /* next an optional Adobe APP14 */
                emit_adobe_app14();
        }

        /// <summary>
        /// Write frame header.
        /// This consists of DQT and SOFn markers.
        /// Note that we do not emit the SOF until we have emitted the DQT(s).
        /// This avoids compatibility problems with incorrect implementations that
        /// try to error-check the quant table numbers as soon as they see the SOF.
        /// </summary>
        public void write_frame_header()
        {
            /* Emit DQT for each quantization table.
             * Note that emit_dqt() suppresses any duplicate tables.
             */
            int prec = 0;
            for (int ci = 0; ci < m_cinfo.m_num_components; ci++)
                prec += emit_dqt(m_cinfo.Component_info[ci].Quant_tbl_no);

            /* now prec is nonzero iff there are any 16-bit quant tables. */

            /* Check for a non-baseline specification.
             * Note we assume that Huffman table numbers won't be changed later.
             */
            bool is_baseline;
            if (m_cinfo.m_progressive_mode || m_cinfo.m_data_precision != 8)
            {
                is_baseline = false;
            }
            else
            {
                is_baseline = true;
                for (int ci = 0; ci < m_cinfo.m_num_components; ci++)
                {
                    if (m_cinfo.Component_info[ci].Dc_tbl_no > 1 || m_cinfo.Component_info[ci].Ac_tbl_no > 1)
                        is_baseline = false;
                }

                if (prec != 0 && is_baseline)
                {
                    is_baseline = false;
                }
            }

            /* Emit the proper SOF marker */
            if (m_cinfo.m_progressive_mode)
                emit_sof(JpegMarkerType.SOF2);    /* SOF code for progressive Huffman */
            else if (is_baseline)
                emit_sof(JpegMarkerType.SOF0);    /* SOF code for baseline implementation */
            else
                emit_sof(JpegMarkerType.SOF1);    /* SOF code for non-baseline Huffman file */
        }

        /// <summary>
        /// Write scan header.
        /// This consists of DHT or DAC markers, optional DRI, and SOS.
        /// Compressed data will be written following the SOS.
        /// </summary>
        public void write_scan_header()
        {
            /* Emit Huffman tables.
             * Note that emit_dht() suppresses any duplicate tables.
             */
            for (int i = 0; i < m_cinfo.m_comps_in_scan; i++)
            {
                int ac_tbl_no = m_cinfo.Component_info[m_cinfo.m_cur_comp_info[i]].Ac_tbl_no;
                int dc_tbl_no = m_cinfo.Component_info[m_cinfo.m_cur_comp_info[i]].Dc_tbl_no;
                if (m_cinfo.m_progressive_mode)
                {
                    /* Progressive mode: only DC or only AC tables are used in one scan */
                    if (m_cinfo.m_Ss == 0)
                    {
                        if (m_cinfo.m_Ah == 0)
                        {
                            /* DC needs no table for refinement scan */
                            emit_dht(dc_tbl_no, false);
                        }
                    }
                    else
                    {
                        emit_dht(ac_tbl_no, true);
                    }
                }
                else
                {
                    /* Sequential mode: need both DC and AC tables */
                    emit_dht(dc_tbl_no, false);
                    emit_dht(ac_tbl_no, true);
                }
            }

            /* Emit DRI if required --- note that DRI value could change for each scan.
             * We avoid wasting space with unnecessary DRIs, however.
             */
            if (m_cinfo.m_restart_interval != m_last_restart_interval)
            {
                emit_dri();
                m_last_restart_interval = m_cinfo.m_restart_interval;
            }

            emit_sos();
        }

        /// <summary>
        /// Write datastream trailer.
        /// </summary>
        public void write_file_trailer()
        {
            emit_marker(JpegMarkerType.EOI);
        }

        /// <summary>
        /// Write an abbreviated table-specification datastream.
        /// This consists of SOI, DQT and DHT tables, and EOI.
        /// Any table that is defined and not marked sent_table = true will be
        /// emitted.  Note that all tables will be marked sent_table = true at exit.
        /// </summary>
        public void write_tables_only()
        {
            emit_marker(JpegMarkerType.SOI);

            for (int i = 0; i < JpegConstants.NumberOfQuantTables; i++)
            {
                if (m_cinfo.m_quant_tbl_ptrs[i] != null)
                    emit_dqt(i);
            }

            for (int i = 0; i < JpegConstants.NumberOfHuffmanTables; i++)
            {
                if (m_cinfo.m_dc_huff_tbl_ptrs[i] != null)
                    emit_dht(i, false);
                if (m_cinfo.m_ac_huff_tbl_ptrs[i] != null)
                    emit_dht(i, true);
            }

            emit_marker(JpegMarkerType.EOI);
        }

        //////////////////////////////////////////////////////////////////////////
        // These routines allow writing an arbitrary marker with parameters.
        // The only intended use is to emit COM or APPn markers after calling
        // write_file_header and before calling write_frame_header.
        // Other uses are not guaranteed to produce desirable results.
        // Counting the parameter bytes properly is the caller's responsibility.

        /// <summary>
        /// Emit an arbitrary marker header
        /// </summary>
        public void write_marker_header(int marker, int datalen)
        {
            if (datalen > 65533)     /* safety check */
                throw new Exception("Bogus marker length");

            emit_marker((JpegMarkerType)marker);

            emit_2bytes(datalen + 2);    /* total length */
        }

        /// <summary>
        /// Emit one byte of marker parameters following write_marker_header
        /// </summary>
        public void write_marker_byte(byte val)
        {
            emit_byte(val);
        }

        //////////////////////////////////////////////////////////////////////////
        // Routines to write specific marker types.
        //

        /// <summary>
        /// Emit a SOS marker
        /// </summary>
        private void emit_sos()
        {
            emit_marker(JpegMarkerType.SOS);

            emit_2bytes(2 * m_cinfo.m_comps_in_scan + 2 + 1 + 3); /* length */

            emit_byte(m_cinfo.m_comps_in_scan);

            for (int i = 0; i < m_cinfo.m_comps_in_scan; i++)
            {
                int componentIndex = m_cinfo.m_cur_comp_info[i];
                emit_byte(m_cinfo.Component_info[componentIndex].Component_id);

                int td = m_cinfo.Component_info[componentIndex].Dc_tbl_no;
                int ta = m_cinfo.Component_info[componentIndex].Ac_tbl_no;
                if (m_cinfo.m_progressive_mode)
                {
                    /* Progressive mode: only DC or only AC tables are used in one scan;
                     * furthermore, Huffman coding of DC refinement uses no table at all.
                     * We emit 0 for unused field(s); this is recommended by the P&M text
                     * but does not seem to be specified in the standard.
                     */
                    if (m_cinfo.m_Ss == 0)
                    {
                        /* DC scan */
                        ta = 0;
                        if (m_cinfo.m_Ah != 0)
                        {
                            /* no DC table either */
                            td = 0;
                        }
                    }
                    else
                    {
                        /* AC scan */
                        td = 0;
                    }
                }

                emit_byte((td << 4) + ta);
            }

            emit_byte(m_cinfo.m_Ss);
            emit_byte(m_cinfo.m_Se);
            emit_byte((m_cinfo.m_Ah << 4) + m_cinfo.m_Al);
        }

        /// <summary>
        /// Emit a SOF marker
        /// </summary>
        private void emit_sof(JpegMarkerType code)
        {
            emit_marker(code);

            emit_2bytes(3 * m_cinfo.m_num_components + 2 + 5 + 1); /* length */

            /* Make sure image isn't bigger than SOF field can handle */
            if (m_cinfo.m_image_height > 65535 || m_cinfo.m_image_width > 65535)
                throw new Exception(String.Format("Maximum supported image dimension is {0} pixels", 65535));

            emit_byte(m_cinfo.m_data_precision);
            emit_2bytes(m_cinfo.m_image_height);
            emit_2bytes(m_cinfo.m_image_width);

            emit_byte(m_cinfo.m_num_components);

            for (int ci = 0; ci < m_cinfo.m_num_components; ci++)
            {
                JpegComponent componentInfo = m_cinfo.Component_info[ci];
                emit_byte(componentInfo.Component_id);
                emit_byte((componentInfo.H_samp_factor << 4) + componentInfo.V_samp_factor);
                emit_byte(componentInfo.Quant_tbl_no);
            }
        }

        /// <summary>
        /// Emit an Adobe APP14 marker
        /// </summary>
        private void emit_adobe_app14()
        {
            /*
             * Length of APP14 block    (2 bytes)
             * Block ID         (5 bytes - ASCII "Adobe")
             * Version Number       (2 bytes - currently 100)
             * Flags0           (2 bytes - currently 0)
             * Flags1           (2 bytes - currently 0)
             * Color transform      (1 byte)
             *
             * Although Adobe TN 5116 mentions Version = 101, all the Adobe files
             * now in circulation seem to use Version = 100, so that's what we write.
             *
             * We write the color transform byte as 1 if the JPEG color space is
             * YCbCr, 2 if it's YCCK, 0 otherwise.  Adobe's definition has to do with
             * whether the encoder performed a transformation, which is pretty useless.
             */

            emit_marker(JpegMarkerType.APP14);

            emit_2bytes(2 + 5 + 2 + 2 + 2 + 1); /* length */

            emit_byte(0x41); /* Identifier: ASCII "Adobe" */
            emit_byte(0x64);
            emit_byte(0x6F);
            emit_byte(0x62);
            emit_byte(0x65);
            emit_2bytes(100);    /* Version */
            emit_2bytes(0);  /* Flags0 */
            emit_2bytes(0);  /* Flags1 */
            switch (m_cinfo.m_jpeg_color_space)
            {
                case ColorSpace.YCbCr:
                    emit_byte(1);    /* Color transform = 1 */
                    break;
                case ColorSpace.YCCK:
                    emit_byte(2);    /* Color transform = 2 */
                    break;
                default:
                    emit_byte(0);    /* Color transform = 0 */
                    break;
            }
        }

        /// <summary>
        /// Emit a DRI marker
        /// </summary>
        private void emit_dri()
        {
            emit_marker(JpegMarkerType.DRI);

            emit_2bytes(4);  /* fixed length */

            emit_2bytes(m_cinfo.m_restart_interval);
        }

        /// <summary>
        /// Emit a DHT marker
        /// </summary>
        private void emit_dht(int index, bool is_ac)
        {
            JpegHuffmanTable htbl = m_cinfo.m_dc_huff_tbl_ptrs[index];
            if (is_ac)
            {
                htbl = m_cinfo.m_ac_huff_tbl_ptrs[index];
                index += 0x10; /* output index has AC bit set */
            }

            if (htbl == null)
                throw new Exception(String.Format("Huffman table 0x{0:X2} was not defined", index));

            if (!htbl.Sent_table)
            {
                emit_marker(JpegMarkerType.DHT);

                int length = 0;
                for (int i = 1; i <= 16; i++)
                    length += htbl.Bits[i];

                emit_2bytes(length + 2 + 1 + 16);
                emit_byte(index);

                for (int i = 1; i <= 16; i++)
                    emit_byte(htbl.Bits[i]);

                for (int i = 0; i < length; i++)
                    emit_byte(htbl.Huffval[i]);

                htbl.Sent_table = true;
            }
        }

        /// <summary>
        /// Emit a DQT marker
        /// </summary>
        /// <param name="index">The index.</param>
        /// <returns>the precision used (0 = 8bits, 1 = 16bits) for baseline checking</returns>
        private int emit_dqt(int index)
        {
            JpegQuantizationTable qtbl = m_cinfo.m_quant_tbl_ptrs[index];
            if (qtbl == null)
                throw new Exception(String.Format("Quantization table 0x{0:X2} was not defined", index));

            int prec = 0;
            for (int i = 0; i < JpegConstants.DCTSize2; i++)
            {
                if (qtbl.quantval[i] > 255)
                    prec = 1;
            }

            if (!qtbl.Sent_table)
            {
                emit_marker(JpegMarkerType.DQT);

                emit_2bytes(prec != 0 ? JpegConstants.DCTSize2 * 2 + 1 + 2 : JpegConstants.DCTSize2 + 1 + 2);

                emit_byte(index + (prec << 4));

                for (int i = 0; i < JpegConstants.DCTSize2; i++)
                {
                    /* The table entries must be emitted in zigzag order. */
                    int qval = qtbl.quantval[JpegUtils.jpeg_natural_order[i]];

                    if (prec != 0)
                        emit_byte(qval >> 8);

                    emit_byte(qval & 0xFF);
                }

                qtbl.Sent_table = true;
            }

            return prec;
        }

        /// <summary>
        /// Emit a JFIF-compliant APP0 marker
        /// </summary>
        private void emit_jfif_app0()
        {
            /*
             * Length of APP0 block (2 bytes)
             * Block ID         (4 bytes - ASCII "JFIF")
             * Zero byte            (1 byte to terminate the ID string)
             * Version Major, Minor (2 bytes - major first)
             * Units            (1 byte - 0x00 = none, 0x01 = inch, 0x02 = cm)
             * Xdpu         (2 bytes - dots per unit horizontal)
             * Ydpu         (2 bytes - dots per unit vertical)
             * Thumbnail X size     (1 byte)
             * Thumbnail Y size     (1 byte)
             */

            emit_marker(JpegMarkerType.APP0);

            emit_2bytes(2 + 4 + 1 + 2 + 1 + 2 + 2 + 1 + 1); /* length */

            emit_byte(0x4A); /* Identifier: ASCII "JFIF" */
            emit_byte(0x46);
            emit_byte(0x49);
            emit_byte(0x46);
            emit_byte(0);
            emit_byte(m_cinfo.m_JFIF_major_version); /* Version fields */
            emit_byte(m_cinfo.m_JFIF_minor_version);
            emit_byte((int)m_cinfo.m_density_unit); /* Pixel size information */
            emit_2bytes(m_cinfo.m_X_density);
            emit_2bytes(m_cinfo.m_Y_density);
            emit_byte(0);        /* No thumbnail image */
            emit_byte(0);
        }

        //////////////////////////////////////////////////////////////////////////
        // Basic output routines.
        // 
        // Note that we do not support suspension while writing a marker.
        // Therefore, an application using suspension must ensure that there is
        // enough buffer space for the initial markers (typ. 600-700 bytes) before
        // calling jpeg_start_compress, and enough space to write the trailing EOI
        // (a few bytes) before calling jpeg_finish_compress.  Multi-pass compression
        // modes are not supported at all with suspension, so those two are the only
        // points where markers will be written.


        /// <summary>
        /// Emit a marker code
        /// </summary>
        private void emit_marker(JpegMarkerType mark)
        {
            emit_byte(0xFF);
            emit_byte((int)mark);
        }

        /// <summary>
        /// Emit a 2-byte integer; these are always MSB first in JPEG files
        /// </summary>
        private void emit_2bytes(int value)
        {
            emit_byte((value >> 8) & 0xFF);
            emit_byte(value & 0xFF);
        }

        /// <summary>
        /// Emit a byte
        /// </summary>
        private void emit_byte(int val)
        {
            if (!m_cinfo.m_dest.emit_byte(val))
                throw new Exception("Suspension not allowed here");
        }
    }
    #endregion

    #region JpegQuantizationTable
    /// <summary>
    /// DCT coefficient quantization table.
    /// </summary>
    public class JpegQuantizationTable
    {
        /// <summary>
        /// This field is used only during compression.  It's initialized false when
        /// the table is created, and set true when it's been output to the file.
        /// You could suppress output of a table by setting this to true.
        /// </summary>
        private bool m_sent_table;

        /// <summary>
        /// This array gives the coefficient quantizers in natural array order
        /// (not the zigzag order in which they are stored in a JPEG DQT marker).
        /// CAUTION: IJG versions prior to v6a kept this array in zigzag order.
        /// </summary>
        internal readonly short[] quantval = new short[JpegConstants.DCTSize2];

        internal JpegQuantizationTable()
        {
        }

        /// <summary>
        /// Gets or sets a value indicating whether the table has been output to file.
        /// </summary>
        /// <value>It's initialized <c>false</c> when the table is created, and set 
        /// <c>true</c> when it's been output to the file. You could suppress output of a table by setting this to <c>true</c>.
        /// </value>
        /// <remarks>This property is used only during compression.</remarks>
        /// <seealso cref="JpegCompressor.jpeg_suppress_tables"/>
        public bool Sent_table
        {
            get { return m_sent_table; }
            set { m_sent_table = value; }
        }
    }
    #endregion

    #region JpegScanInfo
    /// <summary>
    /// The script for encoding a multiple-scan file is an array of these:
    /// </summary>
    class JpegScanInfo
    {
        public int comps_in_scan;      /* number of components encoded in this scan */
        public int[] component_index = new int[JpegConstants.MaxComponentsInScan]; /* their SOF/comp_info[] indexes */
        public int Ss;
        public int Se;         /* progressive JPEG spectral selection parms */
        public int Ah;
        public int Al;         /* progressive JPEG successive approx. parms */
    }
    #endregion

    #region JpegSource
    /// <summary>
    /// Data source object for decompression.
    /// </summary>
    public abstract class Jpeg_Source
    {
        private byte[] m_next_input_byte;
        private int m_bytes_in_buffer; /* # of bytes remaining (unread) in buffer */
        private int m_position;

        /// <summary>
        /// Initializes this instance.
        /// </summary>
        public abstract void init_source();

        /// <summary>
        /// Fills input buffer
        /// </summary>
        /// <returns><c>true</c> if operation succeed; otherwise, <c>false</c></returns>
        public abstract bool fill_input_buffer();

        /// <summary>
        /// Initializes the internal buffer.
        /// </summary>
        /// <param name="buffer">The buffer.</param>
        /// <param name="size">The size.</param>
        protected void initInternalBuffer(byte[] buffer, int size)
        {
            m_bytes_in_buffer = size;
            m_next_input_byte = buffer;
            m_position = 0;
        }

        /// <summary>
        /// Skip data - used to skip over a potentially large amount of
        /// uninteresting data (such as an APPn marker).
        /// </summary>
        /// <param name="num_bytes">The number of bytes to skip.</param>
        /// <remarks>Writers of suspendable-input applications must note that skip_input_data
        /// is not granted the right to give a suspension return.  If the skip extends
        /// beyond the data currently in the buffer, the buffer can be marked empty so
        /// that the next read will cause a fill_input_buffer call that can suspend.
        /// Arranging for additional bytes to be discarded before reloading the input
        /// buffer is the application writer's problem.</remarks>
        public virtual void skip_input_data(int num_bytes)
        {
            /* Just a dumb implementation for now.  Could use fseek() except
            * it doesn't work on pipes.  Not clear that being smart is worth
            * any trouble anyway --- large skips are infrequent.
            */
            if (num_bytes > 0)
            {
                while (num_bytes > m_bytes_in_buffer)
                {
                    num_bytes -= m_bytes_in_buffer;
                    fill_input_buffer();
                    /* note we assume that fill_input_buffer will never return false,
                    * so suspension need not be handled.
                    */
                }

                m_position += num_bytes;
                m_bytes_in_buffer -= num_bytes;
            }
        }

        /// <summary>
        /// This is the default resync_to_restart method for data source 
        /// managers to use if they don't have any better approach.
        /// </summary>
        /// <param name="cinfo">An instance of <see cref="JpegDecompressor"/></param>
        /// <param name="desired">The desired</param>
        /// <returns><c>false</c> if suspension is required.</returns>
        /// <remarks>That method assumes that no backtracking is possible. 
        /// Some data source managers may be able to back up, or may have 
        /// additional knowledge about the data which permits a more 
        /// intelligent recovery strategy; such managers would
        /// presumably supply their own resync method.<br/><br/>
        /// 
        /// read_restart_marker calls resync_to_restart if it finds a marker other than
        /// the restart marker it was expecting.  (This code is *not* used unless
        /// a nonzero restart interval has been declared.)  cinfo.unread_marker is
        /// the marker code actually found (might be anything, except 0 or FF).
        /// The desired restart marker number (0..7) is passed as a parameter.<br/><br/>
        /// 
        /// This routine is supposed to apply whatever error recovery strategy seems
        /// appropriate in order to position the input stream to the next data segment.
        /// Note that cinfo.unread_marker is treated as a marker appearing before
        /// the current data-source input point; usually it should be reset to zero
        /// before returning.<br/><br/>
        /// 
        /// This implementation is substantially constrained by wanting to treat the
        /// input as a data stream; this means we can't back up.  Therefore, we have
        /// only the following actions to work with:<br/>
        /// 1. Simply discard the marker and let the entropy decoder resume at next
        /// byte of file.<br/>
        /// 2. Read forward until we find another marker, discarding intervening
        /// data.  (In theory we could look ahead within the current bufferload,
        /// without having to discard data if we don't find the desired marker.
        /// This idea is not implemented here, in part because it makes behavior
        /// dependent on buffer size and chance buffer-boundary positions.)<br/>
        /// 3. Leave the marker unread (by failing to zero cinfo.unread_marker).
        /// This will cause the entropy decoder to process an empty data segment,
        /// inserting dummy zeroes, and then we will reprocess the marker.<br/>
        /// 
        /// #2 is appropriate if we think the desired marker lies ahead, while #3 is
        /// appropriate if the found marker is a future restart marker (indicating
        /// that we have missed the desired restart marker, probably because it got
        /// corrupted).<br/>
        /// We apply #2 or #3 if the found marker is a restart marker no more than
        /// two counts behind or ahead of the expected one.  We also apply #2 if the
        /// found marker is not a legal JPEG marker code (it's certainly bogus data).
        /// If the found marker is a restart marker more than 2 counts away, we do #1
        /// (too much risk that the marker is erroneous; with luck we will be able to
        /// resync at some future point).<br/>
        /// For any valid non-restart JPEG marker, we apply #3.  This keeps us from
        /// overrunning the end of a scan.  An implementation limited to single-scan
        /// files might find it better to apply #2 for markers other than EOI, since
        /// any other marker would have to be bogus data in that case.</remarks>
        public virtual bool resync_to_restart(JpegDecompressor cinfo, int desired)
        {
            /* Outer loop handles repeated decision after scanning forward. */
            int action = 1;
            for (; ; )
            {
                if (cinfo.m_unread_marker < (int)JpegMarkerType.SOF0)
                {
                    /* invalid marker */
                    action = 2;
                }
                else if (cinfo.m_unread_marker < (int)JpegMarkerType.RST0 ||
                    cinfo.m_unread_marker > (int)JpegMarkerType.RST7)
                {
                    /* valid non-restart marker */
                    action = 3;
                }
                else
                {
                    if (cinfo.m_unread_marker == ((int)JpegMarkerType.RST0 + ((desired + 1) & 7))
                        || cinfo.m_unread_marker == ((int)JpegMarkerType.RST0 + ((desired + 2) & 7)))
                    {
                        /* one of the next two expected restarts */
                        action = 3;
                    }
                    else if (cinfo.m_unread_marker == ((int)JpegMarkerType.RST0 + ((desired - 1) & 7)) ||
                        cinfo.m_unread_marker == ((int)JpegMarkerType.RST0 + ((desired - 2) & 7)))
                    {
                        /* a prior restart, so advance */
                        action = 2;
                    }
                    else
                    {
                        /* desired restart or too far away */
                        action = 1;
                    }
                }

                switch (action)
                {
                    case 1:
                        /* Discard marker and let entropy decoder resume processing. */
                        cinfo.m_unread_marker = 0;
                        return true;
                    case 2:
                        /* Scan to the next marker, and repeat the decision loop. */
                        if (!cinfo.m_marker.next_marker())
                            return false;
                        break;
                    case 3:
                        /* Return without advancing past this marker. */
                        /* Entropy decoder will be forced to process an empty segment. */
                        return true;
                }
            }
        }

        /// <summary>
        /// Terminate source - called by jpeg_finish_decompress
        /// after all data has been read.  Often a no-op.
        /// </summary>
        /// <remarks>NB: <b>not</b> called by jpeg_abort or jpeg_destroy; surrounding
        /// application must deal with any cleanup that should happen even
        /// for error exit.</remarks>
        public virtual void term_source()
        {
        }

        /// <summary>
        /// Reads two bytes interpreted as an unsigned 16-bit integer.
        /// </summary>
        /// <param name="V">The result.</param>
        /// <returns><c>true</c> if operation succeed; otherwise, <c>false</c></returns>
        public virtual bool GetTwoBytes(out int V)
        {
            if (!MakeByteAvailable())
            {
                V = 0;
                return false;
            }

            m_bytes_in_buffer--;
            V = m_next_input_byte[m_position] << 8;
            m_position++;

            if (!MakeByteAvailable())
                return false;

            m_bytes_in_buffer--;
            V += m_next_input_byte[m_position];
            m_position++;
            return true;
        }

        /// <summary>
        /// Read a byte into variable V.
        /// If must suspend, take the specified action (typically "return false").
        /// </summary>
        /// <param name="V">The result.</param>
        /// <returns><c>true</c> if operation succeed; otherwise, <c>false</c></returns>
        public virtual bool GetByte(out int V)
        {
            if (!MakeByteAvailable())
            {
                V = 0;
                return false;
            }

            m_bytes_in_buffer--;
            V = m_next_input_byte[m_position];
            m_position++;
            return true;
        }

        /// <summary>
        /// Gets the bytes.
        /// </summary>
        /// <param name="dest">The destination.</param>
        /// <param name="amount">The amount.</param>
        /// <returns>The number of available bytes.</returns>
        public virtual int GetBytes(byte[] dest, int amount)
        {
            int avail = amount;
            if (avail > m_bytes_in_buffer)
                avail = m_bytes_in_buffer;

            for (int i = 0; i < avail; i++)
            {
                dest[i] = m_next_input_byte[m_position];
                m_position++;
                m_bytes_in_buffer--;
            }

            return avail;
        }

        /// <summary>
        /// Functions for fetching data from the data source module.
        /// </summary>
        /// <returns><c>true</c> if operation succeed; otherwise, <c>false</c></returns>
        /// <remarks>At all times, cinfo.src.next_input_byte and .bytes_in_buffer reflect
        /// the current restart point; we update them only when we have reached a
        /// suitable place to restart if a suspension occurs.</remarks>
        public virtual bool MakeByteAvailable()
        {
            if (m_bytes_in_buffer == 0)
            {
                if (!fill_input_buffer())
                    return false;
            }

            return true;
        }
    }
    #endregion

    #region JpegUpsampler
    /// <summary>
    /// Upsampling (note that upsampler must also call color converter)
    /// </summary>
    abstract class JpegUpsampler
    {
        protected bool m_need_context_rows; /* true if need rows above & below */

        public abstract void start_pass();
        public abstract void upsample(ComponentBuffer[] input_buf, ref int in_row_group_ctr, int in_row_groups_avail, byte[][] output_buf, ref int out_row_ctr, int out_rows_avail);

        public bool NeedContextRows()
        {
            return m_need_context_rows;
        }
    }
    #endregion

    #region JpegUtils
    class JpegUtils
    {
        /*
        * jpeg_natural_order[i] is the natural-order position of the i'th element
        * of zigzag order.
        *
        * When reading corrupted data, the Huffman decoders could attempt
        * to reference an entry beyond the end of this array (if the decoded
        * zero run length reaches past the end of the block).  To prevent
        * wild stores without adding an inner-loop test, we put some extra
        * "63"s after the real entries.  This will cause the extra coefficient
        * to be stored in location 63 of the block, not somewhere random.
        * The worst case would be a run-length of 15, which means we need 16
        * fake entries.
        */
        public static int[] jpeg_natural_order = 
        { 
            0, 1, 8, 16, 9, 2, 3, 10, 17, 24, 32, 25, 18, 11, 4, 5, 12,
            19, 26, 33, 40, 48, 41, 34, 27, 20, 13, 6, 7, 14, 21, 28, 35,
            42, 49, 56, 57, 50, 43, 36, 29, 22, 15, 23, 30, 37, 44, 51,
            58, 59, 52, 45, 38, 31, 39, 46, 53, 60, 61, 54, 47, 55, 62,
            63, 63, 63, 63, 63, 63, 63, 63, 63,
            /* extra entries for safety in decoder */
            63, 63, 63, 63, 63, 63, 63, 63 
        };

