Render SVG images

README.md

@@ -2,13 +2,11 @@
 
 Very fast Markdown parser & HTML renderer implemented in WebAssembly
 
-- Zero dependencies
-- Portable & safe
+- Zero dependencies (31 kB gzipped)
+- Portable & safe (WASM executes in isolated memory and can run almost anywhere)
 - [Simple API](#api)
-- [Fast and efficient](#benchmarks)
-- JS + WASM is only 31 kB gzipped
-
-Based on [md4c](http://github.com/mity/md4c)
+- [Very fast](#benchmarks)
+- Based on [md4c](http://github.com/mity/md4c) — compliant to the CommonMark specification
 
 
 ## Examples
@@ -91,29 +89,22 @@ See [`test/benchmark`](test/benchmark#readme) for more information.
  * parse reads markdown source at s and converts it to HTML.
  * When output is a byte array, it will be a reference.
  */
-export function parse(s :Source, o? :ParseOptions & { asMemoryView? :never|false }) :string
-export function parse(s :Source, o? :ParseOptions & { asMemoryView :true }) :Uint8Array
+export function parse(s :Source, o? :ParseOptions & { bytes? :never|false }) :string
+export function parse(s :Source, o? :ParseOptions & { bytes :true }) :Uint8Array
 
 /** Markdown source code can be provided as a JavaScript string or UTF8 encoded data */
-type Source = string|ArrayLike<number>
+type Source = string | ArrayLike<number>
 
 /** Options for the parse function */
 export interface ParseOptions {
-  /**
-   * Customize parsing.
-   * If not provided, the following flags are used, equating to github-style parsing:
-   *   COLLAPSE_WHITESPACE
-   *   PERMISSIVE_ATX_HEADERS
-   *   PERMISSIVE_URL_AUTO_LINKS
-   *   STRIKETHROUGH
-   *   TABLES
-   *   TASK_LISTS
-   */
+  /** Customize parsing. Defaults to ParseFlags.DEFAULT */
   parseFlags? :ParseFlags
 
+  /** Select output format. Defaults to "html" */
+  format? : "html" | "xhtml"
+
   /**
-   * asMemoryView=true causes parse() to return a view of heap memory as a Uint8Array,
-   * instead of a string.
+   * bytes=true causes parse() to return the result as a Uint8Array instead of a string.
    *
    * The returned Uint8Array is only valid until the next call to parse().
    * If you need to keep the returned data around, call Uint8Array.slice() to make a copy,
@@ -122,9 +113,36 @@ export interface ParseOptions {
    * This only provides a performance benefit when you never need to convert the output
    * to a string. In most cases you're better off leaving this unset or false.
    */
+  bytes? :boolean
+
+  /** Allow "javascript:" in links */
+  allowJSURIs? :boolean
+
+  /**
+   * Optional callback which if provided is called for each code block.
+   * langname holds the "language tag", if any, of the block.
+   *
+   * The returned value is inserted into the resulting HTML verbatim, without HTML escaping.
+   * Thus, you should take care of properly escaping any special HTML characters.
+   *
+   * If the function returns null or undefined, or an exception occurs, the body will be
+   * included as-is after going through HTML escaping.
+   *
+   * Note that use of this callback has an adverse impact on performance as it casues
+   * calls and data to be bridged between WASM and JS on every invocation.
+   */
+  onCodeBlock? :(langname :string, body :UTF8Bytes) => Uint8Array|string|null|undefined
+
+  /** @depreceated use "bytes" instead (v1.1.1) */
   asMemoryView? :boolean
 }
 
+/** UTF8Bytes is a Uint8Array representing UTF8 text  */
+export interface UTF8Bytes extends Uint8Array {
+  /** toString returns a UTF8 decoded string (lazily decoded and cached) */
+  toString() :string
+}
+
 /** Flags that customize Markdown parsing */
 export enum ParseFlags {
   /** In TEXT, collapse non-trivial whitespace into single ' ' */ COLLAPSE_WHITESPACE,
@@ -140,13 +158,24 @@ export enum ParseFlags {
   /** Enable tables extension. */                                 TABLES,
   /** Enable task list extension. */                              TASK_LISTS,
   /** Enable wiki links extension. */                             WIKI_LINKS,
+  /** Enable underline extension (disables '_' for emphasis) */   UNDERLINE,
 
-  /** Default flags */                                            DEFAULT,
-  /** Shorthand for NO_HTML_BLOCKS | NO_HTML_SPANS */             NO_HTML,
+  /** Default flags are:
+   *    COLLAPSE_WHITESPACE |
+   *    PERMISSIVE_ATX_HEADERS |
+   *    PERMISSIVE_URL_AUTO_LINKS |
+   *    STRIKETHROUGH |
+   *    TABLES |
+   *    TASK_LISTS
+   */
+  DEFAULT,
+
+  /** Shorthand for NO_HTML_BLOCKS | NO_HTML_SPANS */
+  NO_HTML,
 }
 ```
 
-See `markdown.d.ts`
+See [`markdown.d.ts`](markdown.d.ts)
 
 
 ## Building from source

docs/index.html

@@ -106,7 +106,7 @@ h6:hover .anchor:before {
     <c>
       <p>
         Fast Markdown parser and HTML renderer implemented in WebAssembly.
-        <a href="https://github.com/rsms/markdown-wasm">Learn more &amp; download on GitHub</a>
+        <a href="https://github.com/rsms/markdown-wasm">Get it from GitHub</a>
       </p>
     </c>
 

example/example.html

@@ -4,7 +4,7 @@
 <p>This is a paragraph</p>
 <h2><a id="another" class="anchor" aria-hidden="true" href="#another"></a>Another</h2>
 <p>This is a paragraph with style <em>italic</em> <em>italic</em> <b>bold</b> <b>bold</b></p>
-<p><img src="https://rsms.me/image.png" alt="image">
+<p><img src="https://rsms.me/raster/examples/image1.jpg" alt="image">
 <img src="https://rsms.me/image.png?without-alt" alt=""></p>
 <p><em>Hello <a href="https://rsms.me/">link</a> lol</em></p>
 <p>Hello <a href="https://rsms.me/"><em>link</em></a> lol &quot;cat&quot;</p>
@@ -13,16 +13,16 @@
 <p>line 1
 line 2</p>
 <h2><a id="code-poetry" class="anchor" aria-hidden="true" href="#code-poetry"></a>Code &amp; Poetry</h2>
-<pre><code>You can also indent
-blocks to display
-code or poetry.
+<pre><code>YOU CAN ALSO INDENT
+BLOCKS TO DISPLAY
+CODE OR POETRY.
 
-Indented code/poetry blocks
-can be hard-wrapped.
+INDENTED CODE/POETRY BLOCKS
+CAN BE HARD-WRAPPED.