        /* We assume that right shift corresponds to signed division by 2 with
        * rounding towards minus infinity.  This is correct for typical "arithmetic
        * shift" instructions that shift in copies of the sign bit.
        * RIGHT_SHIFT provides a proper signed right shift of an int quantity.
        * It is only applied with constant shift counts.  SHIFT_TEMPS must be
        * included in the variables of any routine using RIGHT_SHIFT.
        */
        public static int RIGHT_SHIFT(int x, int shft)
        {
            return (x >> shft);
        }

        /* Descale and correctly round an int value that's scaled by N bits.
        * We assume RIGHT_SHIFT rounds towards minus infinity, so adding
        * the fudge factor is correct for either sign of X.
        */
        public static int DESCALE(int x, int n)
        {
            return RIGHT_SHIFT(x + (1 << (n - 1)), n);
        }

        //////////////////////////////////////////////////////////////////////////
        // Arithmetic utilities

        /// <summary>
        /// Compute a/b rounded up to next integer, ie, ceil(a/b)
        /// Assumes a >= 0, b > 0
        /// </summary>
        public static int jdiv_round_up(int a, int b)
        {
            return (a + b - 1) / b;
        }

        /// <summary>
        /// Compute a rounded up to next multiple of b, ie, ceil(a/b)*b
        /// Assumes a >= 0, b > 0
        /// </summary>
        public static int jround_up(int a, int b)
        {
            a += b - 1;
            return a - (a % b);
        }

        /// <summary>
        /// Copy some rows of samples from one place to another.
        /// num_rows rows are copied from input_array[source_row++]
        /// to output_array[dest_row++]; these areas may overlap for duplication.
        /// The source and destination arrays must be at least as wide as num_cols.
        /// </summary>
        public static void jcopy_sample_rows(ComponentBuffer input_array, int source_row, byte[][] output_array, int dest_row, int num_rows, int num_cols)
        {
            for (int row = 0; row < num_rows; row++)
                Buffer.BlockCopy(input_array[source_row + row], 0, output_array[dest_row + row], 0, num_cols);
        }

        public static void jcopy_sample_rows(ComponentBuffer input_array, int source_row, ComponentBuffer output_array, int dest_row, int num_rows, int num_cols)
        {
            for (int row = 0; row < num_rows; row++)
                Buffer.BlockCopy(input_array[source_row + row], 0, output_array[dest_row + row], 0, num_cols);
        }

        public static void jcopy_sample_rows(byte[][] input_array, int source_row, byte[][] output_array, int dest_row, int num_rows, int num_cols)
        {
            for (int row = 0; row < num_rows; row++)
                Buffer.BlockCopy(input_array[source_row++], 0, output_array[dest_row++], 0, num_cols);
        }
    }
    #endregion

    #region JpegVirtualArray
    /// <summary>
    /// JPEG virtual array.
    /// </summary>
    /// <typeparam name="T">The type of array's elements.</typeparam>
    /// <remarks>You can't create virtual array manually. For creation use methods
    /// <see cref="JpegCommonBase.CreateSamplesArray"/> and
    /// <see cref="JpegCommonBase.CreateBlocksArray"/>.
    /// </remarks>
    public class JpegVirtualArray<T>
    {
        internal delegate T[][] Allocator(int width, int height);

        private JpegCommonBase m_cinfo;

        /// <summary>
        /// the in-memory buffer
        /// </summary>
        private T[][] m_buffer;

        /// <summary>
        /// Request a virtual 2-D array
        /// </summary>
        /// <param name="width">Width of array</param>
        /// <param name="height">Total virtual array height</param>
        /// <param name="allocator">The allocator.</param>
        internal JpegVirtualArray(int width, int height, Allocator allocator)
        {
            m_cinfo = null;
            m_buffer = allocator(width, height);

            if (m_buffer == null)
                throw new Exception("Filling of 'm_buffer' Failed!");
        }

        /// <summary>
        /// Gets or sets the error processor.
        /// </summary>
        /// <value>The error processor.<br/>
        /// Default value: <c>null</c>
        /// </value>
        /// <remarks>Uses only for calling 
        /// <see cref="M:BitMiracle.LibJpeg.Classic.JpegCommonBase.ERREXIT(BitMiracle.LibJpeg.Classic.J_MESSAGE_CODE)">JpegCommonBase.ERREXIT</see>
        /// on error.</remarks>
        public JpegCommonBase ErrorProcessor
        {
            get { return m_cinfo; }
            set { m_cinfo = value; }
        }

        /// <summary>
        /// Access the part of a virtual array.
        /// </summary>
        /// <param name="startRow">The first row in required block.</param>
        /// <param name="numberOfRows">The number of required rows.</param>
        /// <returns>The required part of virtual array.</returns>
        public T[][] Access(int startRow, int numberOfRows)
        {
            /* debugging check */
            if (startRow + numberOfRows > m_buffer.Length)
            {
                throw new InvalidOperationException("Bogus virtual array access");
            }

            /* Return proper part of the buffer */
            T[][] ret = new T[numberOfRows][];
            for (int i = 0; i < numberOfRows; i++)
                ret[i] = m_buffer[startRow + i];

            return ret;
        }
    }
    #endregion

    #region MergedUpsampler
    class MergedUpsampler : JpegUpsampler
    {
        private const int SCALEBITS = 16;  /* speediest right-shift on some machines */
        private const int ONE_HALF = 1 << (SCALEBITS - 1);

        private JpegDecompressor m_cinfo;

        private bool m_use_2v_upsample;

        /* Private state for YCC->RGB conversion */
        private int[] m_Cr_r_tab;      /* => table for Cr to R conversion */
        private int[] m_Cb_b_tab;      /* => table for Cb to B conversion */
        private int[] m_Cr_g_tab;        /* => table for Cr to G conversion */
        private int[] m_Cb_g_tab;        /* => table for Cb to G conversion */

        /* For 2:1 vertical sampling, we produce two output rows at a time.
        * We need a "spare" row buffer to hold the second output row if the
        * application provides just a one-row buffer; we also use the spare
        * to discard the dummy last row if the image height is odd.
        */
        private byte[] m_spare_row;
        private bool m_spare_full;        /* T if spare buffer is occupied */

        private int m_out_row_width;   /* samples per output row */
        private int m_rows_to_go;  /* counts rows remaining in image */

        public MergedUpsampler(JpegDecompressor cinfo)
        {
            m_cinfo = cinfo;
            m_need_context_rows = false;

            m_out_row_width = cinfo.m_output_width * cinfo.m_out_color_components;

            if (cinfo.m_max_v_samp_factor == 2)
            {
                m_use_2v_upsample = true;
                /* Allocate a spare row buffer */
                m_spare_row = new byte[m_out_row_width];
            }
            else
            {
                m_use_2v_upsample = false;
            }

            build_ycc_rgb_table();
        }

        /// <summary>
        /// Initialize for an upsampling pass.
        /// </summary>
        public override void start_pass()
        {
            /* Mark the spare buffer empty */
            m_spare_full = false;

            /* Initialize total-height counter for detecting bottom of image */
            m_rows_to_go = m_cinfo.m_output_height;
        }

        public override void upsample(ComponentBuffer[] input_buf, ref int in_row_group_ctr, int in_row_groups_avail, byte[][] output_buf, ref int out_row_ctr, int out_rows_avail)
        {
            if (m_use_2v_upsample)
                merged_2v_upsample(input_buf, ref in_row_group_ctr, output_buf, ref out_row_ctr, out_rows_avail);
            else
                merged_1v_upsample(input_buf, ref in_row_group_ctr, output_buf, ref out_row_ctr);
        }

        /// <summary>
        /// Control routine to do upsampling (and color conversion).
        /// The control routine just handles the row buffering considerations.
        /// 1:1 vertical sampling case: much easier, never need a spare row.
        /// </summary>
        private void merged_1v_upsample(ComponentBuffer[] input_buf, ref int in_row_group_ctr, byte[][] output_buf, ref int out_row_ctr)
        {
            /* Just do the upsampling. */
            h2v1_merged_upsample(input_buf, in_row_group_ctr, output_buf, out_row_ctr);

            /* Adjust counts */
            out_row_ctr++;
            in_row_group_ctr++;
        }

        /// <summary>
        /// Control routine to do upsampling (and color conversion).
        /// The control routine just handles the row buffering considerations.
        /// 2:1 vertical sampling case: may need a spare row.
        /// </summary>
        private void merged_2v_upsample(ComponentBuffer[] input_buf, ref int in_row_group_ctr, byte[][] output_buf, ref int out_row_ctr, int out_rows_avail)
        {
            int num_rows;        /* number of rows returned to caller */
            if (m_spare_full)
            {
                /* If we have a spare row saved from a previous cycle, just return it. */
                byte[][] temp = new byte[1][];
                temp[0] = m_spare_row;
                JpegUtils.jcopy_sample_rows(temp, 0, output_buf, out_row_ctr, 1, m_out_row_width);
                num_rows = 1;
                m_spare_full = false;
            }
            else
            {
                /* Figure number of rows to return to caller. */
                num_rows = 2;

                /* Not more than the distance to the end of the image. */
                if (num_rows > m_rows_to_go)
                    num_rows = m_rows_to_go;

                /* And not more than what the client can accept: */
                out_rows_avail -= out_row_ctr;
                if (num_rows > out_rows_avail)
                    num_rows = out_rows_avail;

                /* Create output pointer array for upsampler. */
                byte[][] work_ptrs = new byte[2][];
                work_ptrs[0] = output_buf[out_row_ctr];
                if (num_rows > 1)
                {
                    work_ptrs[1] = output_buf[out_row_ctr + 1];
                }
                else
                {
                    work_ptrs[1] = m_spare_row;
                    m_spare_full = true;
                }

                /* Now do the upsampling. */
                h2v2_merged_upsample(input_buf, in_row_group_ctr, work_ptrs);
            }

            /* Adjust counts */
            out_row_ctr += num_rows;
            m_rows_to_go -= num_rows;

            /* When the buffer is emptied, declare this input row group consumed */
            if (!m_spare_full)
                in_row_group_ctr++;
        }

        /*
         * These are the routines invoked by the control routines to do
         * the actual upsampling/conversion.  One row group is processed per call.
         *
         * Note: since we may be writing directly into application-supplied buffers,
         * we have to be honest about the output width; we can't assume the buffer
         * has been rounded up to an even width.
         */

        /// <summary>
        /// Upsample and color convert for the case of 2:1 horizontal and 1:1 vertical.
        /// </summary>
        private void h2v1_merged_upsample(ComponentBuffer[] input_buf, int in_row_group_ctr, byte[][] output_buf, int outRow)
        {
            int inputIndex0 = 0;
            int inputIndex1 = 0;
            int inputIndex2 = 0;
            int outputIndex = 0;

            byte[] limit = m_cinfo.m_sample_range_limit;
            int limitOffset = m_cinfo.m_sampleRangeLimitOffset;

            /* Loop for each pair of output pixels */
            for (int col = m_cinfo.m_output_width >> 1; col > 0; col--)
            {
                /* Do the chroma part of the calculation */
                int cb = input_buf[1][in_row_group_ctr][inputIndex1];
                inputIndex1++;

                int cr = input_buf[2][in_row_group_ctr][inputIndex2];
                inputIndex2++;

                int cred = m_Cr_r_tab[cr];
                int cgreen = JpegUtils.RIGHT_SHIFT(m_Cb_g_tab[cb] + m_Cr_g_tab[cr], SCALEBITS);
                int cblue = m_Cb_b_tab[cb];

                /* Fetch 2 Y values and emit 2 pixels */
                int y = input_buf[0][in_row_group_ctr][inputIndex0];
                inputIndex0++;

                output_buf[outRow][outputIndex + JpegConstants.Offset_RGB_Red] = limit[limitOffset + y + cred];
                output_buf[outRow][outputIndex + JpegConstants.Offset_RGB_Green] = limit[limitOffset + y + cgreen];
                output_buf[outRow][outputIndex + JpegConstants.Offset_RGB_Blue] = limit[limitOffset + y + cblue];
                outputIndex += JpegConstants.RGB_PixelLength;

                y = input_buf[0][in_row_group_ctr][inputIndex0];
                inputIndex0++;

                output_buf[outRow][outputIndex + JpegConstants.Offset_RGB_Red] = limit[limitOffset + y + cred];
                output_buf[outRow][outputIndex + JpegConstants.Offset_RGB_Green] = limit[limitOffset + y + cgreen];
                output_buf[outRow][outputIndex + JpegConstants.Offset_RGB_Blue] = limit[limitOffset + y + cblue];
                outputIndex += JpegConstants.RGB_PixelLength;
            }

            /* If image width is odd, do the last output column separately */
            if ((m_cinfo.m_output_width & 1) != 0)
            {
                int cb = input_buf[1][in_row_group_ctr][inputIndex1];
                int cr = input_buf[2][in_row_group_ctr][inputIndex2];
                int cred = m_Cr_r_tab[cr];
                int cgreen = JpegUtils.RIGHT_SHIFT(m_Cb_g_tab[cb] + m_Cr_g_tab[cr], SCALEBITS);
                int cblue = m_Cb_b_tab[cb];

                int y = input_buf[0][in_row_group_ctr][inputIndex0];
                output_buf[outRow][outputIndex + JpegConstants.Offset_RGB_Red] = limit[limitOffset + y + cred];
                output_buf[outRow][outputIndex + JpegConstants.Offset_RGB_Green] = limit[limitOffset + y + cgreen];
                output_buf[outRow][outputIndex + JpegConstants.Offset_RGB_Blue] = limit[limitOffset + y + cblue];
            }
        }

        /// <summary>
        /// Upsample and color convert for the case of 2:1 horizontal and 2:1 vertical.
        /// </summary>
        private void h2v2_merged_upsample(ComponentBuffer[] input_buf, int in_row_group_ctr, byte[][] output_buf)
        {
            int inputRow00 = in_row_group_ctr * 2;
            int inputIndex00 = 0;

            int inputRow01 = in_row_group_ctr * 2 + 1;
            int inputIndex01 = 0;

            int inputIndex1 = 0;
            int inputIndex2 = 0;

            int outIndex0 = 0;
            int outIndex1 = 0;

            byte[] limit = m_cinfo.m_sample_range_limit;
            int limitOffset = m_cinfo.m_sampleRangeLimitOffset;

            /* Loop for each group of output pixels */
            for (int col = m_cinfo.m_output_width >> 1; col > 0; col--)
            {
                /* Do the chroma part of the calculation */
                int cb = input_buf[1][in_row_group_ctr][inputIndex1];
                inputIndex1++;

                int cr = input_buf[2][in_row_group_ctr][inputIndex2];
                inputIndex2++;

                int cred = m_Cr_r_tab[cr];
                int cgreen = JpegUtils.RIGHT_SHIFT(m_Cb_g_tab[cb] + m_Cr_g_tab[cr], SCALEBITS);
                int cblue = m_Cb_b_tab[cb];

                /* Fetch 4 Y values and emit 4 pixels */
                int y = input_buf[0][inputRow00][inputIndex00];
                inputIndex00++;

                output_buf[0][outIndex0 + JpegConstants.Offset_RGB_Red] = limit[limitOffset + y + cred];
                output_buf[0][outIndex0 + JpegConstants.Offset_RGB_Green] = limit[limitOffset + y + cgreen];
                output_buf[0][outIndex0 + JpegConstants.Offset_RGB_Blue] = limit[limitOffset + y + cblue];
                outIndex0 += JpegConstants.RGB_PixelLength;

                y = input_buf[0][inputRow00][inputIndex00];
                inputIndex00++;

                output_buf[0][outIndex0 + JpegConstants.Offset_RGB_Red] = limit[limitOffset + y + cred];
                output_buf[0][outIndex0 + JpegConstants.Offset_RGB_Green] = limit[limitOffset + y + cgreen];
                output_buf[0][outIndex0 + JpegConstants.Offset_RGB_Blue] = limit[limitOffset + y + cblue];
                outIndex0 += JpegConstants.RGB_PixelLength;

                y = input_buf[0][inputRow01][inputIndex01];
                inputIndex01++;

                output_buf[1][outIndex1 + JpegConstants.Offset_RGB_Red] = limit[limitOffset + y + cred];
                output_buf[1][outIndex1 + JpegConstants.Offset_RGB_Green] = limit[limitOffset + y + cgreen];
                output_buf[1][outIndex1 + JpegConstants.Offset_RGB_Blue] = limit[limitOffset + y + cblue];
                outIndex1 += JpegConstants.RGB_PixelLength;

                y = input_buf[0][inputRow01][inputIndex01];
                inputIndex01++;

                output_buf[1][outIndex1 + JpegConstants.Offset_RGB_Red] = limit[limitOffset + y + cred];
                output_buf[1][outIndex1 + JpegConstants.Offset_RGB_Green] = limit[limitOffset + y + cgreen];
                output_buf[1][outIndex1 + JpegConstants.Offset_RGB_Blue] = limit[limitOffset + y + cblue];
                outIndex1 += JpegConstants.RGB_PixelLength;
            }

            /* If image width is odd, do the last output column separately */
            if ((m_cinfo.m_output_width & 1) != 0)
            {
                int cb = input_buf[1][in_row_group_ctr][inputIndex1];
                int cr = input_buf[2][in_row_group_ctr][inputIndex2];
                int cred = m_Cr_r_tab[cr];
                int cgreen = JpegUtils.RIGHT_SHIFT(m_Cb_g_tab[cb] + m_Cr_g_tab[cr], SCALEBITS);
                int cblue = m_Cb_b_tab[cb];

                int y = input_buf[0][inputRow00][inputIndex00];
                output_buf[0][outIndex0 + JpegConstants.Offset_RGB_Red] = limit[limitOffset + y + cred];
                output_buf[0][outIndex0 + JpegConstants.Offset_RGB_Green] = limit[limitOffset + y + cgreen];
                output_buf[0][outIndex0 + JpegConstants.Offset_RGB_Blue] = limit[limitOffset + y + cblue];

                y = input_buf[0][inputRow01][inputIndex01];
                output_buf[1][outIndex1 + JpegConstants.Offset_RGB_Red] = limit[limitOffset + y + cred];
                output_buf[1][outIndex1 + JpegConstants.Offset_RGB_Green] = limit[limitOffset + y + cgreen];
                output_buf[1][outIndex1 + JpegConstants.Offset_RGB_Blue] = limit[limitOffset + y + cblue];
            }
        }

        /// <summary>
        /// Initialize tables for YCC->RGB colorspace conversion.
        /// This is taken directly from ColorDeconverter; see that file for more info.
        /// </summary>
        private void build_ycc_rgb_table()
        {
            m_Cr_r_tab = new int[JpegConstants.MaxSampleValue + 1];
            m_Cb_b_tab = new int[JpegConstants.MaxSampleValue + 1];
            m_Cr_g_tab = new int[JpegConstants.MaxSampleValue + 1];
            m_Cb_g_tab = new int[JpegConstants.MaxSampleValue + 1];

            for (int i = 0, x = -JpegConstants.MediumSampleValue; i <= JpegConstants.MaxSampleValue; i++, x++)
            {
                /* i is the actual input pixel value, in the range 0..MaxSampleValue */
                /* The Cb or Cr value we are thinking of is x = i - MediumSampleValue */
                /* Cr=>R value is nearest int to 1.40200 * x */
                m_Cr_r_tab[i] = JpegUtils.RIGHT_SHIFT(FIX(1.40200) * x + ONE_HALF, SCALEBITS);

                /* Cb=>B value is nearest int to 1.77200 * x */
                m_Cb_b_tab[i] = JpegUtils.RIGHT_SHIFT(FIX(1.77200) * x + ONE_HALF, SCALEBITS);

                /* Cr=>G value is scaled-up -0.71414 * x */
                m_Cr_g_tab[i] = (-FIX(0.71414)) * x;

                /* Cb=>G value is scaled-up -0.34414 * x */
                /* We also add in ONE_HALF so that need not do it in inner loop */
                m_Cb_g_tab[i] = (-FIX(0.34414)) * x + ONE_HALF;
            }
        }

        private static int FIX(double x)
        {
            return ((int)((x) * (1L << SCALEBITS) + 0.5));
        }
    }
    #endregion

    #region Pass1ColorQuantizer
    /// <summary>
    /// The main purpose of 1-pass quantization is to provide a fast, if not very
    /// high quality, colormapped output capability.  A 2-pass quantizer usually
    /// gives better visual quality; however, for quantized grayscale output this
    /// quantizer is perfectly adequate.  Dithering is highly recommended with this
    /// quantizer, though you can turn it off if you really want to.
    /// 
    /// In 1-pass quantization the colormap must be chosen in advance of seeing the
    /// image.  We use a map consisting of all combinations of Ncolors[i] color
    /// values for the i'th component.  The Ncolors[] values are chosen so that
    /// their product, the total number of colors, is no more than that requested.
    /// (In most cases, the product will be somewhat less.)
    /// 
    /// Since the colormap is orthogonal, the representative value for each color
    /// component can be determined without considering the other components;
    /// then these indexes can be combined into a colormap index by a standard
    /// N-dimensional-array-subscript calculation.  Most of the arithmetic involved
    /// can be precalculated and stored in the lookup table colorindex[].
    /// colorindex[i][j] maps pixel value j in component i to the nearest
    /// representative value (grid plane) for that component; this index is
    /// multiplied by the array stride for component i, so that the
    /// index of the colormap entry closest to a given pixel value is just
    ///     sum( colorindex[component-number][pixel-component-value] )
    /// Aside from being fast, this scheme allows for variable spacing between
    /// representative values with no additional lookup cost.
    /// 
    /// If gamma correction has been applied in color conversion, it might be wise
    /// to adjust the color grid spacing so that the representative colors are
    /// equidistant in linear space.  At this writing, gamma correction is not
    /// implemented, so nothing is done here.
    /// 
    /// 
    /// Declarations for Floyd-Steinberg dithering.
    /// 
    /// Errors are accumulated into the array fserrors[], at a resolution of
    /// 1/16th of a pixel count.  The error at a given pixel is propagated
    /// to its not-yet-processed neighbors using the standard F-S fractions,
    ///     ...	(here)	7/16
    ///    3/16	5/16	1/16
    /// We work left-to-right on even rows, right-to-left on odd rows.
    /// 
    /// We can get away with a single array (holding one row's worth of errors)
    /// by using it to store the current row's errors at pixel columns not yet
    /// processed, but the next row's errors at columns already processed.  We
    /// need only a few extra variables to hold the errors immediately around the
    /// current column.  (If we are lucky, those variables are in registers, but
    /// even if not, they're probably cheaper to access than array elements are.)
    /// 
    /// The fserrors[] array is indexed [component#][position].
    /// We provide (#columns + 2) entries per component; the extra entry at each
    /// end saves us from special-casing the first and last pixels.
    /// 
    /// 
    /// Declarations for ordered dithering.
    /// 
    /// We use a standard 16x16 ordered dither array.  The basic concept of ordered
    /// dithering is described in many references, for instance Dale Schumacher's
    /// chapter II.2 of Graphics Gems II (James Arvo, ed. Academic Press, 1991).
    /// In place of Schumacher's comparisons against a "threshold" value, we add a
    /// "dither" value to the input pixel and then round the result to the nearest
    /// output value.  The dither value is equivalent to (0.5 - threshold) times
    /// the distance between output values.  For ordered dithering, we assume that
    /// the output colors are equally spaced; if not, results will probably be
    /// worse, since the dither may be too much or too little at a given point.
    /// 
    /// The normal calculation would be to form pixel value + dither, range-limit
    /// this to 0..MaxSampleValue, and then index into the colorindex table as usual.
    /// We can skip the separate range-limiting step by extending the colorindex
    /// table in both directions.
    /// </summary>
    class Pass1ColorQuantizer : ColorQuantizer
    {
        private enum QuantizerType
        {
            color_quantizer3,
            color_quantizer,
            quantize3_ord_dither_quantizer,
            quantize_ord_dither_quantizer,
            quantize_fs_dither_quantizer
        }

        private static int[] RGB_order = { JpegConstants.Offset_RGB_Green, JpegConstants.Offset_RGB_Red, JpegConstants.Offset_RGB_Blue };
        private const int MAX_Q_COMPS = 4; /* max components I can handle */

        private const int ODITHER_SIZE = 16; /* dimension of dither matrix */

        /* NB: if ODITHER_SIZE is not a power of 2, ODITHER_MASK uses will break */
        private const int ODITHER_CELLS = (ODITHER_SIZE * ODITHER_SIZE); /* # cells in matrix */
        private const int ODITHER_MASK = (ODITHER_SIZE - 1); /* mask for wrapping around counters */

        /* Bayer's order-4 dither array.  Generated by the code given in
        * Stephen Hawley's article "Ordered Dithering" in Graphics Gems I.
        * The values in this array must range from 0 to ODITHER_CELLS-1.
        */
        private static byte[][] base_dither_matrix = new byte[][] 
        {
            new byte[] {   0,192, 48,240, 12,204, 60,252,  3,195, 51,243, 15,207, 63,255 },
            new byte[] { 128, 64,176,112,140, 76,188,124,131, 67,179,115,143, 79,191,127 },
            new byte[] {  32,224, 16,208, 44,236, 28,220, 35,227, 19,211, 47,239, 31,223 },
            new byte[] { 160, 96,144, 80,172,108,156, 92,163, 99,147, 83,175,111,159, 95 },
            new byte[] {   8,200, 56,248,  4,196, 52,244, 11,203, 59,251,  7,199, 55,247 },
            new byte[] { 136, 72,184,120,132, 68,180,116,139, 75,187,123,135, 71,183,119 },
            new byte[] {  40,232, 24,216, 36,228, 20,212, 43,235, 27,219, 39,231, 23,215 },
            new byte[] { 168,104,152, 88,164,100,148, 84,171,107,155, 91,167,103,151, 87 },
            new byte[] {   2,194, 50,242, 14,206, 62,254,  1,193, 49,241, 13,205, 61,253 },
            new byte[] { 130, 66,178,114,142, 78,190,126,129, 65,177,113,141, 77,189,125 },
            new byte[] {  34,226, 18,210, 46,238, 30,222, 33,225, 17,209, 45,237, 29,221 },
            new byte[] { 162, 98,146, 82,174,110,158, 94,161, 97,145, 81,173,109,157, 93 },
            new byte[] {  10,202, 58,250,  6,198, 54,246,  9,201, 57,249,  5,197, 53,245 },
            new byte[] { 138, 74,186,122,134, 70,182,118,137, 73,185,121,133, 69,181,117 },
            new byte[] {  42,234, 26,218, 38,230, 22,214, 41,233, 25,217, 37,229, 21,213 },
            new byte[] { 170,106,154, 90,166,102,150, 86,169,105,153, 89,165,101,149, 85 }
        };

        private QuantizerType m_quantizer;

        private JpegDecompressor m_cinfo;

        /* Initially allocated colormap is saved here */
        private byte[][] m_sv_colormap;	/* The color map as a 2-D pixel array */
        private int m_sv_actual;		/* number of entries in use */

        private byte[][] m_colorindex;	/* Precomputed mapping for speed */
        private int[] m_colorindexOffset;

        /* colorindex[i][j] = index of color closest to pixel value j in component i,
        * premultiplied as described above.  Since colormap indexes must fit into
        * bytes, the entries of this array will too.
        */
        private bool m_is_padded;		/* is the colorindex padded for odither? */

        private int[] m_Ncolors = new int[MAX_Q_COMPS];	/* # of values alloced to each component */

        /* Variables for ordered dithering */
        private int m_row_index;		/* cur row's vertical index in dither matrix */
        private int[][][] m_odither = new int[MAX_Q_COMPS][][]; /* one dither array per component */

        /* Variables for Floyd-Steinberg dithering */
        private short[][] m_fserrors = new short[MAX_Q_COMPS][]; /* accumulated errors */
        private bool m_on_odd_row;		/* flag to remember which row we are on */

        /// <summary>
        /// Module initialization routine for 1-pass color quantization.
        /// </summary>
        /// <param name="cinfo">The cinfo.</param>
        public Pass1ColorQuantizer(JpegDecompressor cinfo)
        {
            m_cinfo = cinfo;

            m_fserrors[0] = null; /* Flag FS workspace not allocated */
            m_odither[0] = null;    /* Also flag odither arrays not allocated */

            /* Make sure my internal arrays won't overflow */
            if (cinfo.m_out_color_components > MAX_Q_COMPS)
                throw new Exception(String.Format("Cannot quantize more than {0} color components", MAX_Q_COMPS));

            /* Make sure colormap indexes can be represented by JSAMPLEs */
            if (cinfo.m_desired_number_of_colors > (JpegConstants.MaxSampleValue + 1))
                throw new Exception(String.Format("Cannot quantize to more than {0} colors", JpegConstants.MaxSampleValue + 1));

            /* Create the colormap and color index table. */
            create_colormap();
            create_colorindex();

            /* Allocate Floyd-Steinberg workspace now if requested.
            * We do this now since it is FAR storage and may affect the memory
            * manager's space calculations.  If the user changes to FS dither
            * mode in a later pass, we will allocate the space then, and will
            * possibly overrun the max_memory_to_use setting.
            */
            if (cinfo.m_dither_mode == DitherMode.FloydStein)
                alloc_fs_workspace();
        }

        /// <summary>
        /// Initialize for one-pass color quantization.
        /// </summary>
        public virtual void start_pass(bool is_pre_scan)
        {
            /* Install my colormap. */
            m_cinfo.m_colormap = m_sv_colormap;
            m_cinfo.m_actual_number_of_colors = m_sv_actual;

            /* Initialize for desired dithering mode. */
            switch (m_cinfo.m_dither_mode)
            {
                case DitherMode.None:
                    if (m_cinfo.m_out_color_components == 3)
                        m_quantizer = QuantizerType.color_quantizer3;
                    else
                        m_quantizer = QuantizerType.color_quantizer;

                    break;
                case DitherMode.Ordered:
                    if (m_cinfo.m_out_color_components == 3)
                        m_quantizer = QuantizerType.quantize3_ord_dither_quantizer;
                    else
                        m_quantizer = QuantizerType.quantize3_ord_dither_quantizer;

                    /* initialize state for ordered dither */
                    m_row_index = 0;

                    /* If user changed to ordered dither from another mode,
                     * we must recreate the color index table with padding.
                     * This will cost extra space, but probably isn't very likely.
                     */
                    if (!m_is_padded)
                        create_colorindex();

                    /* Create ordered-dither tables if we didn't already. */
                    if (m_odither[0] == null)
                        create_odither_tables();

                    break;
                case DitherMode.FloydStein:
                    m_quantizer = QuantizerType.quantize_fs_dither_quantizer;

                    /* initialize state for F-S dither */
                    m_on_odd_row = false;

                    /* Allocate Floyd-Steinberg workspace if didn't already. */
                    if (m_fserrors[0] == null)
                        alloc_fs_workspace();

                    /* Initialize the propagated errors to zero. */
                    int arraysize = m_cinfo.m_output_width + 2;
                    for (int i = 0; i < m_cinfo.m_out_color_components; i++)
                        Array.Clear(m_fserrors[i], 0, arraysize);

                    break;
                default:
                    throw new Exception("Unknown Dither Mode");
            }
        }

        public virtual void color_quantize(byte[][] input_buf, int in_row, byte[][] output_buf, int out_row, int num_rows)
        {
            switch (m_quantizer)
            {
                case QuantizerType.color_quantizer3:
                    quantize3(input_buf, in_row, output_buf, out_row, num_rows);
                    break;
                case QuantizerType.color_quantizer:
                    quantize(input_buf, in_row, output_buf, out_row, num_rows);
                    break;
                case QuantizerType.quantize3_ord_dither_quantizer:
                    quantize3_ord_dither(input_buf, in_row, output_buf, out_row, num_rows);
                    break;
                case QuantizerType.quantize_ord_dither_quantizer:
                    quantize_ord_dither(input_buf, in_row, output_buf, out_row, num_rows);
                    break;
                case QuantizerType.quantize_fs_dither_quantizer:
                    quantize_fs_dither(input_buf, in_row, output_buf, out_row, num_rows);
                    break;
                default:
                    throw new Exception("Not implemented yet");
            }
        }

        /// <summary>
        /// Finish up at the end of the pass.
        /// </summary>
        public virtual void finish_pass()
        {
            /* no work in 1-pass case */
        }

        /// <summary>
        /// Switch to a new external colormap between output passes.
        /// Shouldn't get to this!
        /// </summary>
        public virtual void new_color_map()
        {
            throw new Exception("Invalid mode change during color quantization");
        }

        /// <summary>
        /// Map some rows of pixels to the output colormapped representation.
        /// General case, no dithering.
        /// </summary>
        private void quantize(byte[][] input_buf, int in_row, byte[][] output_buf, int out_row, int num_rows)
        {
            int nc = m_cinfo.m_out_color_components;

            for (int row = 0; row < num_rows; row++)
            {
                int inIndex = 0;
                int inRow = in_row + row;

                int outIndex = 0;
                int outRow = out_row + row;

                for (int col = m_cinfo.m_output_width; col > 0; col--)
                {
                    int pixcode = 0;
                    for (int ci = 0; ci < nc; ci++)
                    {
                        pixcode += m_colorindex[ci][m_colorindexOffset[ci] + input_buf[inRow][inIndex]];
                        inIndex++;
                    }

                    output_buf[outRow][outIndex] = (byte)pixcode;
                    outIndex++;
                }
            }
        }

        /// <summary>
        /// Map some rows of pixels to the output colormapped representation.
        /// Fast path for out_color_components==3, no dithering
        /// </summary>
        private void quantize3(byte[][] input_buf, int in_row, byte[][] output_buf, int out_row, int num_rows)
        {
            int width = m_cinfo.m_output_width;

            for (int row = 0; row < num_rows; row++)
            {
                int inIndex = 0;
                int inRow = in_row + row;

                int outIndex = 0;
                int outRow = out_row + row;

                for (int col = width; col > 0; col--)
                {
                    int pixcode = m_colorindex[0][m_colorindexOffset[0] + input_buf[inRow][inIndex]];
                    inIndex++;

                    pixcode += m_colorindex[1][m_colorindexOffset[1] + input_buf[inRow][inIndex]];
                    inIndex++;

                    pixcode += m_colorindex[2][m_colorindexOffset[2] + input_buf[inRow][inIndex]];
                    inIndex++;

                    output_buf[outRow][outIndex] = (byte)pixcode;
                    outIndex++;
                }
            }
        }

        /// <summary>
        /// Map some rows of pixels to the output colormapped representation.
        /// General case, with ordered dithering.
        /// </summary>
        private void quantize_ord_dither(byte[][] input_buf, int in_row, byte[][] output_buf, int out_row, int num_rows)
        {
            int nc = m_cinfo.m_out_color_components;
            int width = m_cinfo.m_output_width;

            for (int row = 0; row < num_rows; row++)
            {
                /* Initialize output values to 0 so can process components separately */
                Array.Clear(output_buf[out_row + row], 0, width);

                int row_index = m_row_index;
                for (int ci = 0; ci < nc; ci++)
                {
                    int inputIndex = ci;
                    int outIndex = 0;
                    int outRow = out_row + row;

                    int col_index = 0;
                    for (int col = width; col > 0; col--)
                    {
                        /* Form pixel value + dither, range-limit to 0..MaxSampleValue,
                         * select output value, accumulate into output code for this pixel.
                         * Range-limiting need not be done explicitly, as we have extended
                         * the colorindex table to produce the right answers for out-of-range
                         * inputs.  The maximum dither is +- MaxSampleValue; this sets the
                         * required amount of padding.
                         */
                        output_buf[outRow][outIndex] += m_colorindex[ci][m_colorindexOffset[ci] + input_buf[in_row + row][inputIndex] + m_odither[ci][row_index][col_index]];
                        inputIndex += nc;
                        outIndex++;
                        col_index = (col_index + 1) & ODITHER_MASK;
                    }
                }

                /* Advance row index for next row */
                row_index = (row_index + 1) & ODITHER_MASK;
                m_row_index = row_index;
            }
        }

        /// <summary>
        /// Map some rows of pixels to the output colormapped representation.
        /// Fast path for out_color_components==3, with ordered dithering
        /// </summary>
        private void quantize3_ord_dither(byte[][] input_buf, int in_row, byte[][] output_buf, int out_row, int num_rows)
        {
            int width = m_cinfo.m_output_width;

            for (int row = 0; row < num_rows; row++)
            {
                int row_index = m_row_index;
                int inRow = in_row + row;
                int inIndex = 0;

                int outIndex = 0;
                int outRow = out_row + row;

                int col_index = 0;
                for (int col = width; col > 0; col--)
                {
                    int pixcode = m_colorindex[0][m_colorindexOffset[0] + input_buf[inRow][inIndex] + m_odither[0][row_index][col_index]];
                    inIndex++;

                    pixcode += m_colorindex[1][m_colorindexOffset[1] + input_buf[inRow][inIndex] + m_odither[1][row_index][col_index]];
                    inIndex++;

                    pixcode += m_colorindex[2][m_colorindexOffset[2] + input_buf[inRow][inIndex] + m_odither[2][row_index][col_index]];
                    inIndex++;

                    output_buf[outRow][outIndex] = (byte)pixcode;
                    outIndex++;

                    col_index = (col_index + 1) & ODITHER_MASK;
                }

                row_index = (row_index + 1) & ODITHER_MASK;
                m_row_index = row_index;
            }
        }

        /// <summary>
        /// Map some rows of pixels to the output colormapped representation.
        /// General case, with Floyd-Steinberg dithering
        /// </summary>
        private void quantize_fs_dither(byte[][] input_buf, int in_row, byte[][] output_buf, int out_row, int num_rows)
        {
            int nc = m_cinfo.m_out_color_components;
            int width = m_cinfo.m_output_width;

            byte[] limit = m_cinfo.m_sample_range_limit;
            int limitOffset = m_cinfo.m_sampleRangeLimitOffset;

            for (int row = 0; row < num_rows; row++)
            {
                /* Initialize output values to 0 so can process components separately */
                Array.Clear(output_buf[out_row + row], 0, width);

                for (int ci = 0; ci < nc; ci++)
                {
                    int inRow = in_row + row;
                    int inIndex = ci;

                    int outIndex = 0;
                    int outRow = out_row + row;

                    int errorIndex = 0;
                    int dir;            /* 1 for left-to-right, -1 for right-to-left */
                    if (m_on_odd_row)
                    {
                        /* work right to left in this row */
                        inIndex += (width - 1) * nc; /* so point to rightmost pixel */
                        outIndex += width - 1;
                        dir = -1;
                        errorIndex = width + 1; /* => entry after last column */
                    }
                    else
                    {
                        /* work left to right in this row */
                        dir = 1;
                        errorIndex = 0; /* => entry before first column */
                    }
                    int dirnc = dir * nc;

                    /* Preset error values: no error propagated to first pixel from left */
                    int cur = 0;
                    /* and no error propagated to row below yet */
                    int belowerr = 0;
                    int bpreverr = 0;

                    for (int col = width; col > 0; col--)
                    {
                        /* cur holds the error propagated from the previous pixel on the
                         * current line.  Add the error propagated from the previous line
                         * to form the complete error correction term for this pixel, and
                         * round the error term (which is expressed * 16) to an integer.
                         * RIGHT_SHIFT rounds towards minus infinity, so adding 8 is correct
                         * for either sign of the error value.
                         * Note: errorIndex is for *previous* column's array entry.
                         */
                        cur = JpegUtils.RIGHT_SHIFT(cur + m_fserrors[ci][errorIndex + dir] + 8, 4);

                        /* Form pixel value + error, and range-limit to 0..MaxSampleValue.
                         * The maximum error is +- MaxSampleValue; this sets the required size
                         * of the range_limit array.
                         */
                        cur += input_buf[inRow][inIndex];
                        cur = limit[limitOffset + cur];

                        /* Select output value, accumulate into output code for this pixel */
                        int pixcode = m_colorindex[ci][m_colorindexOffset[ci] + cur];
                        output_buf[outRow][outIndex] += (byte)pixcode;

                        /* Compute actual representation error at this pixel */
                        /* Note: we can do this even though we don't have the final */
                        /* pixel code, because the colormap is orthogonal. */
                        cur -= m_sv_colormap[ci][pixcode];

                        /* Compute error fractions to be propagated to adjacent pixels.
                         * Add these into the running sums, and simultaneously shift the
                         * next-line error sums left by 1 column.
                         */
                        int bnexterr = cur;
                        int delta = cur * 2;
                        cur += delta;       /* form error * 3 */
                        m_fserrors[ci][errorIndex + 0] = (short)(bpreverr + cur);
                        cur += delta;       /* form error * 5 */
                        bpreverr = belowerr + cur;
                        belowerr = bnexterr;
                        cur += delta;       /* form error * 7 */

                        /* At this point cur contains the 7/16 error value to be propagated
                         * to the next pixel on the current line, and all the errors for the
                         * next line have been shifted over. We are therefore ready to move on.
                         */
                        inIndex += dirnc; /* advance input to next column */
                        outIndex += dir;  /* advance output to next column */
                        errorIndex += dir;    /* advance errorIndex to current column */
                    }

                    /* Post-loop cleanup: we must unload the final error value into the
                     * final fserrors[] entry.  Note we need not unload belowerr because
                     * it is for the dummy column before or after the actual array.
                     */
                    m_fserrors[ci][errorIndex + 0] = (short)bpreverr; /* unload prev err into array */
                }

                m_on_odd_row = (m_on_odd_row ? false : true);
            }
        }

        /// <summary>
        /// Create the colormap.
        /// </summary>
        private void create_colormap()
        {
            /* Select number of colors for each component */
            int total_colors = select_ncolors(m_Ncolors);

            /* Allocate and fill in the colormap. */
            /* The colors are ordered in the map in standard row-major order, */
            /* i.e. rightmost (highest-indexed) color changes most rapidly. */
            byte[][] colormap = JpegCommonBase.AllocJpegSamples(total_colors, m_cinfo.m_out_color_components);

            /* blksize is number of adjacent repeated entries for a component */
            /* blkdist is distance between groups of identical entries for a component */
            int blkdist = total_colors;
            for (int i = 0; i < m_cinfo.m_out_color_components; i++)
            {
                /* fill in colormap entries for i'th color component */
                int nci = m_Ncolors[i]; /* # of distinct values for this color */
                int blksize = blkdist / nci;
                for (int j = 0; j < nci; j++)
                {
                    /* Compute j'th output value (out of nci) for component */
                    int val = output_value(j, nci - 1);

                    /* Fill in all colormap entries that have this value of this component */
                    for (int ptr = j * blksize; ptr < total_colors; ptr += blkdist)
                    {
                        /* fill in blksize entries beginning at ptr */
                        for (int k = 0; k < blksize; k++)
                            colormap[i][ptr + k] = (byte)val;
                    }
                }

                /* blksize of this color is blkdist of next */
                blkdist = blksize;
            }

            /* Save the colormap in private storage,
             * where it will survive color quantization mode changes.
             */
            m_sv_colormap = colormap;
            m_sv_actual = total_colors;
        }

        /// <summary>
        /// Create the color index table.
        /// </summary>
        private void create_colorindex()
        {
            /* For ordered dither, we pad the color index tables by MaxSampleValue in
             * each direction (input index values can be -MaxSampleValue .. 2*MaxSampleValue).
             * This is not necessary in the other dithering modes.  However, we
             * flag whether it was done in case user changes dithering mode.
             */
            int pad;
            if (m_cinfo.m_dither_mode == DitherMode.Ordered)
            {
                pad = JpegConstants.MaxSampleValue * 2;
                m_is_padded = true;
            }
            else
            {
                pad = 0;
                m_is_padded = false;
            }

            m_colorindex = JpegCommonBase.AllocJpegSamples(JpegConstants.MaxSampleValue + 1 + pad, m_cinfo.m_out_color_components);
            m_colorindexOffset = new int[m_cinfo.m_out_color_components];

            /* blksize is number of adjacent repeated entries for a component */
            int blksize = m_sv_actual;
            for (int i = 0; i < m_cinfo.m_out_color_components; i++)
            {
                /* fill in colorindex entries for i'th color component */
                int nci = m_Ncolors[i]; /* # of distinct values for this color */
                blksize = blksize / nci;

                /* adjust colorindex pointers to provide padding at negative indexes. */
                if (pad != 0)
                    m_colorindexOffset[i] += JpegConstants.MaxSampleValue;

                /* in loop, val = index of current output value, */
                /* and k = largest j that maps to current val */
                int val = 0;
                int k = largest_input_value(0, nci - 1);
                for (int j = 0; j <= JpegConstants.MaxSampleValue; j++)
                {
                    while (j > k)
                    {
                        /* advance val if past boundary */
                        k = largest_input_value(++val, nci - 1);
                    }

                    /* premultiply so that no multiplication needed in main processing */
                    m_colorindex[i][m_colorindexOffset[i] + j] = (byte)(val * blksize);
                }

                /* Pad at both ends if necessary */
                if (pad != 0)
                {
                    for (int j = 1; j <= JpegConstants.MaxSampleValue; j++)
                    {
                        m_colorindex[i][m_colorindexOffset[i] + -j] = m_colorindex[i][m_colorindexOffset[i]];
                        m_colorindex[i][m_colorindexOffset[i] + JpegConstants.MaxSampleValue + j] = m_colorindex[i][m_colorindexOffset[i] + JpegConstants.MaxSampleValue];
                    }
                }
            }
        }

        /// <summary>
        /// Create the ordered-dither tables.
        /// Components having the same number of representative colors may 
        /// share a dither table.
        /// </summary>
        private void create_odither_tables()
        {
            for (int i = 0; i < m_cinfo.m_out_color_components; i++)
            {
                int nci = m_Ncolors[i]; /* # of distinct values for this color */

                /* search for matching prior component */
                int foundPos = -1;
                for (int j = 0; j < i; j++)
                {
                    if (nci == m_Ncolors[j])
                    {
                        foundPos = j;
                        break;
                    }
                }

                if (foundPos == -1)
                {
                    /* need a new table? */
                    m_odither[i] = make_odither_array(nci);
                }
                else
                    m_odither[i] = m_odither[foundPos];
            }
        }

        /// <summary>
        /// Allocate workspace for Floyd-Steinberg errors.
        /// </summary>
        private void alloc_fs_workspace()
        {
            for (int i = 0; i < m_cinfo.m_out_color_components; i++)
                m_fserrors[i] = new short[m_cinfo.m_output_width + 2];
        }

        /* 
         * Policy-making subroutines for create_colormap and create_colorindex.
         * These routines determine the colormap to be used.  The rest of the module
         * only assumes that the colormap is orthogonal.
         *
         *  * select_ncolors decides how to divvy up the available colors
         *    among the components.
         *  * output_value defines the set of representative values for a component.
         *  * largest_input_value defines the mapping from input values to
         *    representative values for a component.
         * Note that the latter two routines may impose different policies for
         * different components, though this is not currently done.
         */

        /// <summary>
        /// Return largest input value that should map to j'th output value
        /// Must have largest(j=0) >= 0, and largest(j=maxj) >= MaxSampleValue
        /// </summary>
        private static int largest_input_value(int j, int maxj)
        {
            /* Breakpoints are halfway between values returned by output_value */
            return (int)(((2 * j + 1) * JpegConstants.MaxSampleValue + maxj) / (2 * maxj));
        }

        /// <summary>
        /// Return j'th output value, where j will range from 0 to maxj
        /// The output values must fall in 0..MaxSampleValue in increasing order
        /// </summary>
        private static int output_value(int j, int maxj)
        {
            /* We always provide values 0 and MaxSampleValue for each component;
             * any additional values are equally spaced between these limits.
             * (Forcing the upper and lower values to the limits ensures that
             * dithering can't produce a color outside the selected gamut.)
             */
            return (int)((j * JpegConstants.MaxSampleValue + maxj / 2) / maxj);
        }

        /// <summary>
        /// Determine allocation of desired colors to components,
        /// and fill in Ncolors[] array to indicate choice.
        /// Return value is total number of colors (product of Ncolors[] values).
        /// </summary>
        private int select_ncolors(int[] Ncolors)
        {
            int nc = m_cinfo.m_out_color_components; /* number of color components */
            int max_colors = m_cinfo.m_desired_number_of_colors;

            /* We can allocate at least the nc'th root of max_colors per component. */
            /* Compute floor(nc'th root of max_colors). */
            int iroot = 1;
            long temp = 0;
            do
            {
                iroot++;
                temp = iroot;       /* set temp = iroot ** nc */
                for (int i = 1; i < nc; i++)
                    temp *= iroot;
            }
            while (temp <= max_colors); /* repeat till iroot exceeds root */

            /* now iroot = floor(root) */
            iroot--;

            /* Must have at least 2 color values per component */
            if (iroot < 2)
                throw new Exception(String.Format("Cannot quantize to fewer than {0} colors", (int)temp));

            /* Initialize to iroot color values for each component */
            int total_colors = 1;
            for (int i = 0; i < nc; i++)
            {
                Ncolors[i] = iroot;
                total_colors *= iroot;
            }

            /* We may be able to increment the count for one or more components without
             * exceeding max_colors, though we know not all can be incremented.
             * Sometimes, the first component can be incremented more than once!
             * (Example: for 16 colors, we start at 2*2*2, go to 3*2*2, then 4*2*2.)
             * In RGB colorspace, try to increment G first, then R, then B.
             */
            bool changed = false;
            do
            {
                changed = false;
                for (int i = 0; i < nc; i++)
                {
                    int j = (m_cinfo.m_out_color_space == ColorSpace.RGB ? RGB_order[i] : i);
                    /* calculate new total_colors if Ncolors[j] is incremented */
                    temp = total_colors / Ncolors[j];
                    temp *= Ncolors[j] + 1; /* done in long arith to avoid oflo */

                    if (temp > max_colors)
                        break;          /* won't fit, done with this pass */

                    Ncolors[j]++;       /* OK, apply the increment */
                    total_colors = (int)temp;
                    changed = true;
                }
            }
            while (changed);

            return total_colors;
        }

        /// <summary>
        /// Create an ordered-dither array for a component having ncolors
        /// distinct output values.
        /// </summary>
        private static int[][] make_odither_array(int ncolors)
        {
            int[][] odither = new int[ODITHER_SIZE][];
            for (int i = 0; i < ODITHER_SIZE; i++)
                odither[i] = new int[ODITHER_SIZE];

            /* The inter-value distance for this color is MaxSampleValue/(ncolors-1).
             * Hence the dither value for the matrix cell with fill order f
             * (f=0..N-1) should be (N-1-2*f)/(2*N) * MaxSampleValue/(ncolors-1).
             * On 16-bit-int machine, be careful to avoid overflow.
             */
            int den = 2 * ODITHER_CELLS * (ncolors - 1);
            for (int j = 0; j < ODITHER_SIZE; j++)
            {
                for (int k = 0; k < ODITHER_SIZE; k++)
                {
                    int num = ((int)(ODITHER_CELLS - 1 - 2 * ((int)base_dither_matrix[j][k]))) * JpegConstants.MaxSampleValue;

                    /* Ensure round towards zero despite C's lack of consistency
                     * about rounding negative values in integer division...
                     */
                    odither[j][k] = num < 0 ? -((-num) / den) : num / den;
                }
            }

            return odither;
        }
    }
    #endregion

    #region Pass2ColorQuantizer
    /// <summary>
    /// This module implements the well-known Heckbert paradigm for color
    /// quantization.  Most of the ideas used here can be traced back to
    /// Heckbert's seminal paper
    /// Heckbert, Paul.  "Color Image Quantization for Frame Buffer Display",
    /// Proc. SIGGRAPH '82, Computer Graphics v.16 #3 (July 1982), pp 297-304.
    /// 
    /// In the first pass over the image, we accumulate a histogram showing the
    /// usage count of each possible color.  To keep the histogram to a reasonable
    /// size, we reduce the precision of the input; typical practice is to retain
    /// 5 or 6 bits per color, so that 8 or 4 different input values are counted
    /// in the same histogram cell.
    /// 
    /// Next, the color-selection step begins with a box representing the whole
    /// color space, and repeatedly splits the "largest" remaining box until we
    /// have as many boxes as desired colors.  Then the mean color in each
    /// remaining box becomes one of the possible output colors.
    /// 
    /// The second pass over the image maps each input pixel to the closest output
    /// color (optionally after applying a Floyd-Steinberg dithering correction).
    /// This mapping is logically trivial, but making it go fast enough requires
    /// considerable care.
    /// 
    /// Heckbert-style quantizers vary a good deal in their policies for choosing
    /// the "largest" box and deciding where to cut it.  The particular policies
    /// used here have proved out well in experimental comparisons, but better ones
    /// may yet be found.
    /// 
    /// In earlier versions of the IJG code, this module quantized in YCbCr color
    /// space, processing the raw upsampled data without a color conversion step.
    /// This allowed the color conversion math to be done only once per colormap
    /// entry, not once per pixel.  However, that optimization precluded other
    /// useful optimizations (such as merging color conversion with upsampling)
    /// and it also interfered with desired capabilities such as quantizing to an
    /// externally-supplied colormap.  We have therefore abandoned that approach.
    /// The present code works in the post-conversion color space, typically RGB.
    /// 
    /// To improve the visual quality of the results, we actually work in scaled
    /// RGB space, giving G distances more weight than R, and R in turn more than
    /// B.  To do everything in integer math, we must use integer scale factors.
    /// The 2/3/1 scale factors used here correspond loosely to the relative
    /// weights of the colors in the NTSC grayscale equation.
    /// If you want to use this code to quantize a non-RGB color space, you'll
    /// probably need to change these scale factors.
    /// 
    /// First we have the histogram data structure and routines for creating it.
    /// 
    /// The number of bits of precision can be adjusted by changing these symbols.
    /// We recommend keeping 6 bits for G and 5 each for R and B.
    /// If you have plenty of memory and cycles, 6 bits all around gives marginally
    /// better results; if you are short of memory, 5 bits all around will save
    /// some space but degrade the results.
    /// To maintain a fully accurate histogram, we'd need to allocate a "long"
    /// (preferably unsigned long) for each cell.  In practice this is overkill;
    /// we can get by with 16 bits per cell.  Few of the cell counts will overflow,
    /// and clamping those that do overflow to the maximum value will give close-
    /// enough results.  This reduces the recommended histogram size from 256Kb
    /// to 128Kb, which is a useful savings on PC-class machines.
    /// (In the second pass the histogram space is re-used for pixel mapping data;
    /// in that capacity, each cell must be able to store zero to the number of
    /// desired colors.  16 bits/cell is plenty for that too.)
    /// Since the JPEG code is intended to run in small memory model on 80x86
    /// machines, we can't just allocate the histogram in one chunk.  Instead
    /// of a true 3-D array, we use a row of pointers to 2-D arrays.  Each
    /// pointer corresponds to a C0 value (typically 2^5 = 32 pointers) and
    /// each 2-D array has 2^6*2^5 = 2048 or 2^6*2^6 = 4096 entries.  Note that
    /// on 80x86 machines, the pointer row is in near memory but the actual
    /// arrays are in far memory (same arrangement as we use for image arrays).
    /// 
    /// 
    /// Declarations for Floyd-Steinberg dithering.
    /// 
    /// Errors are accumulated into the array fserrors[], at a resolution of
    /// 1/16th of a pixel count.  The error at a given pixel is propagated
    /// to its not-yet-processed neighbors using the standard F-S fractions,
    ///     ... (here)  7/16
    /// 3/16    5/16    1/16
    /// We work left-to-right on even rows, right-to-left on odd rows.
    /// 
    /// We can get away with a single array (holding one row's worth of errors)
    /// by using it to store the current row's errors at pixel columns not yet
    /// processed, but the next row's errors at columns already processed.  We
    /// need only a few extra variables to hold the errors immediately around the
    /// current column.  (If we are lucky, those variables are in registers, but
    /// even if not, they're probably cheaper to access than array elements are.)
    /// 
    /// The fserrors[] array has (#columns + 2) entries; the extra entry at
    /// each end saves us from special-casing the first and last pixels.
    /// Each entry is three values long, one value for each color component.
    /// </summary>
    class Pass2ColorQuantizer : ColorQuantizer
    {
        private struct box
        {
            /* The bounds of the box (inclusive); expressed as histogram indexes */
            public int c0min;
            public int c0max;
            public int c1min;
            public int c1max;
            public int c2min;
            public int c2max;
            /* The volume (actually 2-norm) of the box */
            public int volume;
            /* The number of nonzero histogram cells within this box */
            public long colorcount;
        }

        private enum QuantizerType
        {
            prescan_quantizer,
            pass2_fs_dither_quantizer,
            pass2_no_dither_quantizer
        }

        private const int MAXNUMCOLORS = (JpegConstants.MaxSampleValue + 1); /* maximum size of colormap */