 </code></pre>
 <p><b>Or, wrap your code in three backticks:</b></p>
-<pre><code class="language-js">function codeBlocks() {
-  return &quot;Can be inserted&quot;
+<pre><code class="language-js">FUNCTION CODEBLOCKS() {
+  RETURN &quot;CAN BE INSERTED&quot;
 }
 </code></pre>
 <h3><a id="block-quotes" class="anchor" aria-hidden="true" href="#block-quotes"></a>Block Quotes</h3>
@@ -79,3 +79,4 @@ breaks.</p>
 <h2><a id="anot-her" class="anchor" aria-hidden="true" href="#anot-her"></a>Anöt######her!</h2>
 <h2><a id="anot-her" class="anchor" aria-hidden="true" href="#anot-her"></a>?!Anöt//her!!</h2>
 <h2><a id="" class="anchor" aria-hidden="true" href="#"></a>?!!</h2>
+<p><a href="">XSS test</a></p>

example/example.js

@@ -1,10 +1,15 @@
-const RELEASE = process.argv.includes("-release")
-
 const fs = require("fs")
-const md = RELEASE ? require("../build/release/md.js") : require("../build/debug/md.js")
+const md = require("../dist/markdown.node.js")
+// const md = require("../build/debug/markdown.node.js")
 
 const source = fs.readFileSync(__dirname + "/example.md")
-const outbuf = md.parse(source, { asMemoryView: true })
+const outbuf = md.parse(source, {
+  bytes: true,
+  onCodeBlock(lang, body) {
+    console.log(`onCodeBlock (${lang})`)
+    return html_escape(body.toString().toUpperCase())
+  },
+})
 const outfile = __dirname + "/example.html"
 console.log("write", outfile)
 fs.writeFileSync(outfile, outbuf)
@@ -12,13 +17,37 @@ fs.writeFileSync(outfile, outbuf)
 console.log(fs.readFileSync(outfile, "utf8"))
 
 // mini benchmark
-if (RELEASE) {
-  console.log("benchmark start")
+if (process.argv.includes("-bench")) {
+  benchmark("bytes", {
+    bytes: true,
+  })
+  benchmark("bytes + onCodeBlock", {
+    bytes: true,
+    onCodeBlock(lang, body) {
+      return null // causes the body to be HTML-escaped & included as is
+    },
+  })
+}
+
+function benchmark(name, options) {
+  console.log(`benchmark start ${name} (sampling ~2s of data)`)
   const timeStart = Date.now()
-  const iterations = 10000
-  for (let i = 0; i < iterations; i++) {
-    global["dont-optimize-away"] = md.parse(source, { asMemoryView: true })
+  const N = 1000, T = 2000
+  let ntotal = 0
+  while (Date.now() - timeStart < 2000) {
+    for (let i = 0; i < N; i++) {
+      global["dont-optimize-away"] = md.parse(source, options)
+    }
+    ntotal += N
   }
   const timeSpent = Date.now() - timeStart
-  console.log(`benchmark end -- avg parse time: ${(timeSpent / iterations).toFixed(2)}ms`)
+  console.log(
+    `benchmark end ${name} -- avg parse time: ` +
+    `${((timeSpent / ntotal) * 1000).toFixed(1)}us`)
+}
+
+function html_escape(str) {
+  return str.replace(/[&<>'"]/g, tag => ({
+    '&': '&amp;','<': '&lt;','>': '&gt;',"'": '&#39;','"': '&quot;'
+  }[tag]))
 }

example/example.md

@@ -82,3 +82,5 @@ function codeBlocks() {
 ## ?!Anöt//her!!
 
 ## ?!!
+
+[XSS test](javAscRipt:alert("xss"))

markdown.d.ts

@@ -2,8 +2,8 @@
  * parse reads markdown source at s and converts it to HTML.
  * When output is a byte array, it will be a reference.
  */
-export function parse(s :Source, o? :ParseOptions & { asMemoryView? :never|false }) :string
-export function parse(s :Source, o? :ParseOptions & { asMemoryView :true }) :Uint8Array
+export function parse(s :Source, o? :ParseOptions & { bytes? :never|false }) :string
+export function parse(s :Source, o? :ParseOptions & { bytes :true }) :Uint8Array
 
 /** Markdown source code can be provided as a JavaScript string or UTF8 encoded data */
 type Source = string | ArrayLike<number>
@@ -17,8 +17,7 @@ export interface ParseOptions {
   format? : "html" | "xhtml"
 
   /**
-   * asMemoryView=true causes parse() to return a view of heap memory as a Uint8Array,
-   * instead of a string.
+   * bytes=true causes parse() to return the result as a Uint8Array instead of a string.
    *
    * The returned Uint8Array is only valid until the next call to parse().
    * If you need to keep the returned data around, call Uint8Array.slice() to make a copy,
@@ -27,9 +26,36 @@ export interface ParseOptions {
    * This only provides a performance benefit when you never need to convert the output
    * to a string. In most cases you're better off leaving this unset or false.
    */
+  bytes? :boolean
+
+  /** Allow "javascript:" in links */
+  allowJSURIs? :boolean
+
+  /**
+   * Optional callback which if provided is called for each code block.
+   * langname holds the "language tag", if any, of the block.
+   *
+   * The returned value is inserted into the resulting HTML verbatim, without HTML escaping.
+   * Thus, you should take care of properly escaping any special HTML characters.
+   *
+   * If the function returns null or undefined, or an exception occurs, the body will be
+   * included as-is after going through HTML escaping.
+   *
+   * Note that use of this callback has an adverse impact on performance as it casues
+   * calls and data to be bridged between WASM and JS on every invocation.
+   */
+  onCodeBlock? :(langname :string, body :UTF8Bytes) => Uint8Array|string|null|undefined
+
+  /** @depreceated use "bytes" instead (v1.1.1) */
   asMemoryView? :boolean
 }
 
+/** UTF8Bytes is a Uint8Array representing UTF8 text  */
+export interface UTF8Bytes extends Uint8Array {
+  /** toString returns a UTF8 decoded string (lazily decoded and cached) */
+  toString() :string
+}
+
 /** Flags that customize Markdown parsing */
 export enum ParseFlags {
   /** In TEXT, collapse non-trivial whitespace into single ' ' */ COLLAPSE_WHITESPACE,

misc/dist.sh

@@ -49,6 +49,9 @@ echo "" ; echo "wasmc -clean"
 echo "" ; echo "./test/test.sh"
 ./test/test.sh
 
+# web site
+cp dist/markdown.js dist/markdown.js.map dist/markdown.wasm ./docs/
+
 # commit, tag and push git
 echo "Ready to commit, publish & push:"
 echo ""

package-lock.json

@@ -1,13 +1,31 @@
 {
   "name": "markdown-wasm",
-  "version": "1.0.0",
-  "lockfileVersion": 1,
+  "version": "1.2.0",
+  "lockfileVersion": 2,
   "requires": true,
+  "packages": {
+    "": {
+      "version": "1.2.0",
+      "license": "MIT",
+      "devDependencies": {
+        "wasmc": "^2.2.1"
+      }
+    },
+    "node_modules/wasmc": {
+      "version": "2.2.1",
+      "resolved": "https://registry.npmjs.org/wasmc/-/wasmc-2.2.1.tgz",
+      "integrity": "sha512-uV/nhZdfnffBfLv2WJWWfRoJwk+5phhPobg7fqnJoRpXdUV5eShRb6tOXoco987cTxarkBymYwGufMdC4dPXAw==",
+      "dev": true,
+      "bin": {
+        "wasmc": "wasmc"
+      }
+    }
+  },
   "dependencies": {
     "wasmc": {
-      "version": "2.2.0",
-      "resolved": "https://registry.npmjs.org/wasmc/-/wasmc-2.2.0.tgz",
-      "integrity": "sha512-Pp2B42WuhL7JjslKDlGbpoV8WYpGAz0ct1XYTQWIN99cVsj7nfEuhD/jKwivLJHOFSOzMZZqvQH4m5BPQmTmCg==",
+      "version": "2.2.1",
+      "resolved": "https://registry.npmjs.org/wasmc/-/wasmc-2.2.1.tgz",
+      "integrity": "sha512-uV/nhZdfnffBfLv2WJWWfRoJwk+5phhPobg7fqnJoRpXdUV5eShRb6tOXoco987cTxarkBymYwGufMdC4dPXAw==",
       "dev": true
     }
   }

package.json

@@ -1,6 +1,6 @@
 {
   "name": "markdown-wasm",
-  "version": "1.1.0",
+  "version": "1.2.0",
   "description": "Markdown parser and html generator implemented in WebAssembly",
   "browser": "dist/markdown.js",
   "main": "dist/markdown.node.js",
@@ -35,6 +35,6 @@
   "author": "Rasmus Andersson <https://rsms.me/>",
   "license": "MIT",
   "devDependencies": {
-    "wasmc": "^2.2.0"
+    "wasmc": "^2.2.1"
   }
 }

src/common.h

@@ -54,7 +54,21 @@ typedef int32_t  i32;
 #endif
 #if DEBUG
 #include <stdio.h>
-  #define dlog(...)  printf(__VA_ARGS__)
+  #define dlog(fmt, ...)  printf(fmt "\n", ##__VA_ARGS__)
 #else
   #define dlog(...)
 #endif /* DEBUG > 0 */
+
+#include "wbuf.h"
+
+// these should be in sync with "OutputFlags" in md.js
+typedef enum OutputFlags {
+  OutputFlagHTML       = 1 << 0,
+  OutputFlagXHTML      = 1 << 1,
+  OutputFlagAllowJSURI = 1 << 2, // allow "javascript:" URIs in links
+} OutputFlags;
+
+typedef int(*JSTextFilterFun)(
+  const char*  metaptr, u32 metalen,
+  const char*  inptr,   u32 inlen,
+  const char** outptrp); // return outlen

src/fmt_html.c

@@ -25,17 +25,18 @@
 
 #include <string.h>
 #include <ctype.h>
+#include <strings.h>
 
 #include "common.h"
 #include "fmt_html.h"
 #include "md4c.h"
 
-typedef struct HtmlRenderer_st {
-  WBuf* outbuf;
-  int   imgnest;
-  int   addanchor;
-  u32   flags;
-} HtmlRenderer;
+// typedef struct FmtHTML_st {
+//   WBuf* outbuf;
+//   int   imgnest;
+//   int   addanchor;
+//   u32   flags;
+// } FmtHTML;
 
 
 static char htmlEscapeMap[256] = {
@@ -61,20 +62,20 @@ static char htmlEscapeMap[256] = {
 static const char ucReplacementUTF8[] = { 0xef, 0xbf, 0xbd };
 
 
-static inline void render_text(HtmlRenderer* r, const char* pch, size_t len) {
+static inline void render_text(FmtHTML* r, const char* pch, size_t len) {
   WBufAppendBytes(r->outbuf, pch, len);
 }
 
-static inline void render_literal(HtmlRenderer* r, const char* cs) {
+static inline void render_literal(FmtHTML* r, const char* cs) {
   WBufAppendBytes(r->outbuf, cs, strlen(cs));
 }
 
-static inline void render_char(HtmlRenderer* r, char c) {
+static inline void render_char(FmtHTML* r, char c) {
   WBufAppendc(r->outbuf, c);
 }
 
 
-static void render_html_escaped(HtmlRenderer* r, const char* data, size_t size) {
+static void render_html_escaped(FmtHTML* r, const char* data, size_t size) {
   MD_OFFSET beg = 0;
   MD_OFFSET off = 0;
 
@@ -167,7 +168,7 @@ static size_t WBufAppendSlug(WBuf* b, const char* pch, size_t len) {
 }
 
 
-static void render_attribute(HtmlRenderer* r, const MD_ATTRIBUTE* attr) {
+static void render_attribute(FmtHTML* r, const MD_ATTRIBUTE* attr) {
   int i;
   for (i = 0; attr->substr_offsets[i] < attr->size; i++) {
     MD_TEXTTYPE type = attr->substr_types[i];
@@ -183,7 +184,7 @@ static void render_attribute(HtmlRenderer* r, const MD_ATTRIBUTE* attr) {
 }
 
 
-static void render_open_ol_block(HtmlRenderer* r, const MD_BLOCK_OL_DETAIL* det) {
+static void render_open_ol_block(FmtHTML* r, const MD_BLOCK_OL_DETAIL* det) {
   if (det->start == 1) {
     render_literal(r, "<ol>\n");
   } else {
@@ -193,7 +194,7 @@ static void render_open_ol_block(HtmlRenderer* r, const MD_BLOCK_OL_DETAIL* det)
   }
 }
 
-static void render_open_li_block(HtmlRenderer* r, const MD_BLOCK_LI_DETAIL* det) {
+static void render_open_li_block(FmtHTML* r, const MD_BLOCK_LI_DETAIL* det) {
   if (det->is_task) {
     render_literal(r, "<li class=\"task-list-item\"><input type=\"checkbox\" disabled");
     if (det->task_mark == 'x' || det->task_mark == 'X') {
@@ -205,7 +206,7 @@ static void render_open_li_block(HtmlRenderer* r, const MD_BLOCK_LI_DETAIL* det)
   }
 }
 
-static void render_open_code_block(HtmlRenderer* r, const MD_BLOCK_CODE_DETAIL* det) {
+static void render_open_code_block(FmtHTML* r, const MD_BLOCK_CODE_DETAIL* det) {
   render_literal(r, "<pre><code");
   if (det->lang.text != NULL) {
     render_literal(r, " class=\"language-");
@@ -213,9 +214,41 @@ static void render_open_code_block(HtmlRenderer* r, const MD_BLOCK_CODE_DETAIL*
     render_char(r, '"');
   }
   render_char(r, '>');
+  r->codeBlockNest++;
 }
 
-static void render_open_td_block(HtmlRenderer* r, bool isTH, const MD_BLOCK_TD_DETAIL* det) {
+static void render_close_code_block(FmtHTML* r, const MD_BLOCK_CODE_DETAIL* det) {
+  dlog("end code block (lang \"%.*s\")", (int)det->lang.size, det->lang.text);
+
+  r->codeBlockNest--;
+
+  if (r->onCodeBlock) {
+    const char* text = r->tmpbuf.start;
+    size_t len = WBufLen(&r->tmpbuf);
+
+    int outlen = -1;
+
+    if (len <= 0x7FFFFFFF) {
+      const char* outptr = NULL;
+      outlen = r->onCodeBlock(det->lang.text, (u32)det->lang.size, text, (u32)len, &outptr);
+      if (outlen > 0 && outptr != NULL)
+        WBufAppendBytes(r->outbuf, outptr, (size_t)outlen);
+      if (outptr != NULL)
+        free((void*)outptr);
+    }
+
+    if (outlen < 0) {
+      // The function failed or opted out of taking care of formatting
+      render_html_escaped(r, text, len);
+    }
+
+    WBufReset(&r->tmpbuf);
+  }
+
+  render_literal(r, "</code></pre>\n");
+}
+
+static void render_open_td_block(FmtHTML* r, bool isTH, const MD_BLOCK_TD_DETAIL* det) {
   render_text(r, isTH ? "<th" : "<td", 3);
   switch (det->align) {
     case MD_ALIGN_LEFT:   render_literal(r, " align=\"left\">"); break;
@@ -225,9 +258,21 @@ static void render_open_td_block(HtmlRenderer* r, bool isTH, const MD_BLOCK_TD_D
   }
 }
 
-static void render_open_a_span(HtmlRenderer* r, const MD_SPAN_A_DETAIL* det) {
+static bool is_javascript_uri(const MD_CHAR* text, size_t len) {
+  return (
+    len >= strlen("javascript:") &&
+    strncasecmp(text, "javascript:", strlen("javascript:")) == 0
+  );
+}
+
+static void render_open_a_span(FmtHTML* r, const MD_SPAN_A_DETAIL* det) {
   render_literal(r, "<a href=\"");
-  render_attribute(r, &det->href);
+  // skip "javascript:" URIs unless explicitly allowed
+  if ((r->flags & OutputFlagAllowJSURI) != 0 ||
+      !is_javascript_uri(det->href.text, det->href.size))
+  {
+    render_attribute(r, &det->href);
+  }
   if (det->title.text != NULL) {
     render_literal(r, "\" title=\"");
     render_attribute(r, &det->title);
@@ -235,23 +280,23 @@ static void render_open_a_span(HtmlRenderer* r, const MD_SPAN_A_DETAIL* det) {
   render_literal(r, "\">");
 }
 
-static void render_open_img_span(HtmlRenderer* r, const MD_SPAN_IMG_DETAIL* det) {
+static void render_open_img_span(FmtHTML* r, const MD_SPAN_IMG_DETAIL* det) {
   render_literal(r, "<img src=\"");
   render_attribute(r, &det->src);
   render_literal(r, "\" alt=\"");
   r->imgnest++;
 }
 
-static void render_close_img_span(HtmlRenderer* r, const MD_SPAN_IMG_DETAIL* det) {