        /* These will do the right thing for either R,G,B or B,G,R color order,
        * but you may not like the results for other color orders.
        */
        private const int HIST_C0_BITS = 5;     /* bits of precision in R/B histogram */
        private const int HIST_C1_BITS = 6;     /* bits of precision in G histogram */
        private const int HIST_C2_BITS = 5;     /* bits of precision in B/R histogram */

        /* Number of elements along histogram axes. */
        private const int HIST_C0_ELEMS = (1 << HIST_C0_BITS);
        private const int HIST_C1_ELEMS = (1 << HIST_C1_BITS);
        private const int HIST_C2_ELEMS = (1 << HIST_C2_BITS);

        /* These are the amounts to shift an input value to get a histogram index. */
        private const int C0_SHIFT = (JpegConstants.BitsInSample - HIST_C0_BITS);
        private const int C1_SHIFT = (JpegConstants.BitsInSample - HIST_C1_BITS);
        private const int C2_SHIFT = (JpegConstants.BitsInSample - HIST_C2_BITS);

        private const int R_SCALE = 2;       /* scale R distances by this much */
        private const int G_SCALE = 3;       /* scale G distances by this much */
        private const int B_SCALE = 1;       /* and B by this much */

        /* log2(histogram cells in update box) for each axis; this can be adjusted */
        private const int BOX_C0_LOG = (HIST_C0_BITS - 3);
        private const int BOX_C1_LOG = (HIST_C1_BITS - 3);
        private const int BOX_C2_LOG = (HIST_C2_BITS - 3);

        private const int BOX_C0_ELEMS = (1 << BOX_C0_LOG); /* # of hist cells in update box */
        private const int BOX_C1_ELEMS = (1 << BOX_C1_LOG);
        private const int BOX_C2_ELEMS = (1 << BOX_C2_LOG);

        private const int BOX_C0_SHIFT = (C0_SHIFT + BOX_C0_LOG);
        private const int BOX_C1_SHIFT = (C1_SHIFT + BOX_C1_LOG);
        private const int BOX_C2_SHIFT = (C2_SHIFT + BOX_C2_LOG);

        private QuantizerType m_quantizer;

        private bool m_useFinishPass1;

        private JpegDecompressor m_cinfo;

        /* Space for the eventually created colormap is stashed here */
        private byte[][] m_sv_colormap;  /* colormap allocated at init time */
        private int m_desired;            /* desired # of colors = size of colormap */

        /* Variables for accumulating image statistics */
        private ushort[][] m_histogram;     /* pointer to the histogram */

        private bool m_needs_zeroed;      /* true if next pass must zero histogram */

        /* Variables for Floyd-Steinberg dithering */
        private short[] m_fserrors;      /* accumulated errors */
        private bool m_on_odd_row;        /* flag to remember which row we are on */
        private int[] m_error_limiter;     /* table for clamping the applied error */

        /// <summary>
        /// Module initialization routine for 2-pass color quantization.
        /// </summary>
        public Pass2ColorQuantizer(JpegDecompressor cinfo)
        {
            m_cinfo = cinfo;

            /* Make sure jdmaster didn't give me a case I can't handle */
            if (cinfo.m_out_color_components != 3)
                throw new Exception("Unable to handle anything other than 3 color components!");

            /* Allocate the histogram/inverse colormap storage */
            m_histogram = new ushort[HIST_C0_ELEMS][];
            for (int i = 0; i < HIST_C0_ELEMS; i++)
                m_histogram[i] = new ushort[HIST_C1_ELEMS * HIST_C2_ELEMS];

            m_needs_zeroed = true; /* histogram is garbage now */

            /* Allocate storage for the completed colormap, if required.
            * We do this now since it is FAR storage and may affect
            * the memory manager's space calculations.
            */
            if (cinfo.m_enable_2pass_quant)
            {
                /* Make sure color count is acceptable */
                int desired_local = cinfo.m_desired_number_of_colors;

                /* Lower bound on # of colors ... somewhat arbitrary as long as > 0 */
                if (desired_local < 8)
                    throw new Exception("Cannot quantize to fewer than 8 colors");

                /* Make sure colormap indexes can be represented by JSAMPLEs */
                if (desired_local > MAXNUMCOLORS)
                    throw new Exception(String.Format("Cannot quantize to more than {0} colors", MAXNUMCOLORS));

                m_sv_colormap = JpegCommonBase.AllocJpegSamples(desired_local, 3);
                m_desired = desired_local;
            }

            /* Only F-S dithering or no dithering is supported. */
            /* If user asks for ordered dither, give him F-S. */
            if (cinfo.m_dither_mode != DitherMode.None)
                cinfo.m_dither_mode = DitherMode.FloydStein;

            /* Allocate Floyd-Steinberg workspace if necessary.
            * This isn't really needed until pass 2, but again it is FAR storage.
            * Although we will cope with a later change in dither_mode,
            * we do not promise to honor max_memory_to_use if dither_mode changes.
            */
            if (cinfo.m_dither_mode == DitherMode.FloydStein)
            {
                m_fserrors = new short[(cinfo.m_output_width + 2) * 3];

                /* Might as well create the error-limiting table too. */
                init_error_limit();
            }
        }

        /// <summary>
        /// Initialize for each processing pass.
        /// </summary>
        public virtual void start_pass(bool is_pre_scan)
        {
            /* Only F-S dithering or no dithering is supported. */
            /* If user asks for ordered dither, give him F-S. */
            if (m_cinfo.m_dither_mode != DitherMode.None)
                m_cinfo.m_dither_mode = DitherMode.FloydStein;

            if (is_pre_scan)
            {
                /* Set up method pointers */
                m_quantizer = QuantizerType.prescan_quantizer;
                m_useFinishPass1 = true;
                m_needs_zeroed = true; /* Always zero histogram */
            }
            else
            {
                /* Set up method pointers */
                if (m_cinfo.m_dither_mode == DitherMode.FloydStein)
                    m_quantizer = QuantizerType.pass2_fs_dither_quantizer;
                else
                    m_quantizer = QuantizerType.pass2_no_dither_quantizer;

                m_useFinishPass1 = false;

                /* Make sure color count is acceptable */
                int i = m_cinfo.m_actual_number_of_colors;
                if (i < 1)
                    throw new Exception("Cannot quantize to less than 1 color");

                if (i > MAXNUMCOLORS)
                    throw new Exception(String.Format("Cannot quantize to more than {0} colors", MAXNUMCOLORS));

                if (m_cinfo.m_dither_mode == DitherMode.FloydStein)
                {
                    /* Allocate Floyd-Steinberg workspace if we didn't already. */
                    if (m_fserrors == null)
                    {
                        int arraysize = (m_cinfo.m_output_width + 2) * 3;
                        m_fserrors = new short[arraysize];
                    }
                    else
                    {
                        /* Initialize the propagated errors to zero. */
                        Array.Clear(m_fserrors, 0, m_fserrors.Length);
                    }

                    /* Make the error-limit table if we didn't already. */
                    if (m_error_limiter == null)
                        init_error_limit();

                    m_on_odd_row = false;
                }
            }

            /* Zero the histogram or inverse color map, if necessary */
            if (m_needs_zeroed)
            {
                for (int i = 0; i < HIST_C0_ELEMS; i++)
                    Array.Clear(m_histogram[i], 0, m_histogram[i].Length);

                m_needs_zeroed = false;
            }
        }

        public virtual void color_quantize(byte[][] input_buf, int in_row, byte[][] output_buf, int out_row, int num_rows)
        {
            switch (m_quantizer)
            {
                case QuantizerType.prescan_quantizer:
                    prescan_quantize(input_buf, in_row, num_rows);
                    break;
                case QuantizerType.pass2_fs_dither_quantizer:
                    pass2_fs_dither(input_buf, in_row, output_buf, out_row, num_rows);
                    break;
                case QuantizerType.pass2_no_dither_quantizer:
                    pass2_no_dither(input_buf, in_row, output_buf, out_row, num_rows);
                    break;
                default:
                    throw new Exception("Specified Quantizer Type not implemented");
            }
        }

        public virtual void finish_pass()
        {
            if (m_useFinishPass1)
                finish_pass1();
        }

        /// <summary>
        /// Switch to a new external colormap between output passes.
        /// </summary>
        public virtual void new_color_map()
        {
            /* Reset the inverse color map */
            m_needs_zeroed = true;
        }

        /// <summary>
        /// Prescan some rows of pixels.
        /// In this module the prescan simply updates the histogram, which has been
        /// initialized to zeroes by start_pass.
        /// An output_buf parameter is required by the method signature, but no data
        /// is actually output (in fact the buffer controller is probably passing a
        /// null pointer).
        /// </summary>
        private void prescan_quantize(byte[][] input_buf, int in_row, int num_rows)
        {
            for (int row = 0; row < num_rows; row++)
            {
                int inputIndex = 0;
                for (int col = m_cinfo.m_output_width; col > 0; col--)
                {
                    int rowIndex = (int)input_buf[in_row + row][inputIndex] >> C0_SHIFT;
                    int columnIndex = ((int)input_buf[in_row + row][inputIndex + 1] >> C1_SHIFT) * HIST_C2_ELEMS +
                        ((int)input_buf[in_row + row][inputIndex + 2] >> C2_SHIFT);

                    /* increment pixel value, check for overflow and undo increment if so. */
                    m_histogram[rowIndex][columnIndex]++;
                    if (m_histogram[rowIndex][columnIndex] <= 0)
                        m_histogram[rowIndex][columnIndex]--;

                    inputIndex += 3;
                }
            }
        }

        /// <summary>
        /// Map some rows of pixels to the output colormapped representation.
        /// This version performs Floyd-Steinberg dithering
        /// </summary>
        private void pass2_fs_dither(byte[][] input_buf, int in_row, byte[][] output_buf, int out_row, int num_rows)
        {
            byte[] limit = m_cinfo.m_sample_range_limit;
            int limitOffset = m_cinfo.m_sampleRangeLimitOffset;

            for (int row = 0; row < num_rows; row++)
            {
                int inputPixelIndex = 0;
                int outputPixelIndex = 0;
                int errorIndex = 0;
                int dir;            /* +1 or -1 depending on direction */
                int dir3;           /* 3*dir, for advancing inputIndex & errorIndex */
                if (m_on_odd_row)
                {
                    /* work right to left in this row */
                    inputPixelIndex += (m_cinfo.m_output_width - 1) * 3;   /* so point to rightmost pixel */
                    outputPixelIndex += m_cinfo.m_output_width - 1;
                    dir = -1;
                    dir3 = -3;
                    errorIndex = (m_cinfo.m_output_width + 1) * 3; /* => entry after last column */
                    m_on_odd_row = false; /* flip for next time */
                }
                else
                {
                    /* work left to right in this row */
                    dir = 1;
                    dir3 = 3;
                    errorIndex = 0; /* => entry before first real column */
                    m_on_odd_row = true; /* flip for next time */
                }

                /* Preset error values: no error propagated to first pixel from left */
                /* current error or pixel value */
                int cur0 = 0;
                int cur1 = 0;
                int cur2 = 0;
                /* and no error propagated to row below yet */
                /* error for pixel below cur */
                int belowerr0 = 0;
                int belowerr1 = 0;
                int belowerr2 = 0;
                /* error for below/prev col */
                int bpreverr0 = 0;
                int bpreverr1 = 0;
                int bpreverr2 = 0;

                for (int col = m_cinfo.m_output_width; col > 0; col--)
                {
                    /* curN holds the error propagated from the previous pixel on the
                     * current line.  Add the error propagated from the previous line
                     * to form the complete error correction term for this pixel, and
                     * round the error term (which is expressed * 16) to an integer.
                     * RIGHT_SHIFT rounds towards minus infinity, so adding 8 is correct
                     * for either sign of the error value.
                     * Note: errorIndex is for *previous* column's array entry.
                     */
                    cur0 = JpegUtils.RIGHT_SHIFT(cur0 + m_fserrors[errorIndex + dir3] + 8, 4);
                    cur1 = JpegUtils.RIGHT_SHIFT(cur1 + m_fserrors[errorIndex + dir3 + 1] + 8, 4);
                    cur2 = JpegUtils.RIGHT_SHIFT(cur2 + m_fserrors[errorIndex + dir3 + 2] + 8, 4);

                    /* Limit the error using transfer function set by init_error_limit.
                     * See comments with init_error_limit for rationale.
                     */
                    cur0 = m_error_limiter[JpegConstants.MaxSampleValue + cur0];
                    cur1 = m_error_limiter[JpegConstants.MaxSampleValue + cur1];
                    cur2 = m_error_limiter[JpegConstants.MaxSampleValue + cur2];

                    /* Form pixel value + error, and range-limit to 0..MaxSampleValue.
                     * The maximum error is +- MaxSampleValue (or less with error limiting);
                     * this sets the required size of the range_limit array.
                     */
                    cur0 += input_buf[in_row + row][inputPixelIndex];
                    cur1 += input_buf[in_row + row][inputPixelIndex + 1];
                    cur2 += input_buf[in_row + row][inputPixelIndex + 2];
                    cur0 = limit[limitOffset + cur0];
                    cur1 = limit[limitOffset + cur1];
                    cur2 = limit[limitOffset + cur2];

                    /* Index into the cache with adjusted pixel value */
                    int hRow = cur0 >> C0_SHIFT;
                    int hColumn = (cur1 >> C1_SHIFT) * HIST_C2_ELEMS + (cur2 >> C2_SHIFT);

                    /* If we have not seen this color before, find nearest colormap */
                    /* entry and update the cache */
                    if (m_histogram[hRow][hColumn] == 0)
                        fill_inverse_cmap(cur0 >> C0_SHIFT, cur1 >> C1_SHIFT, cur2 >> C2_SHIFT);

                    /* Now emit the colormap index for this cell */
                    int pixcode = m_histogram[hRow][hColumn] - 1;
                    output_buf[out_row + row][outputPixelIndex] = (byte)pixcode;

                    /* Compute representation error for this pixel */
                    cur0 -= m_cinfo.m_colormap[0][pixcode];
                    cur1 -= m_cinfo.m_colormap[1][pixcode];
                    cur2 -= m_cinfo.m_colormap[2][pixcode];

                    /* Compute error fractions to be propagated to adjacent pixels.
                     * Add these into the running sums, and simultaneously shift the
                     * next-line error sums left by 1 column.
                     */
                    int bnexterr = cur0;    /* Process component 0 */
                    int delta = cur0 * 2;
                    cur0 += delta;      /* form error * 3 */
                    m_fserrors[errorIndex] = (short)(bpreverr0 + cur0);
                    cur0 += delta;      /* form error * 5 */
                    bpreverr0 = belowerr0 + cur0;
                    belowerr0 = bnexterr;
                    cur0 += delta;      /* form error * 7 */
                    bnexterr = cur1;    /* Process component 1 */
                    delta = cur1 * 2;
                    cur1 += delta;      /* form error * 3 */
                    m_fserrors[errorIndex + 1] = (short)(bpreverr1 + cur1);
                    cur1 += delta;      /* form error * 5 */
                    bpreverr1 = belowerr1 + cur1;
                    belowerr1 = bnexterr;
                    cur1 += delta;      /* form error * 7 */
                    bnexterr = cur2;    /* Process component 2 */
                    delta = cur2 * 2;
                    cur2 += delta;      /* form error * 3 */
                    m_fserrors[errorIndex + 2] = (short)(bpreverr2 + cur2);
                    cur2 += delta;      /* form error * 5 */
                    bpreverr2 = belowerr2 + cur2;
                    belowerr2 = bnexterr;
                    cur2 += delta;      /* form error * 7 */

                    /* At this point curN contains the 7/16 error value to be propagated
                     * to the next pixel on the current line, and all the errors for the
                     * next line have been shifted over.  We are therefore ready to move on.
                     */
                    inputPixelIndex += dir3;      /* Advance pixel pointers to next column */
                    outputPixelIndex += dir;
                    errorIndex += dir3;       /* advance errorIndex to current column */
                }

                /* Post-loop cleanup: we must unload the final error values into the
                 * final fserrors[] entry.  Note we need not unload belowerrN because
                 * it is for the dummy column before or after the actual array.
                 */
                m_fserrors[errorIndex] = (short)bpreverr0; /* unload prev errs into array */
                m_fserrors[errorIndex + 1] = (short)bpreverr1;
                m_fserrors[errorIndex + 2] = (short)bpreverr2;
            }
        }

        /// <summary>
        /// Map some rows of pixels to the output colormapped representation.
        /// This version performs no dithering
        /// </summary>
        private void pass2_no_dither(byte[][] input_buf, int in_row, byte[][] output_buf, int out_row, int num_rows)
        {
            for (int row = 0; row < num_rows; row++)
            {
                int inRow = row + in_row;
                int inIndex = 0;
                int outIndex = 0;
                int outRow = out_row + row;
                for (int col = m_cinfo.m_output_width; col > 0; col--)
                {
                    /* get pixel value and index into the cache */
                    int c0 = (int)input_buf[inRow][inIndex] >> C0_SHIFT;
                    inIndex++;

                    int c1 = (int)input_buf[inRow][inIndex] >> C1_SHIFT;
                    inIndex++;

                    int c2 = (int)input_buf[inRow][inIndex] >> C2_SHIFT;
                    inIndex++;

                    int hRow = c0;
                    int hColumn = c1 * HIST_C2_ELEMS + c2;

                    /* If we have not seen this color before, find nearest colormap entry */
                    /* and update the cache */
                    if (m_histogram[hRow][hColumn] == 0)
                        fill_inverse_cmap(c0, c1, c2);

                    /* Now emit the colormap index for this cell */
                    output_buf[outRow][outIndex] = (byte)(m_histogram[hRow][hColumn] - 1);
                    outIndex++;
                }
            }
        }

        /// <summary>
        /// Finish up at the end of each pass.
        /// </summary>
        private void finish_pass1()
        {
            /* Select the representative colors and fill in cinfo.colormap */
            m_cinfo.m_colormap = m_sv_colormap;
            select_colors(m_desired);

            /* Force next pass to zero the color index table */
            m_needs_zeroed = true;
        }

        /// <summary>
        /// Compute representative color for a box, put it in colormap[icolor]
        /// </summary>
        private void compute_color(box[] boxlist, int boxIndex, int icolor)
        {
            /* Current algorithm: mean weighted by pixels (not colors) */
            /* Note it is important to get the rounding correct! */
            long total = 0;
            long c0total = 0;
            long c1total = 0;
            long c2total = 0;
            box curBox = boxlist[boxIndex];
            for (int c0 = curBox.c0min; c0 <= curBox.c0max; c0++)
            {
                for (int c1 = curBox.c1min; c1 <= curBox.c1max; c1++)
                {
                    int histogramIndex = c1 * HIST_C2_ELEMS + curBox.c2min;
                    for (int c2 = curBox.c2min; c2 <= curBox.c2max; c2++)
                    {
                        long count = m_histogram[c0][histogramIndex];
                        histogramIndex++;

                        if (count != 0)
                        {
                            total += count;
                            c0total += ((c0 << C0_SHIFT) + ((1 << C0_SHIFT) >> 1)) * count;
                            c1total += ((c1 << C1_SHIFT) + ((1 << C1_SHIFT) >> 1)) * count;
                            c2total += ((c2 << C2_SHIFT) + ((1 << C2_SHIFT) >> 1)) * count;
                        }
                    }
                }
            }

            m_cinfo.m_colormap[0][icolor] = (byte)((c0total + (total >> 1)) / total);
            m_cinfo.m_colormap[1][icolor] = (byte)((c1total + (total >> 1)) / total);
            m_cinfo.m_colormap[2][icolor] = (byte)((c2total + (total >> 1)) / total);
        }

        /// <summary>
        /// Master routine for color selection
        /// </summary>
        private void select_colors(int desired_colors)
        {
            /* Allocate workspace for box list */
            box[] boxlist = new box[desired_colors];

            /* Initialize one box containing whole space */
            int numboxes = 1;
            boxlist[0].c0min = 0;
            boxlist[0].c0max = JpegConstants.MaxSampleValue >> C0_SHIFT;
            boxlist[0].c1min = 0;
            boxlist[0].c1max = JpegConstants.MaxSampleValue >> C1_SHIFT;
            boxlist[0].c2min = 0;
            boxlist[0].c2max = JpegConstants.MaxSampleValue >> C2_SHIFT;

            /* Shrink it to actually-used volume and set its statistics */
            update_box(boxlist, 0);

            /* Perform median-cut to produce final box list */
            numboxes = median_cut(boxlist, numboxes, desired_colors);

            /* Compute the representative color for each box, fill colormap */
            for (int i = 0; i < numboxes; i++)
                compute_color(boxlist, i, i);

            m_cinfo.m_actual_number_of_colors = numboxes;
        }

        /// <summary>
        /// Repeatedly select and split the largest box until we have enough boxes
        /// </summary>
        private int median_cut(box[] boxlist, int numboxes, int desired_colors)
        {
            while (numboxes < desired_colors)
            {
                /* Select box to split.
                 * Current algorithm: by population for first half, then by volume.
                 */
                int foundIndex;
                if (numboxes * 2 <= desired_colors)
                    foundIndex = find_biggest_color_pop(boxlist, numboxes);
                else
                    foundIndex = find_biggest_volume(boxlist, numboxes);

                if (foundIndex == -1)     /* no splittable boxes left! */
                    break;

                /* Copy the color bounds to the new box. */
                boxlist[numboxes].c0max = boxlist[foundIndex].c0max;
                boxlist[numboxes].c1max = boxlist[foundIndex].c1max;
                boxlist[numboxes].c2max = boxlist[foundIndex].c2max;
                boxlist[numboxes].c0min = boxlist[foundIndex].c0min;
                boxlist[numboxes].c1min = boxlist[foundIndex].c1min;
                boxlist[numboxes].c2min = boxlist[foundIndex].c2min;

                /* Choose which axis to split the box on.
                 * Current algorithm: longest scaled axis.
                 * See notes in update_box about scaling distances.
                 */
                int c0 = ((boxlist[foundIndex].c0max - boxlist[foundIndex].c0min) << C0_SHIFT) * R_SCALE;
                int c1 = ((boxlist[foundIndex].c1max - boxlist[foundIndex].c1min) << C1_SHIFT) * G_SCALE;
                int c2 = ((boxlist[foundIndex].c2max - boxlist[foundIndex].c2min) << C2_SHIFT) * B_SCALE;

                /* We want to break any ties in favor of green, then red, blue last.
                 * This code does the right thing for R,G,B or B,G,R color orders only.
                 */
                int cmax = c1;
                int n = 1;

                if (c0 > cmax)
                {
                    cmax = c0;
                    n = 0;
                }

                if (c2 > cmax)
                {
                    n = 2;
                }

                /* Choose split point along selected axis, and update box bounds.
                 * Current algorithm: split at halfway point.
                 * (Since the box has been shrunk to minimum volume,
                 * any split will produce two nonempty subboxes.)
                 * Note that lb value is max for lower box, so must be < old max.
                 */
                int lb;
                switch (n)
                {
                    case 0:
                        lb = (boxlist[foundIndex].c0max + boxlist[foundIndex].c0min) / 2;
                        boxlist[foundIndex].c0max = lb;
                        boxlist[numboxes].c0min = lb + 1;
                        break;
                    case 1:
                        lb = (boxlist[foundIndex].c1max + boxlist[foundIndex].c1min) / 2;
                        boxlist[foundIndex].c1max = lb;
                        boxlist[numboxes].c1min = lb + 1;
                        break;
                    case 2:
                        lb = (boxlist[foundIndex].c2max + boxlist[foundIndex].c2min) / 2;
                        boxlist[foundIndex].c2max = lb;
                        boxlist[numboxes].c2min = lb + 1;
                        break;
                }

                /* Update stats for boxes */
                update_box(boxlist, foundIndex);
                update_box(boxlist, numboxes);
                numboxes++;
            }

            return numboxes;
        }

        /*
         * Next we have the really interesting routines: selection of a colormap
         * given the completed histogram.
         * These routines work with a list of "boxes", each representing a rectangular
         * subset of the input color space (to histogram precision).
         */

        /// <summary>
        /// Find the splittable box with the largest color population
        /// Returns null if no splittable boxes remain
        /// </summary>
        private static int find_biggest_color_pop(box[] boxlist, int numboxes)
        {
            long maxc = 0;
            int which = -1;
            for (int i = 0; i < numboxes; i++)
            {
                if (boxlist[i].colorcount > maxc && boxlist[i].volume > 0)
                {
                    which = i;
                    maxc = boxlist[i].colorcount;
                }
            }

            return which;
        }

        /// <summary>
        /// Find the splittable box with the largest (scaled) volume
        /// Returns null if no splittable boxes remain
        /// </summary>
        private static int find_biggest_volume(box[] boxlist, int numboxes)
        {
            int maxv = 0;
            int which = -1;
            for (int i = 0; i < numboxes; i++)
            {
                if (boxlist[i].volume > maxv)
                {
                    which = i;
                    maxv = boxlist[i].volume;
                }
            }

            return which;
        }

        /// <summary>
        /// Shrink the min/max bounds of a box to enclose only nonzero elements,
        /// and recompute its volume and population
        /// </summary>
        private void update_box(box[] boxlist, int boxIndex)
        {
            box curBox = boxlist[boxIndex];
            bool have_c0min = false;

            if (curBox.c0max > curBox.c0min)
            {
                for (int c0 = curBox.c0min; c0 <= curBox.c0max; c0++)
                {
                    for (int c1 = curBox.c1min; c1 <= curBox.c1max; c1++)
                    {
                        int histogramIndex = c1 * HIST_C2_ELEMS + curBox.c2min;
                        for (int c2 = curBox.c2min; c2 <= curBox.c2max; c2++)
                        {
                            if (m_histogram[c0][histogramIndex++] != 0)
                            {
                                curBox.c0min = c0;
                                have_c0min = true;
                                break;
                            }
                        }