+static void render_close_img_span(FmtHTML* r, const MD_SPAN_IMG_DETAIL* det) {
   if(det->title.text != NULL) {
     render_literal(r, "\" title=\"");
     render_attribute(r, &det->title);
   }
-  render_literal(r, (r->flags & MD_HTML_FLAG_XHTML) ? "\"/>" : "\">");
+  render_literal(r, (r->flags & OutputFlagXHTML) ? "\"/>" : "\">");
   r->imgnest--;
 }
 
-static void render_open_wikilink_span(HtmlRenderer* r, const MD_SPAN_WIKILINK_DETAIL* det) {
+static void render_open_wikilink_span(FmtHTML* r, const MD_SPAN_WIKILINK_DETAIL* det) {
   render_literal(r, "<x-wikilink data-target=\"");
   render_attribute(r, &det->target);
   render_literal(r, "\">");
@@ -266,30 +311,30 @@ static void render_open_wikilink_span(HtmlRenderer* r, const MD_SPAN_WIKILINK_DE
 
 static int enter_block_callback(MD_BLOCKTYPE type, void* detail, void* userdata) {
   static const MD_CHAR* head[6] = { "<h1>", "<h2>", "<h3>", "<h4>", "<h5>", "<h6>" };
-  HtmlRenderer* r = (HtmlRenderer*) userdata;
+  FmtHTML* r = (FmtHTML*) userdata;
 
   switch(type) {
-    case MD_BLOCK_DOC:      /* noop */ break;
-    case MD_BLOCK_QUOTE:    render_literal(r, "<blockquote>\n"); break;
-    case MD_BLOCK_UL:       render_literal(r, "<ul>\n"); break;
-    case MD_BLOCK_OL:       render_open_ol_block(r, (const MD_BLOCK_OL_DETAIL*)detail); break;
-    case MD_BLOCK_LI:       render_open_li_block(r, (const MD_BLOCK_LI_DETAIL*)detail); break;
-    case MD_BLOCK_HR:       render_literal(r, (r->flags & MD_HTML_FLAG_XHTML) ? "<hr/>\n" : "<hr>\n"); break;
+    case MD_BLOCK_DOC:   /* noop */ break;
+    case MD_BLOCK_QUOTE: render_literal(r, "<blockquote>\n"); break;
+    case MD_BLOCK_UL:    render_literal(r, "<ul>\n"); break;
+    case MD_BLOCK_OL:    render_open_ol_block(r, (const MD_BLOCK_OL_DETAIL*)detail); break;
+    case MD_BLOCK_LI:    render_open_li_block(r, (const MD_BLOCK_LI_DETAIL*)detail); break;
+    case MD_BLOCK_HR:    render_literal(r, (r->flags & OutputFlagXHTML) ? "<hr/>\n" : "<hr>\n"); break;
     case MD_BLOCK_H:
     {
       render_literal(r, head[((MD_BLOCK_H_DETAIL*)detail)->level - 1]);
       r->addanchor = 1;
       break;
     }
-    case MD_BLOCK_CODE:     render_open_code_block(r, (const MD_BLOCK_CODE_DETAIL*) detail); break;
-    case MD_BLOCK_HTML:     /* noop */ break;
-    case MD_BLOCK_P:        render_literal(r, "<p>"); break;
-    case MD_BLOCK_TABLE:    render_literal(r, "<table>\n"); break;
-    case MD_BLOCK_THEAD:    render_literal(r, "<thead>\n"); break;
-    case MD_BLOCK_TBODY:    render_literal(r, "<tbody>\n"); break;
-    case MD_BLOCK_TR:       render_literal(r, "<tr>\n"); break;
-    case MD_BLOCK_TH:       render_open_td_block(r, true, (MD_BLOCK_TD_DETAIL*)detail); break;
-    case MD_BLOCK_TD:       render_open_td_block(r, false, (MD_BLOCK_TD_DETAIL*)detail); break;
+    case MD_BLOCK_CODE:  render_open_code_block(r, (const MD_BLOCK_CODE_DETAIL*) detail); break;
+    case MD_BLOCK_HTML:  /* noop */ break;
+    case MD_BLOCK_P:     render_literal(r, "<p>"); break;
+    case MD_BLOCK_TABLE: render_literal(r, "<table>\n"); break;
+    case MD_BLOCK_THEAD: render_literal(r, "<thead>\n"); break;
+    case MD_BLOCK_TBODY: render_literal(r, "<tbody>\n"); break;
+    case MD_BLOCK_TR:    render_literal(r, "<tr>\n"); break;
+    case MD_BLOCK_TH:    render_open_td_block(r, true, (MD_BLOCK_TD_DETAIL*)detail); break;
+    case MD_BLOCK_TD:    render_open_td_block(r, false, (MD_BLOCK_TD_DETAIL*)detail); break;
   }
 
   return 0;
@@ -297,32 +342,32 @@ static int enter_block_callback(MD_BLOCKTYPE type, void* detail, void* userdata)
 
 static int leave_block_callback(MD_BLOCKTYPE type, void* detail, void* userdata) {
   static const MD_CHAR* head[6] = { "</h1>\n", "</h2>\n", "</h3>\n", "</h4>\n", "</h5>\n", "</h6>\n" };
-  HtmlRenderer* r = (HtmlRenderer*) userdata;
+  FmtHTML* r = (FmtHTML*) userdata;
 
   switch(type) {
-    case MD_BLOCK_DOC:      /*noop*/ break;
-    case MD_BLOCK_QUOTE:    render_literal(r, "</blockquote>\n"); break;
-    case MD_BLOCK_UL:       render_literal(r, "</ul>\n"); break;
-    case MD_BLOCK_OL:       render_literal(r, "</ol>\n"); break;
-    case MD_BLOCK_LI:       render_literal(r, "</li>\n"); break;
-    case MD_BLOCK_HR:       /*noop*/ break;
-    case MD_BLOCK_H:        render_literal(r, head[((MD_BLOCK_H_DETAIL*)detail)->level - 1]); break;
-    case MD_BLOCK_CODE:     render_literal(r, "</code></pre>\n"); break;
-    case MD_BLOCK_HTML:     /* noop */ break;
-    case MD_BLOCK_P:        render_literal(r, "</p>\n"); break;
-    case MD_BLOCK_TABLE:    render_literal(r, "</table>\n"); break;
-    case MD_BLOCK_THEAD:    render_literal(r, "</thead>\n"); break;
-    case MD_BLOCK_TBODY:    render_literal(r, "</tbody>\n"); break;
-    case MD_BLOCK_TR:       render_literal(r, "</tr>\n"); break;
-    case MD_BLOCK_TH:       render_literal(r, "</th>\n"); break;
-    case MD_BLOCK_TD:       render_literal(r, "</td>\n"); break;
+    case MD_BLOCK_DOC:   /*noop*/ break;
+    case MD_BLOCK_QUOTE: render_literal(r, "</blockquote>\n"); break;
+    case MD_BLOCK_UL:    render_literal(r, "</ul>\n"); break;
+    case MD_BLOCK_OL:    render_literal(r, "</ol>\n"); break;
+    case MD_BLOCK_LI:    render_literal(r, "</li>\n"); break;
+    case MD_BLOCK_HR:    /*noop*/ break;
+    case MD_BLOCK_H:     render_literal(r, head[((MD_BLOCK_H_DETAIL*)detail)->level - 1]); break;
+    case MD_BLOCK_CODE:  render_close_code_block(r, (const MD_BLOCK_CODE_DETAIL*)detail); break;
+    case MD_BLOCK_HTML:  /* noop */ break;
+    case MD_BLOCK_P:     render_literal(r, "</p>\n"); break;
+    case MD_BLOCK_TABLE: render_literal(r, "</table>\n"); break;
+    case MD_BLOCK_THEAD: render_literal(r, "</thead>\n"); break;
+    case MD_BLOCK_TBODY: render_literal(r, "</tbody>\n"); break;
+    case MD_BLOCK_TR:    render_literal(r, "</tr>\n"); break;
+    case MD_BLOCK_TH:    render_literal(r, "</th>\n"); break;
+    case MD_BLOCK_TD:    render_literal(r, "</td>\n"); break;
   }
 
   return 0;
 }
 