                        if (have_c0min)
                            break;
                    }

                    if (have_c0min)
                        break;
                }
            }

            bool have_c0max = false;
            if (curBox.c0max > curBox.c0min)
            {
                for (int c0 = curBox.c0max; c0 >= curBox.c0min; c0--)
                {
                    for (int c1 = curBox.c1min; c1 <= curBox.c1max; c1++)
                    {
                        int histogramIndex = c1 * HIST_C2_ELEMS + curBox.c2min;
                        for (int c2 = curBox.c2min; c2 <= curBox.c2max; c2++)
                        {
                            if (m_histogram[c0][histogramIndex++] != 0)
                            {
                                curBox.c0max = c0;
                                have_c0max = true;
                                break;
                            }
                        }

                        if (have_c0max)
                            break;
                    }

                    if (have_c0max)
                        break;
                }
            }

            bool have_c1min = false;
            if (curBox.c1max > curBox.c1min)
            {
                for (int c1 = curBox.c1min; c1 <= curBox.c1max; c1++)
                {
                    for (int c0 = curBox.c0min; c0 <= curBox.c0max; c0++)
                    {
                        int histogramIndex = c1 * HIST_C2_ELEMS + curBox.c2min;
                        for (int c2 = curBox.c2min; c2 <= curBox.c2max; c2++)
                        {
                            if (m_histogram[c0][histogramIndex++] != 0)
                            {
                                curBox.c1min = c1;
                                have_c1min = true;
                                break;
                            }
                        }

                        if (have_c1min)
                            break;
                    }

                    if (have_c1min)
                        break;
                }
            }

            bool have_c1max = false;
            if (curBox.c1max > curBox.c1min)
            {
                for (int c1 = curBox.c1max; c1 >= curBox.c1min; c1--)
                {
                    for (int c0 = curBox.c0min; c0 <= curBox.c0max; c0++)
                    {
                        int histogramIndex = c1 * HIST_C2_ELEMS + curBox.c2min;
                        for (int c2 = curBox.c2min; c2 <= curBox.c2max; c2++)
                        {
                            if (m_histogram[c0][histogramIndex++] != 0)
                            {
                                curBox.c1max = c1;
                                have_c1max = true;
                                break;
                            }
                        }

                        if (have_c1max)
                            break;
                    }

                    if (have_c1max)
                        break;
                }
            }

            bool have_c2min = false;
            if (curBox.c2max > curBox.c2min)
            {
                for (int c2 = curBox.c2min; c2 <= curBox.c2max; c2++)
                {
                    for (int c0 = curBox.c0min; c0 <= curBox.c0max; c0++)
                    {
                        int histogramIndex = curBox.c1min * HIST_C2_ELEMS + c2;
                        for (int c1 = curBox.c1min; c1 <= curBox.c1max; c1++, histogramIndex += HIST_C2_ELEMS)
                        {
                            if (m_histogram[c0][histogramIndex] != 0)
                            {
                                curBox.c2min = c2;
                                have_c2min = true;
                                break;
                            }
                        }

                        if (have_c2min)
                            break;
                    }

                    if (have_c2min)
                        break;
                }
            }

            bool have_c2max = false;
            if (curBox.c2max > curBox.c2min)
            {
                for (int c2 = curBox.c2max; c2 >= curBox.c2min; c2--)
                {
                    for (int c0 = curBox.c0min; c0 <= curBox.c0max; c0++)
                    {
                        int histogramIndex = curBox.c1min * HIST_C2_ELEMS + c2;
                        for (int c1 = curBox.c1min; c1 <= curBox.c1max; c1++, histogramIndex += HIST_C2_ELEMS)
                        {
                            if (m_histogram[c0][histogramIndex] != 0)
                            {
                                curBox.c2max = c2;
                                have_c2max = true;
                                break;
                            }
                        }

                        if (have_c2max)
                            break;
                    }

                    if (have_c2max)
                        break;
                }
            }

            /* Update box volume.
             * We use 2-norm rather than real volume here; this biases the method
             * against making long narrow boxes, and it has the side benefit that
             * a box is splittable iff norm > 0.
             * Since the differences are expressed in histogram-cell units,
             * we have to shift back to byte units to get consistent distances;
             * after which, we scale according to the selected distance scale factors.
             */
            int dist0 = ((curBox.c0max - curBox.c0min) << C0_SHIFT) * R_SCALE;
            int dist1 = ((curBox.c1max - curBox.c1min) << C1_SHIFT) * G_SCALE;
            int dist2 = ((curBox.c2max - curBox.c2min) << C2_SHIFT) * B_SCALE;
            curBox.volume = dist0 * dist0 + dist1 * dist1 + dist2 * dist2;

            /* Now scan remaining volume of box and compute population */
            long ccount = 0;
            for (int c0 = curBox.c0min; c0 <= curBox.c0max; c0++)
            {
                for (int c1 = curBox.c1min; c1 <= curBox.c1max; c1++)
                {
                    int histogramIndex = c1 * HIST_C2_ELEMS + curBox.c2min;
                    for (int c2 = curBox.c2min; c2 <= curBox.c2max; c2++, histogramIndex++)
                    {
                        if (m_histogram[c0][histogramIndex] != 0)
                            ccount++;
                    }
                }
            }

            curBox.colorcount = ccount;
            boxlist[boxIndex] = curBox;
        }

        /// <summary>
        /// Initialize the error-limiting transfer function (lookup table).
        /// The raw F-S error computation can potentially compute error values of up to
        /// +- MaxSampleValue.  But we want the maximum correction applied to a pixel to be
        /// much less, otherwise obviously wrong pixels will be created.  (Typical
        /// effects include weird fringes at color-area boundaries, isolated bright
        /// pixels in a dark area, etc.)  The standard advice for avoiding this problem
        /// is to ensure that the "corners" of the color cube are allocated as output
        /// colors; then repeated errors in the same direction cannot cause cascading
        /// error buildup.  However, that only prevents the error from getting
        /// completely out of hand; Aaron Giles reports that error limiting improves
        /// the results even with corner colors allocated.
        /// A simple clamping of the error values to about +- MaxSampleValue/8 works pretty
        /// well, but the smoother transfer function used below is even better.  Thanks
        /// to Aaron Giles for this idea.
        /// </summary>
        private void init_error_limit()
        {
            m_error_limiter = new int[JpegConstants.MaxSampleValue * 2 + 1];
            int tableOffset = JpegConstants.MaxSampleValue;

            const int STEPSIZE = ((JpegConstants.MaxSampleValue + 1) / 16);

            /* Map errors 1:1 up to +- MaxSampleValue/16 */
            int output = 0;
            int input = 0;
            for (; input < STEPSIZE; input++, output++)
            {
                m_error_limiter[tableOffset + input] = output;
                m_error_limiter[tableOffset - input] = -output;
            }

            /* Map errors 1:2 up to +- 3*MaxSampleValue/16 */
            for (; input < STEPSIZE * 3; input++)
            {
                m_error_limiter[tableOffset + input] = output;
                m_error_limiter[tableOffset - input] = -output;
                output += (input & 1) != 0 ? 1 : 0;
            }

            /* Clamp the rest to final output value (which is (MaxSampleValue+1)/8) */
            for (; input <= JpegConstants.MaxSampleValue; input++)
            {
                m_error_limiter[tableOffset + input] = output;
                m_error_limiter[tableOffset - input] = -output;
            }
        }

        /*
         * These routines are concerned with the time-critical task of mapping input
         * colors to the nearest color in the selected colormap.
         *
         * We re-use the histogram space as an "inverse color map", essentially a
         * cache for the results of nearest-color searches.  All colors within a
         * histogram cell will be mapped to the same colormap entry, namely the one
         * closest to the cell's center.  This may not be quite the closest entry to
         * the actual input color, but it's almost as good.  A zero in the cache
         * indicates we haven't found the nearest color for that cell yet; the array
         * is cleared to zeroes before starting the mapping pass.  When we find the
         * nearest color for a cell, its colormap index plus one is recorded in the
         * cache for future use.  The pass2 scanning routines call fill_inverse_cmap
         * when they need to use an unfilled entry in the cache.
         *
         * Our method of efficiently finding nearest colors is based on the "locally
         * sorted search" idea described by Heckbert and on the incremental distance
         * calculation described by Spencer W. Thomas in chapter III.1 of Graphics
         * Gems II (James Arvo, ed.  Academic Press, 1991).  Thomas points out that
         * the distances from a given colormap entry to each cell of the histogram can
         * be computed quickly using an incremental method: the differences between
         * distances to adjacent cells themselves differ by a constant.  This allows a
         * fairly fast implementation of the "brute force" approach of computing the
         * distance from every colormap entry to every histogram cell.  Unfortunately,
         * it needs a work array to hold the best-distance-so-far for each histogram
         * cell (because the inner loop has to be over cells, not colormap entries).
         * The work array elements have to be ints, so the work array would need
         * 256Kb at our recommended precision.  This is not feasible in DOS machines.
         *
         * To get around these problems, we apply Thomas' method to compute the
         * nearest colors for only the cells within a small subbox of the histogram.
         * The work array need be only as big as the subbox, so the memory usage
         * problem is solved.  Furthermore, we need not fill subboxes that are never
         * referenced in pass2; many images use only part of the color gamut, so a
         * fair amount of work is saved.  An additional advantage of this
         * approach is that we can apply Heckbert's locality criterion to quickly
         * eliminate colormap entries that are far away from the subbox; typically
         * three-fourths of the colormap entries are rejected by Heckbert's criterion,
         * and we need not compute their distances to individual cells in the subbox.
         * The speed of this approach is heavily influenced by the subbox size: too
         * small means too much overhead, too big loses because Heckbert's criterion
         * can't eliminate as many colormap entries.  Empirically the best subbox
         * size seems to be about 1/512th of the histogram (1/8th in each direction).
         *
         * Thomas' article also describes a refined method which is asymptotically
         * faster than the brute-force method, but it is also far more complex and
         * cannot efficiently be applied to small subboxes.  It is therefore not
         * useful for programs intended to be portable to DOS machines.  On machines
         * with plenty of memory, filling the whole histogram in one shot with Thomas'
         * refined method might be faster than the present code --- but then again,
         * it might not be any faster, and it's certainly more complicated.
         */

        /*
         * The next three routines implement inverse colormap filling.  They could
         * all be folded into one big routine, but splitting them up this way saves
         * some stack space (the mindist[] and bestdist[] arrays need not coexist)
         * and may allow some compilers to produce better code by registerizing more
         * inner-loop variables.
         */

        /// <summary>
        /// Locate the colormap entries close enough to an update box to be candidates
        /// for the nearest entry to some cell(s) in the update box.  The update box
        /// is specified by the center coordinates of its first cell.  The number of
        /// candidate colormap entries is returned, and their colormap indexes are
        /// placed in colorlist[].
        /// This routine uses Heckbert's "locally sorted search" criterion to select
        /// the colors that need further consideration.
        /// </summary>
        private int find_nearby_colors(int minc0, int minc1, int minc2, byte[] colorlist)
        {
            /* Compute true coordinates of update box's upper corner and center.
             * Actually we compute the coordinates of the center of the upper-corner
             * histogram cell, which are the upper bounds of the volume we care about.
             * Note that since ">>" rounds down, the "center" values may be closer to
             * min than to max; hence comparisons to them must be "<=", not "<".
             */
            int maxc0 = minc0 + ((1 << BOX_C0_SHIFT) - (1 << C0_SHIFT));
            int centerc0 = (minc0 + maxc0) >> 1;

            int maxc1 = minc1 + ((1 << BOX_C1_SHIFT) - (1 << C1_SHIFT));
            int centerc1 = (minc1 + maxc1) >> 1;

            int maxc2 = minc2 + ((1 << BOX_C2_SHIFT) - (1 << C2_SHIFT));
            int centerc2 = (minc2 + maxc2) >> 1;

            /* For each color in colormap, find:
             *  1. its minimum squared-distance to any point in the update box
             *     (zero if color is within update box);
             *  2. its maximum squared-distance to any point in the update box.
             * Both of these can be found by considering only the corners of the box.
             * We save the minimum distance for each color in mindist[];
             * only the smallest maximum distance is of interest.
             */
            int minmaxdist = 0x7FFFFFFF;
            int[] mindist = new int[MAXNUMCOLORS];    /* min distance to colormap entry i */

            for (int i = 0; i < m_cinfo.m_actual_number_of_colors; i++)
            {
                /* We compute the squared-c0-distance term, then add in the other two. */
                int x = m_cinfo.m_colormap[0][i];
                int min_dist;
                int max_dist;

                if (x < minc0)
                {
                    int tdist = (x - minc0) * R_SCALE;
                    min_dist = tdist * tdist;
                    tdist = (x - maxc0) * R_SCALE;
                    max_dist = tdist * tdist;
                }
                else if (x > maxc0)
                {
                    int tdist = (x - maxc0) * R_SCALE;
                    min_dist = tdist * tdist;
                    tdist = (x - minc0) * R_SCALE;
                    max_dist = tdist * tdist;
                }
                else
                {
                    /* within cell range so no contribution to min_dist */
                    min_dist = 0;
                    if (x <= centerc0)
                    {
                        int tdist = (x - maxc0) * R_SCALE;
                        max_dist = tdist * tdist;
                    }
                    else
                    {
                        int tdist = (x - minc0) * R_SCALE;
                        max_dist = tdist * tdist;
                    }
                }

                x = m_cinfo.m_colormap[1][i];
                if (x < minc1)
                {
                    int tdist = (x - minc1) * G_SCALE;
                    min_dist += tdist * tdist;
                    tdist = (x - maxc1) * G_SCALE;
                    max_dist += tdist * tdist;
                }
                else if (x > maxc1)
                {
                    int tdist = (x - maxc1) * G_SCALE;
                    min_dist += tdist * tdist;
                    tdist = (x - minc1) * G_SCALE;
                    max_dist += tdist * tdist;
                }
                else
                {
                    /* within cell range so no contribution to min_dist */
                    if (x <= centerc1)
                    {
                        int tdist = (x - maxc1) * G_SCALE;
                        max_dist += tdist * tdist;
                    }
                    else
                    {
                        int tdist = (x - minc1) * G_SCALE;
                        max_dist += tdist * tdist;
                    }
                }

                x = m_cinfo.m_colormap[2][i];
                if (x < minc2)
                {
                    int tdist = (x - minc2) * B_SCALE;
                    min_dist += tdist * tdist;
                    tdist = (x - maxc2) * B_SCALE;
                    max_dist += tdist * tdist;
                }
                else if (x > maxc2)
                {
                    int tdist = (x - maxc2) * B_SCALE;
                    min_dist += tdist * tdist;
                    tdist = (x - minc2) * B_SCALE;
                    max_dist += tdist * tdist;
                }
                else
                {
                    /* within cell range so no contribution to min_dist */
                    if (x <= centerc2)
                    {
                        int tdist = (x - maxc2) * B_SCALE;
                        max_dist += tdist * tdist;
                    }
                    else
                    {
                        int tdist = (x - minc2) * B_SCALE;
                        max_dist += tdist * tdist;
                    }
                }

                mindist[i] = min_dist;  /* save away the results */
                if (max_dist < minmaxdist)
                    minmaxdist = max_dist;
            }

            /* Now we know that no cell in the update box is more than minmaxdist
             * away from some colormap entry.  Therefore, only colors that are
             * within minmaxdist of some part of the box need be considered.
             */
            int ncolors = 0;
            for (int i = 0; i < m_cinfo.m_actual_number_of_colors; i++)
            {
                if (mindist[i] <= minmaxdist)
                    colorlist[ncolors++] = (byte)i;
            }

            return ncolors;
        }

        /// <summary>
        /// Find the closest colormap entry for each cell in the update box,
        /// given the list of candidate colors prepared by find_nearby_colors.
        /// Return the indexes of the closest entries in the bestcolor[] array.
        /// This routine uses Thomas' incremental distance calculation method to
        /// find the distance from a colormap entry to successive cells in the box.
        /// </summary>
        private void find_best_colors(int minc0, int minc1, int minc2, int numcolors, byte[] colorlist, byte[] bestcolor)
        {
            /* Nominal steps between cell centers ("x" in Thomas article) */
            const int STEP_C0 = ((1 << C0_SHIFT) * R_SCALE);
            const int STEP_C1 = ((1 << C1_SHIFT) * G_SCALE);
            const int STEP_C2 = ((1 << C2_SHIFT) * B_SCALE);

            /* This array holds the distance to the nearest-so-far color for each cell */
            int[] bestdist = new int[BOX_C0_ELEMS * BOX_C1_ELEMS * BOX_C2_ELEMS];

            /* Initialize best-distance for each cell of the update box */
            int bestIndex = 0;
            for (int i = BOX_C0_ELEMS * BOX_C1_ELEMS * BOX_C2_ELEMS - 1; i >= 0; i--)
            {
                bestdist[bestIndex] = 0x7FFFFFFF;
                bestIndex++;
            }

            /* For each color selected by find_nearby_colors,
             * compute its distance to the center of each cell in the box.
             * If that's less than best-so-far, update best distance and color number.
             */
            for (int i = 0; i < numcolors; i++)
            {
                int icolor = colorlist[i];

                /* Compute (square of) distance from minc0/c1/c2 to this color */
                int inc0 = (minc0 - m_cinfo.m_colormap[0][icolor]) * R_SCALE;
                int dist0 = inc0 * inc0;

                int inc1 = (minc1 - m_cinfo.m_colormap[1][icolor]) * G_SCALE;
                dist0 += inc1 * inc1;

                int inc2 = (minc2 - m_cinfo.m_colormap[2][icolor]) * B_SCALE;
                dist0 += inc2 * inc2;

                /* Form the initial difference increments */
                inc0 = inc0 * (2 * STEP_C0) + STEP_C0 * STEP_C0;
                inc1 = inc1 * (2 * STEP_C1) + STEP_C1 * STEP_C1;
                inc2 = inc2 * (2 * STEP_C2) + STEP_C2 * STEP_C2;

                /* Now loop over all cells in box, updating distance per Thomas method */
                bestIndex = 0;
                int colorIndex = 0;
                int xx0 = inc0;
                for (int ic0 = BOX_C0_ELEMS - 1; ic0 >= 0; ic0--)
                {
                    int dist1 = dist0;
                    int xx1 = inc1;
                    for (int ic1 = BOX_C1_ELEMS - 1; ic1 >= 0; ic1--)
                    {
                        int dist2 = dist1;
                        int xx2 = inc2;
                        for (int ic2 = BOX_C2_ELEMS - 1; ic2 >= 0; ic2--)
                        {
                            if (dist2 < bestdist[bestIndex])
                            {
                                bestdist[bestIndex] = dist2;
                                bestcolor[colorIndex] = (byte)icolor;
                            }

                            dist2 += xx2;
                            xx2 += 2 * STEP_C2 * STEP_C2;
                            bestIndex++;
                            colorIndex++;
                        }

                        dist1 += xx1;
                        xx1 += 2 * STEP_C1 * STEP_C1;
                    }

                    dist0 += xx0;
                    xx0 += 2 * STEP_C0 * STEP_C0;
                }
            }
        }

        /// <summary>
        /// Fill the inverse-colormap entries in the update box that contains
        /// histogram cell c0/c1/c2.  (Only that one cell MUST be filled, but
        /// we can fill as many others as we wish.)
        /// </summary>
        private void fill_inverse_cmap(int c0, int c1, int c2)
        {
            /* Convert cell coordinates to update box ID */
            c0 >>= BOX_C0_LOG;
            c1 >>= BOX_C1_LOG;
            c2 >>= BOX_C2_LOG;

            /* Compute true coordinates of update box's origin corner.
             * Actually we compute the coordinates of the center of the corner
             * histogram cell, which are the lower bounds of the volume we care about.
             */
            int minc0 = (c0 << BOX_C0_SHIFT) + ((1 << C0_SHIFT) >> 1);
            int minc1 = (c1 << BOX_C1_SHIFT) + ((1 << C1_SHIFT) >> 1);
            int minc2 = (c2 << BOX_C2_SHIFT) + ((1 << C2_SHIFT) >> 1);

            /* Determine which colormap entries are close enough to be candidates
             * for the nearest entry to some cell in the update box.
             */
            /* This array lists the candidate colormap indexes. */
            byte[] colorlist = new byte[MAXNUMCOLORS];
            int numcolors = find_nearby_colors(minc0, minc1, minc2, colorlist);

            /* Determine the actually nearest colors. */
            /* This array holds the actually closest colormap index for each cell. */
            byte[] bestcolor = new byte[BOX_C0_ELEMS * BOX_C1_ELEMS * BOX_C2_ELEMS];
            find_best_colors(minc0, minc1, minc2, numcolors, colorlist, bestcolor);

            /* Save the best color numbers (plus 1) in the main cache array */
            c0 <<= BOX_C0_LOG;      /* convert ID back to base cell indexes */
            c1 <<= BOX_C1_LOG;
            c2 <<= BOX_C2_LOG;
            int bestcolorIndex = 0;
            for (int ic0 = 0; ic0 < BOX_C0_ELEMS; ic0++)
            {
                for (int ic1 = 0; ic1 < BOX_C1_ELEMS; ic1++)
                {
                    int histogramIndex = (c1 + ic1) * HIST_C2_ELEMS + c2;
                    for (int ic2 = 0; ic2 < BOX_C2_ELEMS; ic2++)
                    {
                        m_histogram[c0 + ic0][histogramIndex] = (ushort)((int)bestcolor[bestcolorIndex] + 1);
                        histogramIndex++;
                        bestcolorIndex++;
                    }
                }
            }
        }
    }
    #endregion

    #region ProgressiveHuffmanDecoder
    /// <summary>
    /// Expanded entropy decoder object for progressive Huffman decoding.
    /// 
    /// The savable_state sub-record contains fields that change within an MCU,
    /// but must not be updated permanently until we complete the MCU.
    /// </summary>
    class ProgressiveHuffmanDecoder : JpegEntropyDecoder
    {
        private class savable_state
        {
            //savable_state operator=(savable_state src);
            public int EOBRUN;            /* remaining EOBs in EOBRUN */
            public int[] last_dc_val = new int[JpegConstants.MaxComponentsInScan]; /* last DC coef for each component */

            public void Assign(savable_state ss)
            {
                EOBRUN = ss.EOBRUN;
                Buffer.BlockCopy(ss.last_dc_val, 0, last_dc_val, 0, last_dc_val.Length * sizeof(int));
            }
        }

        private enum MCUDecoder
        {
            mcu_DC_first_decoder,
            mcu_AC_first_decoder,
            mcu_DC_refine_decoder,
            mcu_AC_refine_decoder
        }

        private MCUDecoder m_decoder;

        /* These fields are loaded into local variables at start of each MCU.
        * In case of suspension, we exit WITHOUT updating them.
        */
        private SavedBitreadState m_bitstate;    /* Bit buffer at start of MCU */
        private savable_state m_saved = new savable_state();        /* Other state at start of MCU */

        /* These fields are NOT loaded into local working state. */
        private int m_restarts_to_go;    /* MCUs left in this restart interval */

        /* Pointers to derived tables (these workspaces have image lifespan) */
        private DerivedTable[] m_derived_tbls = new DerivedTable[JpegConstants.NumberOfHuffmanTables];

        private DerivedTable m_ac_derived_tbl; /* active table during an AC scan */

        public ProgressiveHuffmanDecoder(JpegDecompressor cinfo)
        {
            m_cinfo = cinfo;

            /* Mark derived tables unallocated */
            for (int i = 0; i < JpegConstants.NumberOfHuffmanTables; i++)
                m_derived_tbls[i] = null;

            /* Create progression status table */
            cinfo.m_coef_bits = new int[cinfo.m_num_components][];
            for (int i = 0; i < cinfo.m_num_components; i++)
                cinfo.m_coef_bits[i] = new int[JpegConstants.DCTSize2];

            for (int ci = 0; ci < cinfo.m_num_components; ci++)
            {
                for (int i = 0; i < JpegConstants.DCTSize2; i++)
                    cinfo.m_coef_bits[ci][i] = -1;
            }
        }

        /// <summary>
        /// Initialize for a Huffman-compressed scan.
        /// </summary>
        public override void start_pass()
        {
            /* Validate scan parameters */
            bool bad = false;
            bool is_DC_band = (m_cinfo.m_Ss == 0);
            if (is_DC_band)
            {
                if (m_cinfo.m_Se != 0)
                    bad = true;
            }
            else
            {
                /* need not check Ss/Se < 0 since they came from unsigned bytes */
                if (m_cinfo.m_Ss > m_cinfo.m_Se || m_cinfo.m_Se >= JpegConstants.DCTSize2)
                    bad = true;

                /* AC scans may have only one component */
                if (m_cinfo.m_comps_in_scan != 1)
                    bad = true;
            }

            if (m_cinfo.m_Ah != 0)
            {
                /* Successive approximation refinement scan: must have Al = Ah-1. */
                if (m_cinfo.m_Al != m_cinfo.m_Ah - 1)
                    bad = true;
            }

            if (m_cinfo.m_Al > 13)
            {
                /* need not check for < 0 */
                bad = true;
            }

            /* Arguably the maximum Al value should be less than 13 for 8-bit precision,
             * but the spec doesn't say so, and we try to be liberal about what we
             * accept.  Note: large Al values could result in out-of-range DC
             * coefficients during early scans, leading to bizarre displays due to
             * overflows in the IDCT math.  But we won't crash.
             */
            if (bad)
                throw new Exception(String.Format("Invalid progressive parameters Ss={0} Se={1} Ah={2} Al={3}", m_cinfo.m_Ss, m_cinfo.m_Se, m_cinfo.m_Ah, m_cinfo.m_Al));

            /* Update progression status, and verify that scan order is legal.
             * Note that inter-scan inconsistencies are treated as warnings
             * not fatal errors ... not clear if this is right way to behave.
             */
            for (int ci = 0; ci < m_cinfo.m_comps_in_scan; ci++)
            {
                int cindex = m_cinfo.Comp_info[m_cinfo.m_cur_comp_info[ci]].Component_index;

                for (int coefi = m_cinfo.m_Ss; coefi <= m_cinfo.m_Se; coefi++)
                {
                    int expected = m_cinfo.m_coef_bits[cindex][coefi];
                    if (expected < 0)
                        expected = 0;

                    m_cinfo.m_coef_bits[cindex][coefi] = m_cinfo.m_Al;
                }
            }