 static int enter_span_callback(MD_SPANTYPE type, void* detail, void* userdata) {
-  HtmlRenderer* r = (HtmlRenderer*) userdata;
+  FmtHTML* r = (FmtHTML*) userdata;
 
   if(r->imgnest > 0) {
     /* We are inside a Markdown image label. Markdown allows to use any
@@ -347,8 +392,8 @@ static int enter_span_callback(MD_SPANTYPE type, void* detail, void* userdata) {
     case MD_SPAN_EM:                render_literal(r, "<em>"); break;
     case MD_SPAN_STRONG:            render_literal(r, "<b>"); break;
     case MD_SPAN_U:                 render_literal(r, "<u>"); break;
-    case MD_SPAN_A:                 render_open_a_span(r, (MD_SPAN_A_DETAIL*) detail); break;
-    case MD_SPAN_IMG:               render_open_img_span(r, (MD_SPAN_IMG_DETAIL*) detail); break;
+    case MD_SPAN_A:                 render_open_a_span(r, (MD_SPAN_A_DETAIL*)detail); break;
+    case MD_SPAN_IMG:               render_open_img_span(r, (MD_SPAN_IMG_DETAIL*)detail); break;
     case MD_SPAN_CODE:              render_literal(r, "<code>"); break;
     case MD_SPAN_DEL:               render_literal(r, "<del>"); break;
     case MD_SPAN_LATEXMATH:         render_literal(r, "<x-equation>"); break;
@@ -360,7 +405,7 @@ static int enter_span_callback(MD_SPANTYPE type, void* detail, void* userdata) {
 }
 
 static int leave_span_callback(MD_SPANTYPE type, void* detail, void* userdata) {
-  HtmlRenderer* r = (HtmlRenderer*) userdata;
+  FmtHTML* r = (FmtHTML*) userdata;
 
   if(r->imgnest > 0) {
     /* Ditto as in enter_span_callback(), except we have to allow the
@@ -387,7 +432,12 @@ static int leave_span_callback(MD_SPANTYPE type, void* detail, void* userdata) {
 }
 
 static int text_callback(MD_TEXTTYPE type, const MD_CHAR* text, MD_SIZE size, void* userdata) {
-  HtmlRenderer* r = (HtmlRenderer*) userdata;
+  FmtHTML* r = (FmtHTML*) userdata;
+
+  if (r->codeBlockNest && r->onCodeBlock) {
+    WBufAppendBytes(&r->tmpbuf, text, size);
+    return 0;
+  }
 
   if (r->addanchor) {
     r->addanchor = 0;
@@ -409,18 +459,18 @@ static int text_callback(MD_TEXTTYPE type, const MD_CHAR* text, MD_SIZE size, vo
     }
   }
 
-  switch(type) {
+  switch (type) {
     case MD_TEXT_NULLCHAR:  render_text(r, ucReplacementUTF8, sizeof(ucReplacementUTF8)); break;
     case MD_TEXT_BR:
       render_literal(
         r,
         r->imgnest == 0 ?
-          ((r->flags & MD_HTML_FLAG_XHTML) ? "<br/>\n" : "<br>\n") :
+          ((r->flags & OutputFlagXHTML) ? "<br/>\n" : "<br>\n") :
           " "
       );
       break;
 
-    render_literal(r, (r->flags & MD_HTML_FLAG_XHTML) ? "<hr/>\n" : "<hr>\n"); break;
+    render_literal(r, (r->flags & OutputFlagXHTML) ? "<hr/>\n" : "<hr>\n"); break;
 
     case MD_TEXT_SOFTBR:    render_literal(r, (r->imgnest == 0 ? "\n" : " ")); break;
     case MD_TEXT_HTML:      render_text(r, text, size); break;
@@ -435,18 +485,15 @@ static int text_callback(MD_TEXTTYPE type, const MD_CHAR* text, MD_SIZE size, vo
 //   dlog("MD4C: %s\n", msg);
 // }
 
-int fmt_html(
-  const MD_CHAR* input,
-  MD_SIZE input_size,
-  WBuf* outbuf,
-  u32 parser_flags,
-  u32 render_flags
-) {
-  HtmlRenderer render = { outbuf, 0, 0, render_flags };
+int fmt_html(const MD_CHAR* input, MD_SIZE input_size, FmtHTML* fmt) {
+  fmt->imgnest = 0;
+  fmt->addanchor = 0;
+  fmt->codeBlockNest = 0;
+  fmt->tmpbuf = (WBuf){0};
 
   MD_PARSER parser = {
     0,
-    parser_flags,
+    fmt->parserFlags,
     enter_block_callback,
     leave_block_callback,
     enter_span_callback,
@@ -456,5 +503,11 @@ int fmt_html(
     NULL
   };
 
-  return md_parse(input, input_size, &parser, (void*) &render);
+  WBufInit(&fmt->tmpbuf);
+
+  int res = md_parse(input, input_size, &parser, (void*)fmt);
+
+  WBufFree(&fmt->tmpbuf);
+
+  return res;
 }

src/fmt_html.h

@@ -1,6 +1,18 @@
 #pragma once
-#include "wbuf.h"
 
-#define MD_HTML_FLAG_XHTML 0x0008 // instead of e.g. <br>, generate <br/>
+typedef struct FmtHTML {
+  OutputFlags flags;
+  u32         parserFlags; // passed along to md_parse
+  WBuf*       outbuf;
 
-int fmt_html(const char* input, u32 inputlen, WBuf* outbuf, u32 parserFlags, u32 renderFlags);
+  // optional callbacks
+  JSTextFilterFun onCodeBlock;
+
+  // internal state
+  int  imgnest;
+  int  addanchor;
+  int  codeBlockNest;
+  WBuf tmpbuf;
+} FmtHTML;
+
+int fmt_html(const char* input, u32 inputlen, FmtHTML* fmt);

src/md.c

@@ -1,16 +1,9 @@
 #include <ctype.h>
 #include "common.h"
 #include "wlib.h"
-#include "wbuf.h"
 #include "fmt_html.h"
 // #include "fmt_json.h"
 
-// these should be in sync with "OutputFlags" in md.js
-typedef enum OutputFlags {
-  OutputFlagHTML  = 1 << 0,
-  OutputFlagXHTML = 1 << 1,
-} OutputFlags;
-
 typedef enum ErrorCode {
   ERR_NONE,
   ERR_MD_PARSE,
@@ -20,7 +13,7 @@ typedef enum ErrorCode {
 
 #if DEBUG
 void __attribute__((constructor)) init() {
-  dlog("WASM INIT\n");
+  dlog("WASM INIT");
 }
 #endif
 
@@ -35,21 +28,24 @@ export size_t parseUTF8(
   u32 inbuflen,
   u32 parser_flags,
   OutputFlags outflags,
-  const char** outptr
+  const char** outptr,
+  JSTextFilterFun onCodeBlock
 ) {
-  dlog("parseUTF8 called with inbufptr=%p  inbuflen=%u\n", inbufptr, inbuflen);
+  dlog("parseUTF8 called with inbufptr=%p  inbuflen=%u", inbufptr, inbuflen);
 
   WBufReset(&outbuf);
 
-  if (outflags & OutputFlagHTML) {
+  if ((outflags & OutputFlagHTML) || (outflags & OutputFlagXHTML)) {
     WBufReserve(&outbuf, inbuflen * 2);  // approximate output size to minimize reallocations
 
-    u32 render_flags = 0;
-    if (outflags & OutputFlagXHTML) {
-      render_flags |= MD_HTML_FLAG_XHTML;
-    }
+    FmtHTML fmt = {
+      .flags = outflags,
+      .parserFlags = parser_flags,
+      .outbuf = &outbuf,
+      .onCodeBlock = onCodeBlock,
+    };
 
-    if (fmt_html(inbufptr, inbuflen, &outbuf, parser_flags, render_flags) != 0) {
+    if (fmt_html(inbufptr, inbuflen, &fmt) != 0) {
       // fmt_html returns status of md_parse which only fails in extreme cases
       // like when out of memory. md4c does not provide error codes or error messages.