            /* Select MCU decoding routine */
            if (m_cinfo.m_Ah == 0)
            {
                if (is_DC_band)
                    m_decoder = MCUDecoder.mcu_DC_first_decoder;
                else
                    m_decoder = MCUDecoder.mcu_AC_first_decoder;
            }
            else
            {
                if (is_DC_band)
                    m_decoder = MCUDecoder.mcu_DC_refine_decoder;
                else
                    m_decoder = MCUDecoder.mcu_AC_refine_decoder;
            }

            for (int ci = 0; ci < m_cinfo.m_comps_in_scan; ci++)
            {
                JpegComponent componentInfo = m_cinfo.Comp_info[m_cinfo.m_cur_comp_info[ci]];
                /* Make sure requested tables are present, and compute derived tables.
                 * We may build same derived table more than once, but it's not expensive.
                 */
                if (is_DC_band)
                {
                    if (m_cinfo.m_Ah == 0)
                    {
                        /* DC refinement needs no table */
                        jpeg_make_d_derived_tbl(true, componentInfo.Dc_tbl_no, ref m_derived_tbls[componentInfo.Dc_tbl_no]);
                    }
                }
                else
                {
                    jpeg_make_d_derived_tbl(false, componentInfo.Ac_tbl_no, ref m_derived_tbls[componentInfo.Ac_tbl_no]);

                    /* remember the single active table */
                    m_ac_derived_tbl = m_derived_tbls[componentInfo.Ac_tbl_no];
                }

                /* Initialize DC predictions to 0 */
                m_saved.last_dc_val[ci] = 0;
            }

            /* Initialize bitread state variables */
            m_bitstate.bits_left = 0;
            m_bitstate.get_buffer = 0; /* unnecessary, but keeps Purify quiet */
            m_insufficient_data = false;

            /* Initialize private state variables */
            m_saved.EOBRUN = 0;

            /* Initialize restart counter */
            m_restarts_to_go = m_cinfo.m_restart_interval;
        }

        public override bool decode_mcu(JpegBlock[] MCU_data)
        {
            switch (m_decoder)
            {
                case MCUDecoder.mcu_DC_first_decoder:
                    return decode_mcu_DC_first(MCU_data);
                case MCUDecoder.mcu_AC_first_decoder:
                    return decode_mcu_AC_first(MCU_data);
                case MCUDecoder.mcu_DC_refine_decoder:
                    return decode_mcu_DC_refine(MCU_data);
                case MCUDecoder.mcu_AC_refine_decoder:
                    return decode_mcu_AC_refine(MCU_data);
            }

            throw new Exception("The specified MCUDecoder is not implemented!");
        }

        /*
         * Huffman MCU decoding.
         * Each of these routines decodes and returns one MCU's worth of
         * Huffman-compressed coefficients. 
         * The coefficients are reordered from zigzag order into natural array order,
         * but are not de-quantized.
         *
         * The i'th block of the MCU is stored into the block pointed to by
         * MCU_data[i].  WE ASSUME THIS AREA IS INITIALLY ZEROED BY THE CALLER.
         *
         * We return false if data source requested suspension.  In that case no
         * changes have been made to permanent state.  (Exception: some output
         * coefficients may already have been assigned.  This is harmless for
         * spectral selection, since we'll just re-assign them on the next call.
         * Successive approximation AC refinement has to be more careful, however.)
         */

        /// <summary>
        /// MCU decoding for DC initial scan (either spectral selection,
        /// or first pass of successive approximation).
        /// </summary>
        private bool decode_mcu_DC_first(JpegBlock[] MCU_data)
        {
            /* Process restart marker if needed; may have to suspend */
            if (m_cinfo.m_restart_interval != 0)
            {
                if (m_restarts_to_go == 0)
                {
                    if (!process_restart())
                        return false;
                }
            }

            /* If we've run out of data, just leave the MCU set to zeroes.
             * This way, we return uniform gray for the remainder of the segment.
             */
            if (!m_insufficient_data)
            {
                /* Load up working state */
                int get_buffer;
                int bits_left;
                WorkingBitreadState br_state = new WorkingBitreadState();
                BITREAD_LOAD_STATE(m_bitstate, out get_buffer, out bits_left, ref br_state);
                savable_state state = new savable_state();
                state.Assign(m_saved);

                /* Outer loop handles each block in the MCU */
                for (int blkn = 0; blkn < m_cinfo.m_blocks_in_MCU; blkn++)
                {
                    int ci = m_cinfo.m_MCU_membership[blkn];

                    /* Decode a single block's worth of coefficients */

                    /* Section F.2.2.1: decode the DC coefficient difference */
                    int s;
                    if (!HUFF_DECODE(out s, ref br_state, m_derived_tbls[m_cinfo.Comp_info[m_cinfo.m_cur_comp_info[ci]].Dc_tbl_no], ref get_buffer, ref bits_left))
                        return false;

                    if (s != 0)
                    {
                        if (!CHECK_BIT_BUFFER(ref br_state, s, ref get_buffer, ref bits_left))
                            return false;

                        int r = GET_BITS(s, get_buffer, ref bits_left);
                        s = HUFF_EXTEND(r, s);
                    }

                    /* Convert DC difference to actual value, update last_dc_val */
                    s += state.last_dc_val[ci];
                    state.last_dc_val[ci] = s;

                    /* Scale and output the coefficient (assumes jpeg_natural_order[0]=0) */
                    MCU_data[blkn][0] = (short)(s << m_cinfo.m_Al);
                }

                /* Completed MCU, so update state */
                BITREAD_SAVE_STATE(ref m_bitstate, get_buffer, bits_left);
                m_saved.Assign(state);
            }

            /* Account for restart interval (no-op if not using restarts) */
            m_restarts_to_go--;

            return true;
        }

        /// <summary>
        /// MCU decoding for AC initial scan (either spectral selection,
        /// or first pass of successive approximation).
        /// </summary>
        private bool decode_mcu_AC_first(JpegBlock[] MCU_data)
        {
            /* Process restart marker if needed; may have to suspend */
            if (m_cinfo.m_restart_interval != 0)
            {
                if (m_restarts_to_go == 0)
                {
                    if (!process_restart())
                        return false;
                }
            }

            /* If we've run out of data, just leave the MCU set to zeroes.
             * This way, we return uniform gray for the remainder of the segment.
             */
            if (!m_insufficient_data)
            {
                /* Load up working state.
                 * We can avoid loading/saving bitread state if in an EOB run.
                 */
                int EOBRUN = m_saved.EOBRUN; /* only part of saved state we need */

                /* There is always only one block per MCU */

                if (EOBRUN > 0)
                {
                    /* if it's a band of zeroes... */
                    /* ...process it now (we do nothing) */
                    EOBRUN--;
                }
                else
                {
                    int get_buffer;
                    int bits_left;
                    WorkingBitreadState br_state = new WorkingBitreadState();
                    BITREAD_LOAD_STATE(m_bitstate, out get_buffer, out bits_left, ref br_state);

                    for (int k = m_cinfo.m_Ss; k <= m_cinfo.m_Se; k++)
                    {
                        int s;
                        if (!HUFF_DECODE(out s, ref br_state, m_ac_derived_tbl, ref get_buffer, ref bits_left))
                            return false;

                        int r = s >> 4;
                        s &= 15;
                        if (s != 0)
                        {
                            k += r;

                            if (!CHECK_BIT_BUFFER(ref br_state, s, ref get_buffer, ref bits_left))
                                return false;

                            r = GET_BITS(s, get_buffer, ref bits_left);
                            s = HUFF_EXTEND(r, s);

                            /* Scale and output coefficient in natural (dezigzagged) order */
                            MCU_data[0][JpegUtils.jpeg_natural_order[k]] = (short)(s << m_cinfo.m_Al);
                        }
                        else
                        {
                            if (r == 15)
                            {
                                /* ZRL */
                                k += 15;        /* skip 15 zeroes in band */
                            }
                            else
                            {
                                /* EOBr, run length is 2^r + appended bits */
                                EOBRUN = 1 << r;
                                if (r != 0)
                                {
                                    /* EOBr, r > 0 */
                                    if (!CHECK_BIT_BUFFER(ref br_state, r, ref get_buffer, ref bits_left))
                                        return false;

                                    r = GET_BITS(r, get_buffer, ref bits_left);
                                    EOBRUN += r;
                                }

                                EOBRUN--;       /* this band is processed at this moment */
                                break;      /* force end-of-band */
                            }
                        }
                    }

                    BITREAD_SAVE_STATE(ref m_bitstate, get_buffer, bits_left);
                }

                /* Completed MCU, so update state */
                m_saved.EOBRUN = EOBRUN; /* only part of saved state we need */
            }

            /* Account for restart interval (no-op if not using restarts) */
            m_restarts_to_go--;

            return true;
        }

        /// <summary>
        /// MCU decoding for DC successive approximation refinement scan.
        /// Note: we assume such scans can be multi-component, although the spec
        /// is not very clear on the point.
        /// </summary>
        private bool decode_mcu_DC_refine(JpegBlock[] MCU_data)
        {
            /* Process restart marker if needed; may have to suspend */
            if (m_cinfo.m_restart_interval != 0)
            {
                if (m_restarts_to_go == 0)
                {
                    if (!process_restart())
                        return false;
                }
            }

            /* Not worth the cycles to check insufficient_data here,
             * since we will not change the data anyway if we read zeroes.
             */

            /* Load up working state */
            int get_buffer;
            int bits_left;
            WorkingBitreadState br_state = new WorkingBitreadState();
            BITREAD_LOAD_STATE(m_bitstate, out get_buffer, out bits_left, ref br_state);

            /* Outer loop handles each block in the MCU */

            for (int blkn = 0; blkn < m_cinfo.m_blocks_in_MCU; blkn++)
            {
                /* Encoded data is simply the next bit of the two's-complement DC value */
                if (!CHECK_BIT_BUFFER(ref br_state, 1, ref get_buffer, ref bits_left))
                    return false;

                if (GET_BITS(1, get_buffer, ref bits_left) != 0)
                {
                    /* 1 in the bit position being coded */
                    MCU_data[blkn][0] |= (short)(1 << m_cinfo.m_Al);
                }

                /* Note: since we use |=, repeating the assignment later is safe */
            }

            /* Completed MCU, so update state */
            BITREAD_SAVE_STATE(ref m_bitstate, get_buffer, bits_left);

            /* Account for restart interval (no-op if not using restarts) */
            m_restarts_to_go--;

            return true;
        }

        // There is always only one block per MCU
        private bool decode_mcu_AC_refine(JpegBlock[] MCU_data)
        {
            int p1 = 1 << m_cinfo.m_Al;    /* 1 in the bit position being coded */
            int m1 = -1 << m_cinfo.m_Al; /* -1 in the bit position being coded */

            /* Process restart marker if needed; may have to suspend */
            if (m_cinfo.m_restart_interval != 0)
            {
                if (m_restarts_to_go == 0)
                {
                    if (!process_restart())
                        return false;
                }
            }

            /* If we've run out of data, don't modify the MCU.
             */
            if (!m_insufficient_data)
            {
                /* Load up working state */
                int get_buffer;
                int bits_left;
                WorkingBitreadState br_state = new WorkingBitreadState();
                BITREAD_LOAD_STATE(m_bitstate, out get_buffer, out bits_left, ref br_state);
                int EOBRUN = m_saved.EOBRUN; /* only part of saved state we need */

                /* If we are forced to suspend, we must undo the assignments to any newly
                 * nonzero coefficients in the block, because otherwise we'd get confused
                 * next time about which coefficients were already nonzero.
                 * But we need not undo addition of bits to already-nonzero coefficients;
                 * instead, we can test the current bit to see if we already did it.
                 */
                int num_newnz = 0;
                int[] newnz_pos = new int[JpegConstants.DCTSize2];

                /* initialize coefficient loop counter to start of band */
                int k = m_cinfo.m_Ss;

                if (EOBRUN == 0)
                {
                    for (; k <= m_cinfo.m_Se; k++)
                    {
                        int s;
                        if (!HUFF_DECODE(out s, ref br_state, m_ac_derived_tbl, ref get_buffer, ref bits_left))
                        {
                            undo_decode_mcu_AC_refine(MCU_data, newnz_pos, num_newnz);
                            return false;
                        }

                        int r = s >> 4;
                        s &= 15;
                        if (s != 0)
                        {
                            if (!CHECK_BIT_BUFFER(ref br_state, 1, ref get_buffer, ref bits_left))
                            {
                                undo_decode_mcu_AC_refine(MCU_data, newnz_pos, num_newnz);
                                return false;
                            }

                            if (GET_BITS(1, get_buffer, ref bits_left) != 0)
                            {
                                /* newly nonzero coef is positive */
                                s = p1;
                            }
                            else
                            {
                                /* newly nonzero coef is negative */
                                s = m1;
                            }
                        }
                        else
                        {
                            if (r != 15)
                            {
                                EOBRUN = 1 << r;    /* EOBr, run length is 2^r + appended bits */
                                if (r != 0)
                                {
                                    if (!CHECK_BIT_BUFFER(ref br_state, r, ref get_buffer, ref bits_left))
                                    {
                                        undo_decode_mcu_AC_refine(MCU_data, newnz_pos, num_newnz);
                                        return false;
                                    }

                                    r = GET_BITS(r, get_buffer, ref bits_left);
                                    EOBRUN += r;
                                }
                                break;      /* rest of block is handled by EOB logic */
                            }
                            /* note s = 0 for processing ZRL */
                        }
                        /* Advance over already-nonzero coefs and r still-zero coefs,
                         * appending correction bits to the nonzeroes.  A correction bit is 1
                         * if the absolute value of the coefficient must be increased.
                         */
                        do
                        {
                            int blockIndex = JpegUtils.jpeg_natural_order[k];
                            short thiscoef = MCU_data[0][blockIndex];
                            if (thiscoef != 0)
                            {
                                if (!CHECK_BIT_BUFFER(ref br_state, 1, ref get_buffer, ref bits_left))
                                {
                                    undo_decode_mcu_AC_refine(MCU_data, newnz_pos, num_newnz);
                                    return false;
                                }

                                if (GET_BITS(1, get_buffer, ref bits_left) != 0)
                                {
                                    if ((thiscoef & p1) == 0)
                                    {
                                        /* do nothing if already set it */
                                        if (thiscoef >= 0)
                                            MCU_data[0][blockIndex] += (short)p1;
                                        else
                                            MCU_data[0][blockIndex] += (short)m1;
                                    }
                                }
                            }
                            else
                            {
                                if (--r < 0)
                                    break;      /* reached target zero coefficient */
                            }

                            k++;
                        }
                        while (k <= m_cinfo.m_Se);

                        if (s != 0)
                        {
                            int pos = JpegUtils.jpeg_natural_order[k];

                            /* Output newly nonzero coefficient */
                            MCU_data[0][pos] = (short)s;

                            /* Remember its position in case we have to suspend */
                            newnz_pos[num_newnz++] = pos;
                        }
                    }
                }

                if (EOBRUN > 0)
                {
                    /* Scan any remaining coefficient positions after the end-of-band
                     * (the last newly nonzero coefficient, if any).  Append a correction
                     * bit to each already-nonzero coefficient.  A correction bit is 1
                     * if the absolute value of the coefficient must be increased.
                     */
                    for (; k <= m_cinfo.m_Se; k++)
                    {
                        int blockIndex = JpegUtils.jpeg_natural_order[k];
                        short thiscoef = MCU_data[0][blockIndex];
                        if (thiscoef != 0)
                        {
                            if (!CHECK_BIT_BUFFER(ref br_state, 1, ref get_buffer, ref bits_left))
                            {
                                //undo_decode_mcu_AC_refine(MCU_data[0], newnz_pos, num_newnz);
                                undo_decode_mcu_AC_refine(MCU_data, newnz_pos, num_newnz);
                                return false;
                            }

                            if (GET_BITS(1, get_buffer, ref bits_left) != 0)
                            {
                                if ((thiscoef & p1) == 0)
                                {
                                    /* do nothing if already changed it */
                                    if (thiscoef >= 0)
                                        MCU_data[0][blockIndex] += (short)p1;
                                    else
                                        MCU_data[0][blockIndex] += (short)m1;
                                }
                            }
                        }
                    }

                    /* Count one block completed in EOB run */
                    EOBRUN--;
                }

                /* Completed MCU, so update state */
                BITREAD_SAVE_STATE(ref m_bitstate, get_buffer, bits_left);
                m_saved.EOBRUN = EOBRUN; /* only part of saved state we need */
            }

            /* Account for restart interval (no-op if not using restarts) */
            m_restarts_to_go--;

            return true;
        }

        /// <summary>
        /// Check for a restart marker and resynchronize decoder.
        /// Returns false if must suspend.
        /// </summary>
        private bool process_restart()
        {
            /* Throw away any unused bits remaining in bit buffer; */
            /* include any full bytes in next_marker's count of discarded bytes */
            m_cinfo.m_marker.SkipBytes(m_bitstate.bits_left / 8);
            m_bitstate.bits_left = 0;

            /* Advance past the RSTn marker */
            if (!m_cinfo.m_marker.read_restart_marker())
                return false;

            /* Re-initialize DC predictions to 0 */
            for (int ci = 0; ci < m_cinfo.m_comps_in_scan; ci++)
                m_saved.last_dc_val[ci] = 0;

            /* Re-init EOB run count, too */
            m_saved.EOBRUN = 0;

            /* Reset restart counter */
            m_restarts_to_go = m_cinfo.m_restart_interval;

            /* Reset out-of-data flag, unless read_restart_marker left us smack up
             * against a marker.  In that case we will end up treating the next data
             * segment as empty, and we can avoid producing bogus output pixels by
             * leaving the flag set.
             */
            if (m_cinfo.m_unread_marker == 0)
                m_insufficient_data = false;

            return true;
        }

        /// <summary>
        /// MCU decoding for AC successive approximation refinement scan.
        /// </summary>
        private static void undo_decode_mcu_AC_refine(JpegBlock[] block, int[] newnz_pos, int num_newnz)
        {
            /* Re-zero any output coefficients that we made newly nonzero */
            while (num_newnz > 0)
                block[0][newnz_pos[--num_newnz]] = 0;
        }
    }
    #endregion

    #region ProgressiveHuffmanEncoder
    /// <summary>
    /// Expanded entropy encoder object for progressive Huffman encoding.
    /// </summary>
    class ProgressiveHuffmanEncoder : JpegEntropyEncoder
    {
        private enum MCUEncoder
        {
            mcu_DC_first_encoder,
            mcu_AC_first_encoder,
            mcu_DC_refine_encoder,
            mcu_AC_refine_encoder
        }

        /* MAX_CORR_BITS is the number of bits the AC refinement correction-bit
         * buffer can hold.  Larger sizes may slightly improve compression, but
         * 1000 is already well into the realm of overkill.
         * The minimum safe size is 64 bits.
         */
        private const int MAX_CORR_BITS = 1000; /* Max # of correction bits I can buffer */

        private MCUEncoder m_MCUEncoder;

        /* Mode flag: true for optimization, false for actual data output */
        private bool m_gather_statistics;

        // Bit-level coding status.
        private int m_put_buffer;       /* current bit-accumulation buffer */
        private int m_put_bits;           /* # of bits now in it */

        /* Coding status for DC components */
        private int[] m_last_dc_val = new int[JpegConstants.MaxComponentsInScan]; /* last DC coef for each component */

        /* Coding status for AC components */
        private int m_ac_tbl_no;      /* the table number of the single component */
        private int m_EOBRUN;        /* run length of EOBs */
        private int m_BE;        /* # of buffered correction bits before MCU */
        private char[] m_bit_buffer;       /* buffer for correction bits (1 per char) */
        /* packing correction bits tightly would save some space but cost time... */

        private int m_restarts_to_go;    /* MCUs left in this restart interval */
        private int m_next_restart_num;       /* next restart number to write (0-7) */

        /* Pointers to derived tables (these workspaces have image lifespan).
        * Since any one scan codes only DC or only AC, we only need one set
        * of tables, not one for DC and one for AC.
        */
        private c_derived_tbl[] m_derived_tbls = new c_derived_tbl[JpegConstants.NumberOfHuffmanTables];

        /* Statistics tables for optimization; again, one set is enough */
        private long[][] m_count_ptrs = new long[JpegConstants.NumberOfHuffmanTables][];

        public ProgressiveHuffmanEncoder(JpegCompressor cinfo)
        {
            m_cinfo = cinfo;

            /* Mark tables unallocated */
            for (int i = 0; i < JpegConstants.NumberOfHuffmanTables; i++)
            {
                m_derived_tbls[i] = null;
                m_count_ptrs[i] = null;
            }
        }

        // Initialize for a Huffman-compressed scan using progressive JPEG.
        public override void start_pass(bool gather_statistics)
        {
            m_gather_statistics = gather_statistics;

            /* We assume the scan parameters are already validated. */

            /* Select execution routines */
            bool is_DC_band = (m_cinfo.m_Ss == 0);
            if (m_cinfo.m_Ah == 0)
            {
                if (is_DC_band)
                    m_MCUEncoder = MCUEncoder.mcu_DC_first_encoder;
                else
                    m_MCUEncoder = MCUEncoder.mcu_AC_first_encoder;
            }
            else
            {
                if (is_DC_band)
                {
                    m_MCUEncoder = MCUEncoder.mcu_DC_refine_encoder;
                }
                else
                {
                    m_MCUEncoder = MCUEncoder.mcu_AC_refine_encoder;

                    /* AC refinement needs a correction bit buffer */
                    if (m_bit_buffer == null)
                        m_bit_buffer = new char[MAX_CORR_BITS];
                }
            }

            /* Only DC coefficients may be interleaved, so m_cinfo.comps_in_scan = 1
             * for AC coefficients.
             */
            for (int ci = 0; ci < m_cinfo.m_comps_in_scan; ci++)
            {
                JpegComponent componentInfo = m_cinfo.Component_info[m_cinfo.m_cur_comp_info[ci]];

                /* Initialize DC predictions to 0 */
                m_last_dc_val[ci] = 0;

                /* Get table index */
                int tbl;
                if (is_DC_band)
                {
                    if (m_cinfo.m_Ah != 0) /* DC refinement needs no table */
                        continue;

                    tbl = componentInfo.Dc_tbl_no;
                }
                else
                {
                    m_ac_tbl_no = componentInfo.Ac_tbl_no;
                    tbl = componentInfo.Ac_tbl_no;
                }

                if (m_gather_statistics)
                {
                    /* Check for invalid table index */
                    /* (make_c_derived_tbl does this in the other path) */
                    if (tbl < 0 || tbl >= JpegConstants.NumberOfHuffmanTables)
                        throw new Exception(String.Format("Huffman table 0x{0:X2} was not defined", tbl));

                    /* Allocate and zero the statistics tables */
                    /* Note that jpeg_gen_optimal_table expects 257 entries in each table! */
                    if (m_count_ptrs[tbl] == null)
                        m_count_ptrs[tbl] = new long[257];

                    Array.Clear(m_count_ptrs[tbl], 0, 257);
                }
                else
                {
                    /* Compute derived values for Huffman table */
                    /* We may do this more than once for a table, but it's not expensive */
                    jpeg_make_c_derived_tbl(is_DC_band, tbl, ref m_derived_tbls[tbl]);
                }
            }

            /* Initialize AC stuff */
            m_EOBRUN = 0;
            m_BE = 0;

            /* Initialize bit buffer to empty */
            m_put_buffer = 0;
            m_put_bits = 0;

            /* Initialize restart stuff */
            m_restarts_to_go = m_cinfo.m_restart_interval;
            m_next_restart_num = 0;
        }

        public override bool encode_mcu(JpegBlock[][] MCU_data)
        {
            switch (m_MCUEncoder)
            {
                case MCUEncoder.mcu_DC_first_encoder:
                    return encode_mcu_DC_first(MCU_data);
                case MCUEncoder.mcu_AC_first_encoder:
                    return encode_mcu_AC_first(MCU_data);
                case MCUEncoder.mcu_DC_refine_encoder:
                    return encode_mcu_DC_refine(MCU_data);
                case MCUEncoder.mcu_AC_refine_encoder:
                    return encode_mcu_AC_refine(MCU_data);
            }

            throw new Exception("The specified MCUEncoder is not implemented.");
        }

        public override void finish_pass()
        {
            if (m_gather_statistics)
                finish_pass_gather_phuff();
            else
                finish_pass_phuff();
        }

        /// <summary>
        /// MCU encoding for DC initial scan (either spectral selection,
        /// or first pass of successive approximation).
        /// </summary>
        private bool encode_mcu_DC_first(JpegBlock[][] MCU_data)
        {
            /* Emit restart marker if needed */
            if (m_cinfo.m_restart_interval != 0)
            {
                if (m_restarts_to_go == 0)
                    emit_restart(m_next_restart_num);
            }

            /* Encode the MCU data blocks */
            for (int blkn = 0; blkn < m_cinfo.m_blocks_in_MCU; blkn++)
            {
                /* Compute the DC value after the required point transform by Al.
                 * This is simply an arithmetic right shift.
                 */
                int temp2 = IRIGHT_SHIFT(MCU_data[blkn][0][0], m_cinfo.m_Al);

                /* DC differences are figured on the point-transformed values. */
                int ci = m_cinfo.m_MCU_membership[blkn];
                int temp = temp2 - m_last_dc_val[ci];
                m_last_dc_val[ci] = temp2;

                /* Encode the DC coefficient difference per section G.1.2.1 */
                temp2 = temp;
                if (temp < 0)
                {
                    /* temp is abs value of input */
                    temp = -temp;

                    /* For a negative input, want temp2 = bitwise complement of abs(input) */
                    /* This code assumes we are on a two's complement machine */
                    temp2--;
                }

                /* Find the number of bits needed for the magnitude of the coefficient */
                int nbits = 0;
                while (temp != 0)
                {
                    nbits++;
                    temp >>= 1;
                }

                /* Check for out-of-range coefficient values.
                 * Since we're encoding a difference, the range limit is twice as much.
                 */
                if (nbits > MAX_HUFFMAN_COEF_BITS + 1)
                    throw new Exception("DCT coefficient out of range");

                /* Count/emit the Huffman-coded symbol for the number of bits */
                emit_symbol(m_cinfo.Component_info[m_cinfo.m_cur_comp_info[ci]].Dc_tbl_no, nbits);

                /* Emit that number of bits of the value, if positive, */
                /* or the complement of its magnitude, if negative. */
                if (nbits != 0)
                {
                    /* emit_bits rejects calls with size 0 */
                    emit_bits(temp2, nbits);
                }
            }

            /* Update restart-interval state too */
            if (m_cinfo.m_restart_interval != 0)
            {
                if (m_restarts_to_go == 0)
                {
                    m_restarts_to_go = m_cinfo.m_restart_interval;
                    m_next_restart_num++;
                    m_next_restart_num &= 7;
                }

                m_restarts_to_go--;
            }

            return true;
        }

        /// <summary>
        /// MCU encoding for AC initial scan (either spectral selection,
        /// or first pass of successive approximation).
        /// </summary>
        private bool encode_mcu_AC_first(JpegBlock[][] MCU_data)
        {
            /* Emit restart marker if needed */
            if (m_cinfo.m_restart_interval != 0)
            {
                if (m_restarts_to_go == 0)
                    emit_restart(m_next_restart_num);
            }

            /* Encode the AC coefficients per section G.1.2.2, fig. G.3 */
            /* r = run length of zeros */
            int r = 0;
            for (int k = m_cinfo.m_Ss; k <= m_cinfo.m_Se; k++)
            {
                int temp = MCU_data[0][0][JpegUtils.jpeg_natural_order[k]];
                if (temp == 0)
                {
                    r++;
                    continue;
                }

                /* We must apply the point transform by Al.  For AC coefficients this
                 * is an integer division with rounding towards 0.  To do this portably
                 * in C, we shift after obtaining the absolute value; so the code is
                 * interwoven with finding the abs value (temp) and output bits (temp2).
                 */
                int temp2;
                if (temp < 0)
                {
                    temp = -temp;       /* temp is abs value of input */
                    temp >>= m_cinfo.m_Al;        /* apply the point transform */
                    /* For a negative coef, want temp2 = bitwise complement of abs(coef) */
                    temp2 = ~temp;
                }
                else
                {
                    temp >>= m_cinfo.m_Al;        /* apply the point transform */
                    temp2 = temp;
                }

                /* Watch out for case that nonzero coef is zero after point transform */
                if (temp == 0)
                {
                    r++;
                    continue;
                }

                /* Emit any pending EOBRUN */
                if (m_EOBRUN > 0)
                    emit_eobrun();

                /* if run length > 15, must emit special run-length-16 codes (0xF0) */
                while (r > 15)
                {
                    emit_symbol(m_ac_tbl_no, 0xF0);
                    r -= 16;
                }

                /* Find the number of bits needed for the magnitude of the coefficient */
                int nbits = 1;          /* there must be at least one 1 bit */
                while ((temp >>= 1) != 0)
                    nbits++;

                /* Check for out-of-range coefficient values */
                if (nbits > MAX_HUFFMAN_COEF_BITS)
                    throw new Exception("DCT coefficient out of range");

                /* Count/emit Huffman symbol for run length / number of bits */
                emit_symbol(m_ac_tbl_no, (r << 4) + nbits);

                /* Emit that number of bits of the value, if positive, */
                /* or the complement of its magnitude, if negative. */
                emit_bits(temp2, nbits);

                r = 0;          /* reset zero run length */
            }

            if (r > 0)
            {
                /* If there are trailing zeroes, */
                m_EOBRUN++;      /* count an EOB */
                if (m_EOBRUN == 0x7FFF)
                    emit_eobrun();   /* force it out to avoid overflow */
            }

            /* Update restart-interval state too */
            if (m_cinfo.m_restart_interval != 0)
            {
                if (m_restarts_to_go == 0)
                {
                    m_restarts_to_go = m_cinfo.m_restart_interval;
                    m_next_restart_num++;
                    m_next_restart_num &= 7;
                }
                m_restarts_to_go--;
            }

            return true;
        }

        /// <summary>
        /// MCU encoding for DC successive approximation refinement scan.
        /// Note: we assume such scans can be multi-component, although the spec
        /// is not very clear on the point.
        /// </summary>
        private bool encode_mcu_DC_refine(JpegBlock[][] MCU_data)
        {
            /* Emit restart marker if needed */
            if (m_cinfo.m_restart_interval != 0)
            {
                if (m_restarts_to_go == 0)
                    emit_restart(m_next_restart_num);
            }

            /* Encode the MCU data blocks */
            for (int blkn = 0; blkn < m_cinfo.m_blocks_in_MCU; blkn++)
            {
                /* We simply emit the Al'th bit of the DC coefficient value. */
                int temp = MCU_data[blkn][0][0];
                emit_bits(temp >> m_cinfo.m_Al, 1);
            }

            /* Update restart-interval state too */
            if (m_cinfo.m_restart_interval != 0)
            {
                if (m_restarts_to_go == 0)
                {
                    m_restarts_to_go = m_cinfo.m_restart_interval;
                    m_next_restart_num++;
                    m_next_restart_num &= 7;
                }
                m_restarts_to_go--;
            }

            return true;
        }

        /// <summary>
        /// MCU encoding for AC successive approximation refinement scan.
        /// </summary>
        private bool encode_mcu_AC_refine(JpegBlock[][] MCU_data)
        {
            /* Emit restart marker if needed */
            if (m_cinfo.m_restart_interval != 0)
            {
                if (m_restarts_to_go == 0)
                    emit_restart(m_next_restart_num);
            }

            /* Encode the MCU data block */

            /* It is convenient to make a pre-pass to determine the transformed
             * coefficients' absolute values and the EOB position.
             */
            int EOB = 0;
            int[] absvalues = new int[JpegConstants.DCTSize2];
            for (int k = m_cinfo.m_Ss; k <= m_cinfo.m_Se; k++)
            {
                int temp = MCU_data[0][0][JpegUtils.jpeg_natural_order[k]];

                /* We must apply the point transform by Al.  For AC coefficients this
                 * is an integer division with rounding towards 0.  To do this portably
                 * in C, we shift after obtaining the absolute value.
                 */
                if (temp < 0)
                    temp = -temp;       /* temp is abs value of input */

                temp >>= m_cinfo.m_Al;        /* apply the point transform */
                absvalues[k] = temp;    /* save abs value for main pass */

                if (temp == 1)
                {
                    /* EOB = index of last newly-nonzero coef */
                    EOB = k;
                }
            }

            /* Encode the AC coefficients per section G.1.2.3, fig. G.7 */

            int r = 0;          /* r = run length of zeros */
            int BR = 0;         /* BR = count of buffered bits added now */
            int bitBufferOffset = m_BE; /* Append bits to buffer */

            for (int k = m_cinfo.m_Ss; k <= m_cinfo.m_Se; k++)
            {
                int temp = absvalues[k];
                if (temp == 0)
                {
                    r++;
                    continue;
                }

                /* Emit any required ZRLs, but not if they can be folded into EOB */
                while (r > 15 && k <= EOB)
                {
                    /* emit any pending EOBRUN and the BE correction bits */
                    emit_eobrun();

                    /* Emit ZRL */
                    emit_symbol(m_ac_tbl_no, 0xF0);
                    r -= 16;

                    /* Emit buffered correction bits that must be associated with ZRL */
                    emit_buffered_bits(bitBufferOffset, BR);
                    bitBufferOffset = 0;/* BE bits are gone now */
                    BR = 0;
                }

                /* If the coef was previously nonzero, it only needs a correction bit.
                 * NOTE: a straight translation of the spec's figure G.7 would suggest
                 * that we also need to test r > 15.  But if r > 15, we can only get here
                 * if k > EOB, which implies that this coefficient is not 1.
                 */
                if (temp > 1)
                {
                    /* The correction bit is the next bit of the absolute value. */
                    m_bit_buffer[bitBufferOffset + BR] = (char)(temp & 1);
                    BR++;
                    continue;
                }

                /* Emit any pending EOBRUN and the BE correction bits */
                emit_eobrun();

                /* Count/emit Huffman symbol for run length / number of bits */
                emit_symbol(m_ac_tbl_no, (r << 4) + 1);

                /* Emit output bit for newly-nonzero coef */
                temp = (MCU_data[0][0][JpegUtils.jpeg_natural_order[k]] < 0) ? 0 : 1;
                emit_bits(temp, 1);

                /* Emit buffered correction bits that must be associated with this code */
                emit_buffered_bits(bitBufferOffset, BR);
                bitBufferOffset = 0;/* BE bits are gone now */
                BR = 0;
                r = 0;          /* reset zero run length */
            }

            if (r > 0 || BR > 0)
            {
                /* If there are trailing zeroes, */
                m_EOBRUN++;      /* count an EOB */
                m_BE += BR;      /* concat my correction bits to older ones */

                /* We force out the EOB if we risk either:
                 * 1. overflow of the EOB counter;
                 * 2. overflow of the correction bit buffer during the next MCU.
                 */
                if (m_EOBRUN == 0x7FFF || m_BE > (MAX_CORR_BITS - JpegConstants.DCTSize2 + 1))
                    emit_eobrun();
            }

            /* Update restart-interval state too */
            if (m_cinfo.m_restart_interval != 0)
            {
                if (m_restarts_to_go == 0)
                {
                    m_restarts_to_go = m_cinfo.m_restart_interval;
                    m_next_restart_num++;
                    m_next_restart_num &= 7;
                }
                m_restarts_to_go--;
            }

            return true;
        }

        /// <summary>
        /// Finish up at the end of a Huffman-compressed progressive scan.
        /// </summary>
        private void finish_pass_phuff()
        {
            /* Flush out any buffered data */
            emit_eobrun();
            flush_bits();
        }

        /// <summary>
        /// Finish up a statistics-gathering pass and create the new Huffman tables.
        /// </summary>
        private void finish_pass_gather_phuff()
        {
            /* Flush out buffered data (all we care about is counting the EOB symbol) */
            emit_eobrun();

            /* It's important not to apply jpeg_gen_optimal_table more than once
             * per table, because it clobbers the input frequency counts!
             */
            bool[] did = new bool[JpegConstants.NumberOfHuffmanTables];

            bool is_DC_band = (m_cinfo.m_Ss == 0);
            for (int ci = 0; ci < m_cinfo.m_comps_in_scan; ci++)
            {
                JpegComponent componentInfo = m_cinfo.Component_info[m_cinfo.m_cur_comp_info[ci]];
                int tbl = componentInfo.Ac_tbl_no;

                if (is_DC_band)
                {
                    if (m_cinfo.m_Ah != 0) /* DC refinement needs no table */
                        continue;

                    tbl = componentInfo.Dc_tbl_no;
                }

                if (!did[tbl])
                {
                    JpegHuffmanTable htblptr = null;
                    if (is_DC_band)
                    {
                        if (m_cinfo.m_dc_huff_tbl_ptrs[tbl] == null)
                            m_cinfo.m_dc_huff_tbl_ptrs[tbl] = new JpegHuffmanTable();

                        htblptr = m_cinfo.m_dc_huff_tbl_ptrs[tbl];
                    }
                    else
                    {
                        if (m_cinfo.m_ac_huff_tbl_ptrs[tbl] == null)
                            m_cinfo.m_ac_huff_tbl_ptrs[tbl] = new JpegHuffmanTable();

                        htblptr = m_cinfo.m_ac_huff_tbl_ptrs[tbl];
                    }

                    jpeg_gen_optimal_table(htblptr, m_count_ptrs[tbl]);
                    did[tbl] = true;
                }
            }
        }

        //////////////////////////////////////////////////////////////////////////
        // Outputting bytes to the file.
        // NB: these must be called only when actually outputting,
        // that is, entropy.gather_statistics == false.

        // Emit a byte
        private void emit_byte(int val)
        {
            m_cinfo.m_dest.emit_byte(val);
        }

        /// <summary>
        /// Outputting bits to the file
        /// 
        /// Only the right 24 bits of put_buffer are used; the valid bits are
        /// left-justified in this part.  At most 16 bits can be passed to emit_bits
        /// in one call, and we never retain more than 7 bits in put_buffer
        /// between calls, so 24 bits are sufficient.
        /// </summary>
        private void emit_bits(int code, int size)
        {
            // Emit some bits, unless we are in gather mode
            /* This routine is heavily used, so it's worth coding tightly. */
            int local_put_buffer = code;

            /* if size is 0, caller used an invalid Huffman table entry */
            if (size == 0)
                throw new Exception("Missing Huffman code table entry");

            if (m_gather_statistics)
            {
                /* do nothing if we're only getting stats */
                return;
            }

            local_put_buffer &= (1 << size) - 1; /* mask off any extra bits in code */

            m_put_bits += size;       /* new number of bits in buffer */

            local_put_buffer <<= 24 - m_put_bits; /* align incoming bits */

            local_put_buffer |= m_put_buffer; /* and merge with old buffer contents */

            while (m_put_bits >= 8)
            {
                int c = (local_put_buffer >> 16) & 0xFF;

                emit_byte(c);
                if (c == 0xFF)
                {
                    /* need to stuff a zero byte? */
                    emit_byte(0);
                }
                local_put_buffer <<= 8;
                m_put_bits -= 8;
            }

            m_put_buffer = local_put_buffer; /* update variables */
        }

        private void flush_bits()
        {
            emit_bits(0x7F, 7); /* fill any partial byte with ones */
            m_put_buffer = 0;     /* and reset bit-buffer to empty */
            m_put_bits = 0;
        }

        // Emit (or just count) a Huffman symbol.
        private void emit_symbol(int tbl_no, int symbol)
        {
            if (m_gather_statistics)
                m_count_ptrs[tbl_no][symbol]++;
            else
                emit_bits(m_derived_tbls[tbl_no].ehufco[symbol], m_derived_tbls[tbl_no].ehufsi[symbol]);
        }

        // Emit bits from a correction bit buffer.
        private void emit_buffered_bits(int offset, int nbits)
        {
            if (m_gather_statistics)
            {
                /* no real work */
                return;
            }

            for (int i = 0; i < nbits; i++)
                emit_bits(m_bit_buffer[offset + i], 1);
        }

        // Emit any pending EOBRUN symbol.
        private void emit_eobrun()
        {
            if (m_EOBRUN > 0)
            {
                /* if there is any pending EOBRUN */
                int temp = m_EOBRUN;
                int nbits = 0;
                while ((temp >>= 1) != 0)
                    nbits++;

                /* safety check: shouldn't happen given limited correction-bit buffer */
                if (nbits > 14)
                    throw new Exception("Missing Huffman code table entry");

                emit_symbol(m_ac_tbl_no, nbits << 4);
                if (nbits != 0)
                    emit_bits(m_EOBRUN, nbits);

                m_EOBRUN = 0;

                /* Emit any buffered correction bits */
                emit_buffered_bits(0, m_BE);
                m_BE = 0;
            }
        }

        // Emit a restart marker & resynchronize predictions.
        private void emit_restart(int restart_num)
        {
            emit_eobrun();

            if (!m_gather_statistics)
            {
                flush_bits();
                emit_byte(0xFF);
                emit_byte((int)(JpegMarkerType.RST0 + restart_num));
            }

            if (m_cinfo.m_Ss == 0)
            {
                /* Re-initialize DC predictions to 0 */
                for (int ci = 0; ci < m_cinfo.m_comps_in_scan; ci++)
                    m_last_dc_val[ci] = 0;
            }
            else
            {
                /* Re-initialize all AC-related fields to 0 */
                m_EOBRUN = 0;
                m_BE = 0;
            }
        }

        /// <summary>
        /// IRIGHT_SHIFT is like RIGHT_SHIFT, but works on int rather than int.
        /// We assume that int right shift is unsigned if int right shift is,
        /// which should be safe.
        /// </summary>
        private static int IRIGHT_SHIFT(int x, int shft)
        {
            return (x >> shft);
        }
    }
    #endregion

    #region RawImage
    class RawImage : IRawImage
    {
        private List<SampleRow> m_samples;
        private ColorSpace m_colorspace;

        private int m_currentRow = -1;

        internal RawImage(List<SampleRow> samples, ColorSpace colorspace)
        {
            if (samples == null)
                throw new ArgumentNullException("'samples' Cannot be null!");
            if (samples.Count <= 0)
                throw new Exception("No element specified in 'samples'");
            if (colorspace == ColorSpace.Unknown)
                throw new Exception("Unknown Color Space!");

            m_samples = samples;
            m_colorspace = colorspace;
        }

        public override int Width
        {
            get
            {
                return m_samples[0].Length;
            }
        }

        public override int Height
        {
            get
            {
                return m_samples.Count;
            }
        }

        public override ColorSpace Colorspace
        {
            get
            {
                return m_colorspace;
            }
        }

        public override int ComponentsPerPixel
        {
            get
            {
                return m_samples[0][0].ComponentCount;
            }
        }

        public override void BeginRead()
        {
            m_currentRow = 0;
        }

        public override byte[] GetPixelRow()
        {
            SampleRow row = m_samples[m_currentRow];
            List<byte> result = new List<byte>();
            for (int i = 0; i < row.Length; ++i)
            {
                Sample sample = row[i];
                for (int j = 0; j < sample.ComponentCount; ++j)
                    result.Add((byte)sample[j]);
            }
            ++m_currentRow;
            return result.ToArray();
        }

        public override void EndRead()
        {
        }
    }
    #endregion

    #region Sample
    /// <summary>
    /// Represents a "sample" (you can understand it as a "pixel") of image.
    /// </summary>
    /// <remarks>It's impossible to create an instance of this class directly, 
    /// but you can use existing samples through <see cref="SampleRow"/> collection. 
    /// Usual scenario is to get row of samples from the <see cref="JpegImage.GetRow"/> method.
    /// </remarks>
    public class Sample
    {
        private short[] m_components;
        private byte m_bitsPerComponent;

        internal Sample(BitStream bitStream, byte bitsPerComponent, byte componentCount)
        {
            if (bitStream == null)
                throw new ArgumentNullException("bitStream");

            if (bitsPerComponent <= 0 || bitsPerComponent > 16)
                throw new ArgumentOutOfRangeException("bitsPerComponent");

            if (componentCount <= 0 || componentCount > 5)
                throw new ArgumentOutOfRangeException("componentCount");

            m_bitsPerComponent = bitsPerComponent;

            m_components = new short[componentCount];
            for (short i = 0; i < componentCount; ++i)
                m_components[i] = (short)bitStream.Read(bitsPerComponent);
        }

        internal Sample(short[] components, byte bitsPerComponent)
        {
            if (components == null)
                throw new ArgumentNullException("components");

            if (components.Length == 0 || components.Length > 5)
                throw new ArgumentException("components must be not empty and contain less than 5 elements");

            if (bitsPerComponent <= 0 || bitsPerComponent > 16)
                throw new ArgumentOutOfRangeException("bitsPerComponent");

            m_bitsPerComponent = bitsPerComponent;

            m_components = new short[components.Length];
            Buffer.BlockCopy(components, 0, m_components, 0, components.Length * sizeof(short));
        }

        /// <summary>
        /// Gets the number of bits per color component.
        /// </summary>
        /// <value>The number of bits per color component.</value>
        public byte BitsPerComponent
        {
            get
            {
                return m_bitsPerComponent;
            }
        }

        /// <summary>
        /// Gets the number of color components.
        /// </summary>
        /// <value>The number of color components.</value>
        public byte ComponentCount
        {
            get
            {
                return (byte)m_components.Length;
            }
        }

        /// <summary>
        /// Gets the color component at the specified index.
        /// </summary>
        /// <param name="componentNumber">The number of color component.</param>
        /// <returns>Value of color component.</returns>
        public short this[int componentNumber]
        {
            get
            {
                return GetComponent(componentNumber);
            }
        }

        /// <summary>
        /// Gets the required color component.
        /// </summary>
        /// <param name="componentNumber">The number of color component.</param>
        /// <returns>Value of color component.</returns>
        public short GetComponent(int componentNumber)
        {
            return m_components[componentNumber];
        }
    }
    #endregion

    #region SampleRow
    /// <summary>
    /// Represents a row of image - collection of samples.
    /// </summary>
    public class SampleRow
    {
        private byte[] m_bytes;
        private Sample[] m_samples;

        /// <summary>
        /// Creates a row from raw samples data.
        /// </summary>
        /// <param name="row">Raw description of samples.<br/>
        /// You can pass collection with more than sampleCount samples - only sampleCount samples 
        /// will be parsed and all remaining bytes will be ignored.</param>
        /// <param name="sampleCount">The number of samples in row.</param>
        /// <param name="bitsPerComponent">The number of bits per component.</param>
        /// <param name="componentsPerSample">The number of components per sample.</param>
        public SampleRow(byte[] row, int sampleCount, byte bitsPerComponent, byte componentsPerSample)
        {
            if (row == null)
                throw new ArgumentNullException("row");

            if (row.Length == 0)
                throw new ArgumentException("row is empty");

            if (sampleCount <= 0)
                throw new ArgumentOutOfRangeException("sampleCount");

            if (bitsPerComponent <= 0 || bitsPerComponent > 16)
                throw new ArgumentOutOfRangeException("bitsPerComponent");

            if (componentsPerSample <= 0 || componentsPerSample > 5)
                throw new ArgumentOutOfRangeException("componentsPerSample");

            m_bytes = row;

            using (BitStream bitStream = new BitStream(row))
            {
                m_samples = new Sample[sampleCount];
                for (int i = 0; i < sampleCount; ++i)
                    m_samples[i] = new Sample(bitStream, bitsPerComponent, componentsPerSample);
            }
        }

        /// <summary>
        /// Creates row from an array of components.
        /// </summary>
        /// <param name="sampleComponents">Array of color components.</param>
        /// <param name="bitsPerComponent">The number of bits per component.</param>
        /// <param name="componentsPerSample">The number of components per sample.</param>
        /// <remarks>The difference between this constructor and 
        /// <see cref="M:BitMiracle.LibJpeg.SampleRow.#ctor(System.Byte[],System.Int32,System.Byte,System.Byte)">another one</see> -
        /// this constructor accept an array of prepared color components whereas
        /// another constructor accept raw bytes and parse them.
        /// </remarks>
        internal SampleRow(short[] sampleComponents, byte bitsPerComponent, byte componentsPerSample)
        {
            if (sampleComponents == null)
                throw new ArgumentNullException("sampleComponents");

            if (sampleComponents.Length == 0)
                throw new ArgumentException("row is empty");

            if (bitsPerComponent <= 0 || bitsPerComponent > 16)
                throw new ArgumentOutOfRangeException("bitsPerComponent");

            if (componentsPerSample <= 0 || componentsPerSample > 5)
                throw new ArgumentOutOfRangeException("componentsPerSample");

            int sampleCount = sampleComponents.Length / componentsPerSample;
            m_samples = new Sample[sampleCount];
            for (int i = 0; i < sampleCount; ++i)
            {
                short[] components = new short[componentsPerSample];
                Buffer.BlockCopy(sampleComponents, i * componentsPerSample * sizeof(short), components, 0, componentsPerSample * sizeof(short));
                m_samples[i] = new Sample(components, bitsPerComponent);
            }

            using (BitStream bits = new BitStream())
            {
                for (int i = 0; i < sampleCount; ++i)
                {
                    for (int j = 0; j < componentsPerSample; ++j)
                        bits.Write(sampleComponents[i * componentsPerSample + j], bitsPerComponent);
                }

                m_bytes = new byte[bits.UnderlyingStream.Length];
                bits.UnderlyingStream.Seek(0, System.IO.SeekOrigin.Begin);
                bits.UnderlyingStream.Read(m_bytes, 0, (int)bits.UnderlyingStream.Length);
            }
        }


        /// <summary>
        /// Gets the number of samples in this row.
        /// </summary>
        /// <value>The number of samples.</value>
        public int Length
        {
            get
            {
                return m_samples.Length;
            }
        }


        /// <summary>
        /// Gets the sample at the specified index.
        /// </summary>
        /// <param name="sampleNumber">The number of sample.</param>
        /// <returns>The required sample.</returns>
        public Sample this[int sampleNumber]
        {
            get
            {
                return GetAt(sampleNumber);
            }
        }

        /// <summary>
        /// Gets the sample at the specified index.
        /// </summary>
        /// <param name="sampleNumber">The number of sample.</param>
        /// <returns>The required sample.</returns>
        public Sample GetAt(int sampleNumber)
        {
            return m_samples[sampleNumber];
        }

        /// <summary>
        /// Serializes this row to raw bytes.
        /// </summary>
        /// <returns>The row representation as array of bytes</returns>
        public byte[] ToBytes()
        {
            return m_bytes;
        }
    }
    #endregion

    #region SavedBitreadState
    /// <summary>
    /// Bitreading state saved across MCUs
    /// </summary>
    struct SavedBitreadState
    {
        public int get_buffer;    /* current bit-extraction buffer */
        public int bits_left;      /* # of unused bits in it */
    }
    #endregion

    #region SourceManagerImpl
    /// <summary>
    /// Expanded data source object for stdio input
    /// </summary>
    class SourceManagerImpl : Jpeg_Source
    {
        private const int INPUT_BUF_SIZE = 4096;

        private JpegDecompressor m_cinfo;

        private Stream m_infile;       /* source stream */
        private byte[] m_buffer;     /* start of buffer */
        private bool m_start_of_file; /* have we gotten any data yet? */

        /// <summary>
        /// Initialize source - called by jpeg_read_header
        /// before any data is actually read.
        /// </summary>
        public SourceManagerImpl(JpegDecompressor cinfo)
        {
            m_cinfo = cinfo;
            m_buffer = new byte[INPUT_BUF_SIZE];
        }

        public void Attach(Stream infile)
        {
            m_infile = infile;
            m_infile.Seek(0, SeekOrigin.Begin);
            initInternalBuffer(null, 0);
        }

        public override void init_source()
        {
            /* We reset the empty-input-file flag for each image,
             * but we don't clear the input buffer.
             * This is correct behavior for reading a series of images from one source.
             */
            m_start_of_file = true;
        }

        /// <summary>
        /// Fill the input buffer - called whenever buffer is emptied.
        /// 
        /// In typical applications, this should read fresh data into the buffer
        /// (ignoring the current state of next_input_byte and bytes_in_buffer),
        /// reset the pointer and count to the start of the buffer, and return true
        /// indicating that the buffer has been reloaded.  It is not necessary to
        /// fill the buffer entirely, only to obtain at least one more byte.
        /// 
        /// There is no such thing as an EOF return.  If the end of the file has been
        /// reached, the routine has a choice of ERREXIT() or inserting fake data into
        /// the buffer.  In most cases, generating a warning message and inserting a
        /// fake EOI marker is the best course of action --- this will allow the
        /// decompressor to output however much of the image is there.  However,
        /// the resulting error message is misleading if the real problem is an empty
        /// input file, so we handle that case specially.
        /// 
        /// In applications that need to be able to suspend compression due to input
        /// not being available yet, a false return indicates that no more data can be
        /// obtained right now, but more may be forthcoming later.  In this situation,
        /// the decompressor will return to its caller (with an indication of the
        /// number of scanlines it has read, if any).  The application should resume
        /// decompression after it has loaded more data into the input buffer.  Note
        /// that there are substantial restrictions on the use of suspension --- see
        /// the documentation.
        /// 
        /// When suspending, the decompressor will back up to a convenient restart point
        /// (typically the start of the current MCU). next_input_byte and bytes_in_buffer
        /// indicate where the restart point will be if the current call returns false.
        /// Data beyond this point must be rescanned after resumption, so move it to
        /// the front of the buffer rather than discarding it.
        /// </summary>
        public override bool fill_input_buffer()
        {
            int nbytes = m_infile.Read(m_buffer, 0, INPUT_BUF_SIZE);
            if (nbytes <= 0)
            {
                if (m_start_of_file) /* Treat empty input file as fatal error */
                    throw new Exception("The input file is empty!");

                /* Insert a fake EOI marker */
                m_buffer[0] = (byte)0xFF;
                m_buffer[1] = (byte)JpegMarkerType.EOI;
                nbytes = 2;
            }

            initInternalBuffer(m_buffer, nbytes);
            m_start_of_file = false;

            return true;
        }
    }
    #endregion

    #region TransCoefControllerImpl
    /// <summary>
    /// This is a special implementation of the coefficient
    /// buffer controller.  This is similar to jccoefct.c, but it handles only
    /// output from presupplied virtual arrays.  Furthermore, we generate any
    /// dummy padding blocks on-the-fly rather than expecting them to be present
    /// in the arrays.
    /// </summary>
    class TransCoefControllerImpl : JpegCompressorCoefController
    {
        private JpegCompressor m_cinfo;

        private int m_iMCU_row_num;    /* iMCU row # within image */
        private int m_mcu_ctr;     /* counts MCUs processed in current row */
        private int m_MCU_vert_offset;        /* counts MCU rows within iMCU row */
        private int m_MCU_rows_per_iMCU_row;  /* number of such rows needed */

        /* Virtual block array for each component. */
        private JpegVirtualArray<JpegBlock>[] m_whole_image;

        /* Workspace for constructing dummy blocks at right/bottom edges. */
        private JpegBlock[][] m_dummy_buffer = new JpegBlock[JpegConstants.CompressorMaxBlocksInMCU][];