       WErrSet(ERR_MD_PARSE, "md parser error");

src/md.js

@@ -1,4 +1,10 @@
-import { utf8, withTmpBytePtr, withOutPtr, werrCheck } from "./wlib"
+import {
+  utf8,
+  withTmpBytePtr,
+  withOutPtr,
+  werrCheck,
+  mallocbuf,
+} from "./wlib"
 
 export const ready = Module.ready
 
@@ -35,12 +41,14 @@ export const ParseFlags = {
   NO_HTML: 0x0020 | 0x0040, // NO_HTML_BLOCKS | NO_HTML_SPANS
 }
 
-// these should be in sync with "OutputFlags" in md.c
+// these should be in sync with "OutputFlags" in common.h
 const OutputFlags = {
-  HTML:  1 << 0, // Output HTML
-  XHTML: 1 << 1, // Output XHTML (only has effect with HTML flag set)
+  HTML:       1 << 0, // Output HTML
+  XHTML:      1 << 1, // Output XHTML (only has effect with HTML flag set)
+  AllowJSURI: 1 << 2, // Allow "javascript:" URIs
 }
 
+
 export function parse(source, options) {
   options = options || {}
 
@@ -49,7 +57,8 @@ export function parse(source, options) {
     options.parseFlags
   )
 
-  let outputFlags = 0
+  let outputFlags = options.allowJSURIs ? OutputFlags.AllowJSURI : 0
+
   switch (options.format) {
     case "xhtml":
       outputFlags |= OutputFlags.HTML | OutputFlags.XHTML
@@ -66,11 +75,16 @@ export function parse(source, options) {
       throw new Error(`invalid format "${options.format}"`)
   }
 
-  let buf = typeof source == "string" ? utf8.encode(source) : source
+  let onCodeBlockPtr = options.onCodeBlock ? create_onCodeBlock_fn(options.onCodeBlock) : 0
+
+  let buf = as_byte_array(source)
   let outbuf = withOutPtr(outptr => withTmpBytePtr(buf, (inptr, inlen) =>
-    _parseUTF8(inptr, inlen, parseFlags, outputFlags, outptr)
+    _parseUTF8(inptr, inlen, parseFlags, outputFlags, outptr, onCodeBlockPtr)
   ))
 
+  if (options.onCodeBlock)
+    removeFunction(onCodeBlockPtr)
+
   // check for error and throw if needed
   werrCheck()
 
@@ -79,8 +93,62 @@ export function parse(source, options) {
   //   console.log(utf8.decode(outbuf))
   // }
 
-  if (options.asMemoryView) {
+  if (options.bytes || options.asMemoryView)
     return outbuf
-  }
+
   return utf8.decode(outbuf)
 }
+
+
+function create_onCodeBlock_fn(onCodeBlock) {
+  // See https://emscripten.org/docs/porting/connecting_cpp_and_javascript/
+  //   Interacting-with-code.html#calling-javascript-functions-as-function-pointers-from-c
+  //
+  // Function's C type: JSTextFilterFun
+  // (metaptr ptr, metalen ptr, inptr ptr, inlen ptr, outptr ptr) -> outlen int
+  const fnptr = addFunction(function(metaptr, metalen, inptr, inlen, outptr) {
+    try {
+      // lang is the "language" tag, if any, provided with the code block
+      const lang = metalen > 0 ? utf8.decode(HEAPU8.subarray(metaptr, metaptr + metalen)) : ""
+
+      // body is a view into heap memory of the segment of source (UTF8 bytes)
+      const body = HEAPU8.subarray(inptr, inptr + inlen)
+      let bodystr = undefined
+      body.toString = () => (bodystr || (bodystr = utf8.decode(body)))
+
+      // result is the result from the onCodeBlock function
+      let result = null
+      result = onCodeBlock(lang, body)
+
+      if (result === null || result === undefined) {
+        // Callback indicates that it does not wish to filter.
+        // The md.c implementation will html-encode the body.
+        return -1
+      }
+
+      let resbuf = as_byte_array(result)
+      if (resbuf.length > 0) {
+        // copy resbuf to WASM heap memory
+        const resptr = mallocbuf(resbuf, resbuf.length)
+        // write pointer value
+        HEAPU32[outptr >> 2 /* == outptr / 4 */] = resptr
+        // Note: fmt_html.c calls free(resptr)
+      }
+
+      return resbuf.length
+    } catch (err) {
+      console.error(`error in markdown onCodeBlock callback: ${err.stack||err}`)
+      return -1
+    }
+  }, "iiiiii")
+  return fnptr
+}
+
+
+function as_byte_array(something) {
+  if (typeof something == "string")
+    return utf8.encode(something)
+  if (something instanceof Uint8Array)
+    return something
+  return new Uint8Array(something)
+}

src/md4c.c

@@ -26,6 +26,7 @@
 #include "md4c.h"
 
 #include <limits.h>
+#include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
 
@@ -71,6 +72,12 @@
     #define FALSE           0
 #endif
 
+#if defined(__has_attribute) && __has_attribute(unused)
+    #define MD_UNUSED __attribute__((unused))
+#else
+    #define MD_UNUSED
+#endif
+
 
 /************************
  ***  Internal Types  ***
@@ -465,7 +472,7 @@ md_text_with_null_replacement(MD_CTX* ctx, MD_TEXTTYPE type, const CHAR* str, SZ
 typedef struct MD_UNICODE_FOLD_INFO_tag MD_UNICODE_FOLD_INFO;
 struct MD_UNICODE_FOLD_INFO_tag {
     unsigned codepoints[3];
-    int n_codepoints;
+    unsigned n_codepoints;
 };
 
 
@@ -669,7 +676,7 @@ struct MD_UNICODE_FOLD_INFO_tag {
             const unsigned* map;
             const unsigned* data;
             size_t map_size;
-            int n_codepoints;
+            unsigned n_codepoints;
         } FOLD_MAP_LIST[] = {
             { FOLD_MAP_1, FOLD_MAP_1_DATA, SIZEOF_ARRAY(FOLD_MAP_1), 1 },
             { FOLD_MAP_2, FOLD_MAP_2_DATA, SIZEOF_ARRAY(FOLD_MAP_2), 2 },
@@ -689,12 +696,12 @@ struct MD_UNICODE_FOLD_INFO_tag {
 
         /* Try to locate the codepoint in any of the maps. */
         for(i = 0; i < (int) SIZEOF_ARRAY(FOLD_MAP_LIST); i++) {
-            int index;
+            unsigned index;
 
             index = md_unicode_bsearch__(codepoint, FOLD_MAP_LIST[i].map, FOLD_MAP_LIST[i].map_size);
             if(index >= 0) {
                 /* Found the mapping. */
-                int n_codepoints = FOLD_MAP_LIST[i].n_codepoints;
+                unsigned n_codepoints = FOLD_MAP_LIST[i].n_codepoints;
                 const unsigned* map = FOLD_MAP_LIST[i].map;
                 const unsigned* codepoints = FOLD_MAP_LIST[i].data + (index * n_codepoints);
 
@@ -886,7 +893,7 @@ struct MD_UNICODE_FOLD_INFO_tag {
  * what the caller should allocate.)