        /// <summary>
        /// Initialize coefficient buffer controller.
        /// 
        /// Each passed coefficient array must be the right size for that
        /// coefficient: width_in_blocks wide and height_in_blocks high,
        /// with unit height at least v_samp_factor.
        /// </summary>
        public TransCoefControllerImpl(JpegCompressor cinfo, JpegVirtualArray<JpegBlock>[] coef_arrays)
        {
            m_cinfo = cinfo;

            /* Save pointer to virtual arrays */
            m_whole_image = coef_arrays;

            /* Allocate and pre-zero space for dummy DCT blocks. */
            JpegBlock[] buffer = new JpegBlock[JpegConstants.CompressorMaxBlocksInMCU];
            for (int i = 0; i < JpegConstants.CompressorMaxBlocksInMCU; i++)
                buffer[i] = new JpegBlock();

            for (int i = 0; i < JpegConstants.CompressorMaxBlocksInMCU; i++)
            {
                m_dummy_buffer[i] = new JpegBlock[JpegConstants.CompressorMaxBlocksInMCU - i];
                for (int j = i; j < JpegConstants.CompressorMaxBlocksInMCU; j++)
                    m_dummy_buffer[i][j - i] = buffer[j];
            }
        }

        /// <summary>
        /// Initialize for a processing pass.
        /// </summary>
        public virtual void start_pass(BufferMode pass_mode)
        {
            if (pass_mode != BufferMode.CrankDest)
                throw new Exception("Bogus buffer control mode");

            m_iMCU_row_num = 0;
            start_iMCU_row();
        }

        /// <summary>
        /// Process some data.
        /// We process the equivalent of one fully interleaved MCU row ("iMCU" row)
        /// per call, ie, v_samp_factor block rows for each component in the scan.
        /// The data is obtained from the virtual arrays and fed to the entropy coder.
        /// Returns true if the iMCU row is completed, false if suspended.
        /// 
        /// NB: input_buf is ignored; it is likely to be a null pointer.
        /// </summary>
        public virtual bool compress_data(byte[][][] input_buf)
        {
            /* Align the virtual buffers for the components used in this scan. */
            JpegBlock[][][] buffer = new JpegBlock[JpegConstants.MaxComponentsInScan][][];
            for (int ci = 0; ci < m_cinfo.m_comps_in_scan; ci++)
            {
                JpegComponent componentInfo = m_cinfo.Component_info[m_cinfo.m_cur_comp_info[ci]];
                buffer[ci] = m_whole_image[componentInfo.Component_index].Access(
                    m_iMCU_row_num * componentInfo.V_samp_factor, componentInfo.V_samp_factor);
            }

            /* Loop to process one whole iMCU row */
            int last_MCU_col = m_cinfo.m_MCUs_per_row - 1;
            int last_iMCU_row = m_cinfo.m_total_iMCU_rows - 1;
            JpegBlock[][] MCU_buffer = new JpegBlock[JpegConstants.CompressorMaxBlocksInMCU][];
            for (int yoffset = m_MCU_vert_offset; yoffset < m_MCU_rows_per_iMCU_row; yoffset++)
            {
                for (int MCU_col_num = m_mcu_ctr; MCU_col_num < m_cinfo.m_MCUs_per_row; MCU_col_num++)
                {
                    /* Construct list of pointers to DCT blocks belonging to this MCU */
                    int blkn = 0;           /* index of current DCT block within MCU */
                    for (int ci = 0; ci < m_cinfo.m_comps_in_scan; ci++)
                    {
                        JpegComponent componentInfo = m_cinfo.Component_info[m_cinfo.m_cur_comp_info[ci]];
                        int start_col = MCU_col_num * componentInfo.MCU_width;
                        int blockcnt = (MCU_col_num < last_MCU_col) ? componentInfo.MCU_width : componentInfo.last_col_width;
                        for (int yindex = 0; yindex < componentInfo.MCU_height; yindex++)
                        {
                            int xindex = 0;
                            if (m_iMCU_row_num < last_iMCU_row || yindex + yoffset < componentInfo.last_row_height)
                            {
                                /* Fill in pointers to real blocks in this row */
                                for (xindex = 0; xindex < blockcnt; xindex++)
                                {
                                    int bufLength = buffer[ci][yindex + yoffset].Length;
                                    int start = start_col + xindex;
                                    MCU_buffer[blkn] = new JpegBlock[bufLength - start];
                                    for (int j = start; j < bufLength; j++)
                                        MCU_buffer[blkn][j - start] = buffer[ci][yindex + yoffset][j];

                                    blkn++;
                                }
                            }
                            else
                            {
                                /* At bottom of image, need a whole row of dummy blocks */
                                xindex = 0;
                            }

                            /* Fill in any dummy blocks needed in this row.
                            * Dummy blocks are filled in the same way as in jccoefct.c:
                            * all zeroes in the AC entries, DC entries equal to previous
                            * block's DC value.  The init routine has already zeroed the
                            * AC entries, so we need only set the DC entries correctly.
                            */
                            for (; xindex < componentInfo.MCU_width; xindex++)
                            {
                                MCU_buffer[blkn] = m_dummy_buffer[blkn];
                                MCU_buffer[blkn][0][0] = MCU_buffer[blkn - 1][0][0];
                                blkn++;
                            }
                        }
                    }

                    /* Try to write the MCU. */
                    if (!m_cinfo.m_entropy.encode_mcu(MCU_buffer))
                    {
                        /* Suspension forced; update state counters and exit */
                        m_MCU_vert_offset = yoffset;
                        m_mcu_ctr = MCU_col_num;
                        return false;
                    }
                }

                /* Completed an MCU row, but perhaps not an iMCU row */
                m_mcu_ctr = 0;
            }

            /* Completed the iMCU row, advance counters for next one */
            m_iMCU_row_num++;
            start_iMCU_row();
            return true;
        }

        /// <summary>
        /// Reset within-iMCU-row counters for a new row
        /// </summary>
        private void start_iMCU_row()
        {
            /* In an interleaved scan, an MCU row is the same as an iMCU row.
            * In a noninterleaved scan, an iMCU row has v_samp_factor MCU rows.
            * But at the bottom of the image, process only what's left.
            */
            if (m_cinfo.m_comps_in_scan > 1)
            {
                m_MCU_rows_per_iMCU_row = 1;
            }
            else
            {
                if (m_iMCU_row_num < (m_cinfo.m_total_iMCU_rows - 1))
                    m_MCU_rows_per_iMCU_row = m_cinfo.Component_info[m_cinfo.m_cur_comp_info[0]].V_samp_factor;
                else
                    m_MCU_rows_per_iMCU_row = m_cinfo.Component_info[m_cinfo.m_cur_comp_info[0]].last_row_height;
            }

            m_mcu_ctr = 0;
            m_MCU_vert_offset = 0;
        }
    }
    #endregion

    #region UpsamplerImpl
    class UpsamplerImpl : JpegUpsampler
    {
        private enum ComponentUpsampler
        {
            noop_upsampler,
            fullsize_upsampler,
            h2v1_fancy_upsampler,
            h2v1_upsampler,
            h2v2_fancy_upsampler,
            h2v2_upsampler,
            int_upsampler
        }

        private JpegDecompressor m_cinfo;

        /* Color conversion buffer.  When using separate upsampling and color
        * conversion steps, this buffer holds one upsampled row group until it
        * has been color converted and output.
        * Note: we do not allocate any storage for component(s) which are full-size,
        * ie do not need rescaling.  The corresponding entry of color_buf[] is
        * simply set to point to the input data array, thereby avoiding copying.
        */
        private ComponentBuffer[] m_color_buf = new ComponentBuffer[JpegConstants.MaxComponents];

        // used only for fullsize_upsampler mode
        private int[] m_perComponentOffsets = new int[JpegConstants.MaxComponents];

        /* Per-component upsampling method pointers */
        private ComponentUpsampler[] m_upsampleMethods = new ComponentUpsampler[JpegConstants.MaxComponents];
        private int m_currentComponent; // component being upsampled
        private int m_upsampleRowOffset;

        private int m_next_row_out;       /* counts rows emitted from color_buf */
        private int m_rows_to_go;  /* counts rows remaining in image */

        /* Height of an input row group for each component. */
        private int[] m_rowgroup_height = new int[JpegConstants.MaxComponents];

        /* These arrays save pixel expansion factors so that int_expand need not
        * re-compute them each time.  They are unused for other up-sampling methods.
        */
        private byte[] m_h_expand = new byte[JpegConstants.MaxComponents];
        private byte[] m_v_expand = new byte[JpegConstants.MaxComponents];

        public UpsamplerImpl(JpegDecompressor cinfo)
        {
            m_cinfo = cinfo;
            m_need_context_rows = false; /* until we find out differently */

            if (cinfo.m_CCIR601_sampling)    /* this isn't supported */
                throw new Exception("CCIR601 sampling not implemented yet");

            /* JpegDecompressorMainController doesn't support context rows when min_DCT_scaled_size = 1,
            * so don't ask for it.
            */
            bool do_fancy = cinfo.m_do_fancy_upsampling && cinfo.m_min_DCT_scaled_size > 1;

            /* Verify we can handle the sampling factors, select per-component methods,
            * and create storage as needed.
            */
            for (int ci = 0; ci < cinfo.m_num_components; ci++)
            {
                JpegComponent componentInfo = cinfo.Comp_info[ci];

                /* Compute size of an "input group" after IDCT scaling.  This many samples
                * are to be converted to max_h_samp_factor * max_v_samp_factor pixels.
                */
                int h_in_group = (componentInfo.H_samp_factor * componentInfo.DCT_scaled_size) / cinfo.m_min_DCT_scaled_size;
                int v_in_group = (componentInfo.V_samp_factor * componentInfo.DCT_scaled_size) / cinfo.m_min_DCT_scaled_size;
                int h_out_group = cinfo.m_max_h_samp_factor;
                int v_out_group = cinfo.m_max_v_samp_factor;

                /* save for use later */
                m_rowgroup_height[ci] = v_in_group;
                bool need_buffer = true;
                if (!componentInfo.component_needed)
                {
                    /* Don't bother to upsample an uninteresting component. */
                    m_upsampleMethods[ci] = ComponentUpsampler.noop_upsampler;
                    need_buffer = false;
                }
                else if (h_in_group == h_out_group && v_in_group == v_out_group)
                {
                    /* Fullsize components can be processed without any work. */
                    m_upsampleMethods[ci] = ComponentUpsampler.fullsize_upsampler;
                    need_buffer = false;
                }
                else if (h_in_group * 2 == h_out_group && v_in_group == v_out_group)
                {
                    /* Special cases for 2h1v upsampling */
                    if (do_fancy && componentInfo.downsampled_width > 2)
                        m_upsampleMethods[ci] = ComponentUpsampler.h2v1_fancy_upsampler;
                    else
                        m_upsampleMethods[ci] = ComponentUpsampler.h2v1_upsampler;
                }
                else if (h_in_group * 2 == h_out_group && v_in_group * 2 == v_out_group)
                {
                    /* Special cases for 2h2v upsampling */
                    if (do_fancy && componentInfo.downsampled_width > 2)
                    {
                        m_upsampleMethods[ci] = ComponentUpsampler.h2v2_fancy_upsampler;
                        m_need_context_rows = true;
                    }
                    else
                    {
                        m_upsampleMethods[ci] = ComponentUpsampler.h2v2_upsampler;
                    }
                }
                else if ((h_out_group % h_in_group) == 0 && (v_out_group % v_in_group) == 0)
                {
                    /* Generic integral-factors up-sampling method */
                    m_upsampleMethods[ci] = ComponentUpsampler.int_upsampler;
                    m_h_expand[ci] = (byte)(h_out_group / h_in_group);
                    m_v_expand[ci] = (byte)(v_out_group / v_in_group);
                }
                else
                    throw new Exception("Fractional sampling not implemented yet");

                if (need_buffer)
                {
                    ComponentBuffer cb = new ComponentBuffer();
                    cb.SetBuffer(JpegCommonBase.AllocJpegSamples(JpegUtils.jround_up(cinfo.m_output_width,
                        cinfo.m_max_h_samp_factor), cinfo.m_max_v_samp_factor), null, 0);

                    m_color_buf[ci] = cb;
                }
            }
        }

        /// <summary>
        /// Initialize for an upsampling pass.
        /// </summary>
        public override void start_pass()
        {
            /* Mark the conversion buffer empty */
            m_next_row_out = m_cinfo.m_max_v_samp_factor;

            /* Initialize total-height counter for detecting bottom of image */
            m_rows_to_go = m_cinfo.m_output_height;
        }

        /// <summary>
        /// Control routine to do upsampling (and color conversion).
        /// 
        /// In this version we upsample each component independently.
        /// We upsample one row group into the conversion buffer, then apply
        /// color conversion a row at a time.
        /// </summary>
        public override void upsample(ComponentBuffer[] input_buf, ref int in_row_group_ctr, int in_row_groups_avail, byte[][] output_buf, ref int out_row_ctr, int out_rows_avail)
        {
            /* Fill the conversion buffer, if it's empty */
            if (m_next_row_out >= m_cinfo.m_max_v_samp_factor)
            {
                for (int ci = 0; ci < m_cinfo.m_num_components; ci++)
                {
                    m_perComponentOffsets[ci] = 0;

                    /* Invoke per-component upsample method.*/
                    m_currentComponent = ci;
                    m_upsampleRowOffset = in_row_group_ctr * m_rowgroup_height[ci];
                    upsampleComponent(ref input_buf[ci]);
                }

                m_next_row_out = 0;
            }

            /* Color-convert and emit rows */

            /* How many we have in the buffer: */
            int num_rows = m_cinfo.m_max_v_samp_factor - m_next_row_out;

            /* Not more than the distance to the end of the image.  Need this test
             * in case the image height is not a multiple of max_v_samp_factor:
             */
            if (num_rows > m_rows_to_go)
                num_rows = m_rows_to_go;

            /* And not more than what the client can accept: */
            out_rows_avail -= out_row_ctr;
            if (num_rows > out_rows_avail)
                num_rows = out_rows_avail;

            m_cinfo.m_cconvert.color_convert(m_color_buf, m_perComponentOffsets, m_next_row_out, output_buf, out_row_ctr, num_rows);

            /* Adjust counts */
            out_row_ctr += num_rows;
            m_rows_to_go -= num_rows;
            m_next_row_out += num_rows;

            /* When the buffer is emptied, declare this input row group consumed */
            if (m_next_row_out >= m_cinfo.m_max_v_samp_factor)
                in_row_group_ctr++;
        }

        private void upsampleComponent(ref ComponentBuffer input_data)
        {
            switch (m_upsampleMethods[m_currentComponent])
            {
                case ComponentUpsampler.noop_upsampler:
                    noop_upsample();
                    break;
                case ComponentUpsampler.fullsize_upsampler:
                    fullsize_upsample(ref input_data);
                    break;
                case ComponentUpsampler.h2v1_fancy_upsampler:
                    h2v1_fancy_upsample(m_cinfo.Comp_info[m_currentComponent].downsampled_width, ref input_data);
                    break;
                case ComponentUpsampler.h2v1_upsampler:
                    h2v1_upsample(ref input_data);
                    break;
                case ComponentUpsampler.h2v2_fancy_upsampler:
                    h2v2_fancy_upsample(m_cinfo.Comp_info[m_currentComponent].downsampled_width, ref input_data);
                    break;
                case ComponentUpsampler.h2v2_upsampler:
                    h2v2_upsample(ref input_data);
                    break;
                case ComponentUpsampler.int_upsampler:
                    int_upsample(ref input_data);
                    break;
                default:
                    throw new Exception("The specified Component Up-sampler isn't implemented.");
            }
        }

        /*
         * These are the routines invoked to upsample pixel values
         * of a single component.  One row group is processed per call.
         */

        /// <summary>
        /// This is a no-op version used for "uninteresting" components.
        /// These components will not be referenced by color conversion.
        /// </summary>
        private static void noop_upsample()
        {
            // do nothing
        }

        /// <summary>
        /// For full-size components, we just make color_buf[ci] point at the
        /// input buffer, and thus avoid copying any data.  Note that this is
        /// safe only because sep_upsample doesn't declare the input row group
        /// "consumed" until we are done color converting and emitting it.
        /// </summary>
        private void fullsize_upsample(ref ComponentBuffer input_data)
        {
            m_color_buf[m_currentComponent] = input_data;
            m_perComponentOffsets[m_currentComponent] = m_upsampleRowOffset;
        }

        /// <summary>
        /// Fancy processing for the common case of 2:1 horizontal and 1:1 vertical.
        /// 
        /// The upsampling algorithm is linear interpolation between pixel centers,
        /// also known as a "triangle filter".  This is a good compromise between
        /// speed and visual quality.  The centers of the output pixels are 1/4 and 3/4
        /// of the way between input pixel centers.
        /// 
        /// A note about the "bias" calculations: when rounding fractional values to
        /// integer, we do not want to always round 0.5 up to the next integer.
        /// If we did that, we'd introduce a noticeable bias towards larger values.
        /// Instead, this code is arranged so that 0.5 will be rounded up or down at
        /// alternate pixel locations (a simple ordered dither pattern).
        /// </summary>
        private void h2v1_fancy_upsample(int downsampled_width, ref ComponentBuffer input_data)
        {
            ComponentBuffer output_data = m_color_buf[m_currentComponent];

            for (int inrow = 0; inrow < m_cinfo.m_max_v_samp_factor; inrow++)
            {
                int row = m_upsampleRowOffset + inrow;
                int inIndex = 0;

                int outIndex = 0;

                /* Special case for first column */
                int invalue = input_data[row][inIndex];
                inIndex++;

                output_data[inrow][outIndex] = (byte)invalue;
                outIndex++;
                output_data[inrow][outIndex] = (byte)((invalue * 3 + (int)input_data[row][inIndex] + 2) >> 2);
                outIndex++;

                for (int colctr = downsampled_width - 2; colctr > 0; colctr--)
                {
                    /* General case: 3/4 * nearer pixel + 1/4 * further pixel */
                    invalue = (int)input_data[row][inIndex] * 3;
                    inIndex++;

                    output_data[inrow][outIndex] = (byte)((invalue + (int)input_data[row][inIndex - 2] + 1) >> 2);
                    outIndex++;

                    output_data[inrow][outIndex] = (byte)((invalue + (int)input_data[row][inIndex] + 2) >> 2);
                    outIndex++;
                }

                /* Special case for last column */
                invalue = input_data[row][inIndex];
                output_data[inrow][outIndex] = (byte)((invalue * 3 + (int)input_data[row][inIndex - 1] + 1) >> 2);
                outIndex++;
                output_data[inrow][outIndex] = (byte)invalue;
                outIndex++;
            }
        }

        /// <summary>
        /// Fast processing for the common case of 2:1 horizontal and 1:1 vertical.
        /// It's still a box filter.
        /// </summary>
        private void h2v1_upsample(ref ComponentBuffer input_data)
        {
            ComponentBuffer output_data = m_color_buf[m_currentComponent];

            for (int inrow = 0; inrow < m_cinfo.m_max_v_samp_factor; inrow++)
            {
                int row = m_upsampleRowOffset + inrow;
                int outIndex = 0;

                for (int col = 0; col < m_cinfo.m_output_width; col++)
                {
                    byte invalue = input_data[row][col]; /* don't need GETJSAMPLE() here */
                    output_data[inrow][outIndex] = invalue;
                    outIndex++;
                    output_data[inrow][outIndex] = invalue;
                    outIndex++;
                }
            }
        }

        /// <summary>
        /// Fancy processing for the common case of 2:1 horizontal and 2:1 vertical.
        /// Again a triangle filter; see comments for h2v1 case, above.
        /// 
        /// It is OK for us to reference the adjacent input rows because we demanded
        /// context from the main buffer controller (see initialization code).
        /// </summary>
        private void h2v2_fancy_upsample(int downsampled_width, ref ComponentBuffer input_data)
        {
            ComponentBuffer output_data = m_color_buf[m_currentComponent];

            int inrow = m_upsampleRowOffset;
            int outrow = 0;
            while (outrow < m_cinfo.m_max_v_samp_factor)
            {
                for (int v = 0; v < 2; v++)
                {
                    // nearest input row index
                    int inIndex0 = 0;

                    //next nearest input row index
                    int inIndex1 = 0;
                    int inRow1 = -1;
                    if (v == 0)
                    {
                        /* next nearest is row above */
                        inRow1 = inrow - 1;
                    }
                    else
                    {
                        /* next nearest is row below */
                        inRow1 = inrow + 1;
                    }

                    int row = outrow;
                    int outIndex = 0;
                    outrow++;

                    /* Special case for first column */
                    int thiscolsum = (int)input_data[inrow][inIndex0] * 3 + (int)input_data[inRow1][inIndex1];
                    inIndex0++;
                    inIndex1++;

                    int nextcolsum = (int)input_data[inrow][inIndex0] * 3 + (int)input_data[inRow1][inIndex1];
                    inIndex0++;
                    inIndex1++;

                    output_data[row][outIndex] = (byte)((thiscolsum * 4 + 8) >> 4);
                    outIndex++;

                    output_data[row][outIndex] = (byte)((thiscolsum * 3 + nextcolsum + 7) >> 4);
                    outIndex++;

                    int lastcolsum = thiscolsum;
                    thiscolsum = nextcolsum;

                    for (int colctr = downsampled_width - 2; colctr > 0; colctr--)
                    {
                        /* General case: 3/4 * nearer pixel + 1/4 * further pixel in each */
                        /* dimension, thus 9/16, 3/16, 3/16, 1/16 overall */
                        nextcolsum = (int)input_data[inrow][inIndex0] * 3 + (int)input_data[inRow1][inIndex1];
                        inIndex0++;
                        inIndex1++;

                        output_data[row][outIndex] = (byte)((thiscolsum * 3 + lastcolsum + 8) >> 4);
                        outIndex++;

                        output_data[row][outIndex] = (byte)((thiscolsum * 3 + nextcolsum + 7) >> 4);
                        outIndex++;

                        lastcolsum = thiscolsum;
                        thiscolsum = nextcolsum;
                    }

                    /* Special case for last column */
                    output_data[row][outIndex] = (byte)((thiscolsum * 3 + lastcolsum + 8) >> 4);
                    outIndex++;
                    output_data[row][outIndex] = (byte)((thiscolsum * 4 + 7) >> 4);
                    outIndex++;
                }

                inrow++;
            }
        }

        /// <summary>
        /// Fast processing for the common case of 2:1 horizontal and 2:1 vertical.
        /// It's still a box filter.
        /// </summary>
        private void h2v2_upsample(ref ComponentBuffer input_data)
        {
            ComponentBuffer output_data = m_color_buf[m_currentComponent];

            int inrow = 0;
            int outrow = 0;
            while (outrow < m_cinfo.m_max_v_samp_factor)
            {
                int row = m_upsampleRowOffset + inrow;
                int outIndex = 0;

                for (int col = 0; col < m_cinfo.m_output_width; col++)
                {
                    byte invalue = input_data[row][col]; /* don't need GETJSAMPLE() here */
                    output_data[outrow][outIndex] = invalue;
                    outIndex++;
                    output_data[outrow][outIndex] = invalue;
                    outIndex++;
                }

                JpegUtils.jcopy_sample_rows(output_data, outrow, output_data, outrow + 1, 1, m_cinfo.m_output_width);
                inrow++;
                outrow += 2;
            }
        }

        /// <summary>
        /// This version handles any integral sampling ratios.
        /// This is not used for typical JPEG files, so it need not be fast.
        /// Nor, for that matter, is it particularly accurate: the algorithm is
        /// simple replication of the input pixel onto the corresponding output
        /// pixels.  The hi-falutin sampling literature refers to this as a
        /// "box filter".  A box filter tends to introduce visible artifacts,
        /// so if you are actually going to use 3:1 or 4:1 sampling ratios
        /// you would be well advised to improve this code.
        /// </summary>
        private void int_upsample(ref ComponentBuffer input_data)
        {
            ComponentBuffer output_data = m_color_buf[m_currentComponent];
            int h_expand = m_h_expand[m_currentComponent];
            int v_expand = m_v_expand[m_currentComponent];

            int inrow = 0;
            int outrow = 0;
            while (outrow < m_cinfo.m_max_v_samp_factor)
            {
                /* Generate one output row with proper horizontal expansion */
                int row = m_upsampleRowOffset + inrow;
                for (int col = 0; col < m_cinfo.m_output_width; col++)
                {
                    byte invalue = input_data[row][col]; /* don't need GETJSAMPLE() here */
                    int outIndex = 0;
                    for (int h = h_expand; h > 0; h--)
                    {
                        output_data[outrow][outIndex] = invalue;
                        outIndex++;
                    }
                }

                /* Generate any additional output rows by duplicating the first one */
                if (v_expand > 1)
                {
                    JpegUtils.jcopy_sample_rows(output_data, outrow, output_data,
                        outrow + 1, v_expand - 1, m_cinfo.m_output_width);
                }

                inrow++;
                outrow += v_expand;
            }
        }
    }
    #endregion

    #region Utils
    class Utils
    {
        public static MemoryStream CopyStream(Stream stream)
        {
            if (stream == null)
                throw new ArgumentNullException("stream");

            long positionBefore = stream.Position;
            stream.Seek(0, SeekOrigin.Begin);

            MemoryStream result = new MemoryStream((int)stream.Length);

            byte[] block = new byte[2048];
            for (; ; )
            {
                int bytesRead = stream.Read(block, 0, 2048);
                result.Write(block, 0, bytesRead);
                if (bytesRead < 2048)
                    break;
            }

            stream.Seek(positionBefore, SeekOrigin.Begin);
            return result;
        }

        public static void CMYK2RGB(byte c, byte m, byte y, byte k, out byte red, out byte green, out byte blue)
        {
            float C, M, Y, K;
            C = c / 255.0f;
            M = m / 255.0f;
            Y = y / 255.0f;
            K = k / 255.0f;

            float R, G, B;
            R = C * (1.0f - K) + K;
            G = M * (1.0f - K) + K;
            B = Y * (1.0f - K) + K;

            R = (1.0f - R) * 255.0f + 0.5f;
            G = (1.0f - G) * 255.0f + 0.5f;
            B = (1.0f - B) * 255.0f + 0.5f;

            red = (byte)(R * 255);
            green = (byte)(G * 255);
            blue = (byte)(B * 255);

            //C = (double)c;
            //M = (double)m;
            //Y = (double)y;
            //K = (double)k;

            //C = C / 255.0;
            //M = M / 255.0;
            //Y = Y / 255.0;
            //K = K / 255.0;

            //R = C * (1.0 - K) + K;
            //G = M * (1.0 - K) + K;
            //B = Y * (1.0 - K) + K;

            //R = (1.0 - R) * 255.0 + 0.5;
            //G = (1.0 - G) * 255.0 + 0.5;
            //B = (1.0 - B) * 255.0 + 0.5;

            //r = (byte)R;
            //g = (byte)G;
            //b = (byte)B;

            //rgb = RGB(r, g, b);

            //return rgb;
        }
    }
    #endregion
    
    #region WorkingBitreadState
    /// <summary>
    /// Bitreading working state within an MCU
    /// </summary>
    struct WorkingBitreadState
    {
        public int get_buffer;    /* current bit-extraction buffer */
        public int bits_left;      /* # of unused bits in it */

        /* Pointer needed by jpeg_fill_bit_buffer. */
        public JpegDecompressor cinfo;  /* back link to decompress master record */
    }
    #endregion
}