  */
 static void
-md_merge_lines(MD_CTX* ctx, OFF beg, OFF end, const MD_LINE* lines, int n_lines,
+md_merge_lines(MD_CTX* ctx, OFF beg, OFF end, const MD_LINE* lines,
                CHAR line_break_replacement_char, CHAR* buffer, SZ* p_size)
 {
     CHAR* ptr = buffer;
@@ -921,7 +928,7 @@ md_merge_lines(MD_CTX* ctx, OFF beg, OFF end, const MD_LINE* lines, int n_lines,
 /* Wrapper of md_merge_lines() which allocates new buffer for the output string.
  */
 static int
-md_merge_lines_alloc(MD_CTX* ctx, OFF beg, OFF end, const MD_LINE* lines, int n_lines,
+md_merge_lines_alloc(MD_CTX* ctx, OFF beg, OFF end, const MD_LINE* lines,
                     CHAR line_break_replacement_char, CHAR** p_str, SZ* p_size)
 {
     CHAR* buffer;
@@ -932,7 +939,7 @@ md_merge_lines_alloc(MD_CTX* ctx, OFF beg, OFF end, const MD_LINE* lines, int n_
         return -1;
     }
 
-    md_merge_lines(ctx, beg, end, lines, n_lines,
+    md_merge_lines(ctx, beg, end, lines,
                 line_break_replacement_char, buffer, p_size);
 
     *p_str = buffer;
@@ -1231,7 +1238,8 @@ md_is_html_any(MD_CTX* ctx, const MD_LINE* lines, int n_lines, OFF beg, OFF max_
  ****************************/
 
 static int
-md_is_hex_entity_contents(MD_CTX* ctx, const CHAR* text, OFF beg, OFF max_end, OFF* p_end)
+md_is_hex_entity_contents(MD_UNUSED MD_CTX* ctx, const CHAR* text,
+                          OFF beg, OFF max_end, OFF* p_end)
 {
     OFF off = beg;
 
@@ -1247,7 +1255,8 @@ md_is_hex_entity_contents(MD_CTX* ctx, const CHAR* text, OFF beg, OFF max_end, O
 }
 
 static int
-md_is_dec_entity_contents(MD_CTX* ctx, const CHAR* text, OFF beg, OFF max_end, OFF* p_end)
+md_is_dec_entity_contents(MD_UNUSED MD_CTX* ctx, const CHAR* text,
+                          OFF beg, OFF max_end, OFF* p_end)
 {
     OFF off = beg;
 
@@ -1263,7 +1272,8 @@ md_is_dec_entity_contents(MD_CTX* ctx, const CHAR* text, OFF beg, OFF max_end, O
 }
 
 static int
-md_is_named_entity_contents(MD_CTX* ctx, const CHAR* text, OFF beg, OFF max_end, OFF* p_end)
+md_is_named_entity_contents(MD_UNUSED MD_CTX* ctx, const CHAR* text,
+                            OFF beg, OFF max_end, OFF* p_end)
 {
     OFF off = beg;
 
@@ -1369,7 +1379,7 @@ md_build_attr_append_substr(MD_CTX* ctx, MD_ATTRIBUTE_BUILD* build,
 }
 
 static void
-md_free_attribute(MD_CTX* ctx, MD_ATTRIBUTE_BUILD* build)
+md_free_attribute(MD_UNUSED MD_CTX* ctx, MD_ATTRIBUTE_BUILD* build)
 {
     if(build->substr_alloc > 0) {
         free(build->text);
@@ -2158,7 +2168,7 @@ md_is_link_reference_definition(MD_CTX* ctx, const MD_LINE* lines, int n_lines)
 
     if(label_is_multiline) {
         MD_CHECK(md_merge_lines_alloc(ctx, label_contents_beg, label_contents_end,
-                    lines + label_contents_line_index, n_lines - label_contents_line_index,
+                    lines + label_contents_line_index,
                     _T(' '), &def->label, &def->label_size));
         def->label_needs_free = TRUE;
     } else {
@@ -2168,7 +2178,7 @@ md_is_link_reference_definition(MD_CTX* ctx, const MD_LINE* lines, int n_lines)
 
     if(title_is_multiline) {
         MD_CHECK(md_merge_lines_alloc(ctx, title_contents_beg, title_contents_end,
-                    lines + title_contents_line_index, n_lines - title_contents_line_index,
+                    lines + title_contents_line_index,
                     _T('\n'), &def->title, &def->title_size));
         def->title_needs_free = TRUE;
     } else {
@@ -2221,8 +2231,7 @@ md_is_link_reference(MD_CTX* ctx, const MD_LINE* lines, int n_lines,
         end_line++;
 
     if(beg_line != end_line) {
-        MD_CHECK(md_merge_lines_alloc(ctx, beg, end, beg_line,
-                 n_lines - (beg_line - lines), _T(' '), &label, &label_size));
+        MD_CHECK(md_merge_lines_alloc(ctx, beg, end, beg_line, _T(' '), &label, &label_size));
     } else {
         label = (CHAR*) STR(beg);
         label_size = end - beg;
@@ -2331,7 +2340,7 @@ md_is_inline_link_spec(MD_CTX* ctx, const MD_LINE* lines, int n_lines,
         attr->title_needs_free = FALSE;
     } else {
         MD_CHECK(md_merge_lines_alloc(ctx, title_contents_beg, title_contents_end,
-                    lines + title_contents_line_index, n_lines - title_contents_line_index,
+                    lines + title_contents_line_index,
                     _T('\n'), &attr->title, &attr->title_size));
         attr->title_needs_free = TRUE;
     }
@@ -3378,8 +3387,7 @@ md_analyze_bracket(MD_CTX* ctx, int mark_index)
 }
 
 /* Forward declaration. */
-static void md_analyze_link_contents(MD_CTX* ctx, const MD_LINE* lines, int n_lines,
-                                     int mark_beg, int mark_end);
+static void md_analyze_link_contents(MD_CTX* ctx, int mark_beg, int mark_end);
 
 static int
 md_resolve_links(MD_CTX* ctx, const MD_LINE* lines, int n_lines)
@@ -3498,7 +3506,7 @@ md_resolve_links(MD_CTX* ctx, const MD_LINE* lines, int n_lines)
                 if(delim != NULL) {
                     delim->flags |= MD_MARK_RESOLVED;
                     md_rollback(ctx, opener_index, delim_index, MD_ROLLBACK_ALL);
-                    md_analyze_link_contents(ctx, lines, n_lines, opener_index+1, closer_index);
+                    md_analyze_link_contents(ctx, opener_index+1, closer_index);
                 } else {
                     md_rollback(ctx, opener_index, closer_index, MD_ROLLBACK_ALL);
                 }
@@ -3523,6 +3531,10 @@ md_resolve_links(MD_CTX* ctx, const MD_LINE* lines, int n_lines)
             if(is_link) {
                 /* Eat the 2nd "[...]". */
                 closer->end = next_closer->end;
+
+                /* Do not analyze the label as a standalone link in the next
+                 * iteration. */
+                next_index = ctx->marks[next_index].prev;
             }
         } else {
             if(closer->end < ctx->size  &&  CH(closer->end) == _T('(')) {
@@ -3599,7 +3611,7 @@ md_resolve_links(MD_CTX* ctx, const MD_LINE* lines, int n_lines)
                 last_img_end = closer->end;
             }
 
-            md_analyze_link_contents(ctx, lines, n_lines, opener_index+1, closer_index);
+            md_analyze_link_contents(ctx, opener_index+1, closer_index);
         }
 
         opener_index = next_index;
@@ -3678,7 +3690,7 @@ md_analyze_emph(MD_CTX* ctx, int mark_index)
     /* If we can be a closer, try to resolve with the preceding opener. */
     if(mark->flags & MD_MARK_POTENTIAL_CLOSER) {
         MD_MARK* opener = NULL;
-        int opener_index;
+        int opener_index = 0;
 
         if(mark->ch == _T('*')) {
             MD_MARKCHAIN* opener_chains[6];
@@ -3913,8 +3925,7 @@ md_analyze_permissive_email_autolink(MD_CTX* ctx, int mark_index)
 }
 
 static inline void
-md_analyze_marks(MD_CTX* ctx, const MD_LINE* lines, int n_lines,
-                 int mark_beg, int mark_end, const CHAR* mark_chars)
+md_analyze_marks(MD_CTX* ctx, int mark_beg, int mark_end, const CHAR* mark_chars)
 {
     int i = mark_beg;
 
@@ -3972,10 +3983,10 @@ md_analyze_inlines(MD_CTX* ctx, const MD_LINE* lines, int n_lines, int table_mod
 
     /* We analyze marks in few groups to handle their precedence. */
     /* (1) Entities; code spans; autolinks; raw HTML. */
-    md_analyze_marks(ctx, lines, n_lines, 0, ctx->n_marks, _T("&"));
+    md_analyze_marks(ctx, 0, ctx->n_marks, _T("&"));
 
     /* (2) Links. */
-    md_analyze_marks(ctx, lines, n_lines, 0, ctx->n_marks, _T("[]!"));
+    md_analyze_marks(ctx, 0, ctx->n_marks, _T("[]!"));
     MD_CHECK(md_resolve_links(ctx, lines, n_lines));
     BRACKET_OPENERS.head = -1;
     BRACKET_OPENERS.tail = -1;
@@ -3990,24 +4001,23 @@ md_analyze_inlines(MD_CTX* ctx, const MD_LINE* lines, int n_lines, int table_mod
         TABLECELLBOUNDARIES.head = -1;
         TABLECELLBOUNDARIES.tail = -1;
         ctx->n_table_cell_boundaries = 0;
-        md_analyze_marks(ctx, lines, n_lines, 0, ctx->n_marks, _T("|"));
+        md_analyze_marks(ctx, 0, ctx->n_marks, _T("|"));
         return ret;
     }
 
     /* (4) Emphasis and strong emphasis; permissive autolinks. */
-    md_analyze_link_contents(ctx, lines, n_lines, 0, ctx->n_marks);
+    md_analyze_link_contents(ctx, 0, ctx->n_marks);
 
 abort:
     return ret;
 }
 
 static void
-md_analyze_link_contents(MD_CTX* ctx, const MD_LINE* lines, int n_lines,
-                         int mark_beg, int mark_end)
+md_analyze_link_contents(MD_CTX* ctx, int mark_beg, int mark_end)
 {
     int i;
 
-    md_analyze_marks(ctx, lines, n_lines, mark_beg, mark_end, _T("*_~$@:."));
+    md_analyze_marks(ctx, mark_beg, mark_end, _T("*_~$@:."));
 
     for(i = OPENERS_CHAIN_FIRST; i <= OPENERS_CHAIN_LAST; i++) {
         ctx->mark_chains[i].head = -1;
@@ -5654,7 +5664,7 @@ md_line_indentation(MD_CTX* ctx, unsigned total_indent, OFF beg, OFF* p_end)
     return indent - total_indent;
 }
 
-static const MD_LINE_ANALYSIS md_dummy_blank_line = { MD_LINE_BLANK, 0 };
+static const MD_LINE_ANALYSIS md_dummy_blank_line = { MD_LINE_BLANK, 0, 0, 0, 0 };
 
 /* Analyze type of the line and find some its properties. This serves as a
  * main input for determining type and boundaries of a block. */

src/wbuf.c

@@ -1,4 +1,4 @@
-#include "wbuf.h"
+#include "common.h"
 
 void WBufInit(WBuf* b) {
   b->start = 0;

src/wbuf.h

@@ -1,5 +1,4 @@
 #pragma once
-#include "common.h"
 
 typedef struct WBuf_s {
   char* start; // pointer to start of data

test/benchmark/bench.js

@@ -6,6 +6,7 @@ const Path = require('path')
 const commonmark = require('commonmark')
 const Showdown = require('showdown')
 const marked = require('marked')
+const { Remarkable } = require('remarkable');
 const markdownit = require('markdown-it')('commonmark')
 const markdown_wasm = require('../../dist/markdown.node.js')
 
@@ -22,6 +23,9 @@ var showdown = new Showdown.Converter()
 var parser = new commonmark.Parser()
 var renderer = new commonmark.HtmlRenderer()
 
+// setup remarkable
+var remarkable = new Remarkable();
+
 // parse CLI input
 let filename = process.argv[2]
 if (!filename) {
@@ -76,6 +80,9 @@ function benchmarkFile(benchfile) {
   .add('marked', function() {
     marked(contents);
   })
+  .add('remarkable', function() {
+    remarkable.render(contents);
+  })
   .add('markdown-it', function() {
     markdownit.render(contents);
   })

test/benchmark/package.json

@@ -18,6 +18,7 @@
     "d3-node": "^2.2.2",
     "markdown-it": "^10.0.0",
     "marked": "^0.7.0",
+    "remarkable": "^2.0.1",
     "showdown": "^1.9.1",
     "svgo": "^1.3.2"
   }

test/benchmark/results/avg-ops-per-sec.svg

@@ -1 +1 @@
-<svg xmlns="http://www.w3.org/2000/svg" width="840" height="264" font-family="-apple-system,BlinkMacSystemFont,&quot;Segoe UI&quot;,Helvetica,Arial,sans-serif" font-size="16"><g text-anchor="end" font-weight="500"><text alignment-baseline="central" x="156" y="68" dx="-10">markdown-wasm</text><text alignment-baseline="central" x="156" y="104" dx="-10">markdown-it</text><text alignment-baseline="central" x="156" y="140" dx="-10">marked</text><text alignment-baseline="central" x="156" y="176" dx="-10">commonmark</text><text alignment-baseline="central" x="156" y="212" dx="-10">showdown</text></g><path fill="#4e79a7" d="M156 52h668v32H156z"/><path fill="#f28e2c" d="M156 88h290.307v32H156z"/><path fill="#e15759" d="M156 124h263.486v32H156z"/><path fill="#76b7b2" d="M156 160h226.65v32H156z"/><path fill="#59a14f" d="M156 196h29.116v32H156z"/><g fill="#fff" text-anchor="end" font-weight="500"><text x="824" y="68" alignment-baseline="central" dx="-5">195,065</text><text x="446.307" y="104" alignment-baseline="central" dx="-5">84,774</text><text x="419.486" y="140" alignment-baseline="central" dx="-5">76,942</text><text x="382.65" y="176" alignment-baseline="central" dx="-5">66,185</text><text x="185.116" y="212" alignment-baseline="central" dx="5" fill="#000" text-anchor="start">8,502</text></g><g fill="none" font-size="12.8" text-anchor="middle" opacity=".5"><g class="tick"><path stroke="currentColor" d="M156.5 48v-6"/><text fill="currentColor" y="-9" dy="0em" transform="translate(156.5 48)">0</text></g><g class="tick"><path stroke="currentColor" d="M224.99 48v-6"/><text fill="currentColor" y="-9" dy="0em" transform="translate(224.99 48)">20,000</text></g><g class="tick"><path stroke="currentColor" d="M293.48 48v-6"/><text fill="currentColor" y="-9" dy="0em" transform="translate(293.48 48)">40,000</text></g><g class="tick"><path stroke="currentColor" d="M361.97 48v-6"/><text fill="currentColor" y="-9" dy="0em" transform="translate(361.97 48)">60,000</text></g><g class="tick"><path stroke="currentColor" d="M430.46 48v-6"/><text fill="currentColor" y="-9" dy="0em" transform="translate(430.46 48)">80,000</text></g><g class="tick"><path stroke="currentColor" d="M498.95 48v-6"/><text fill="currentColor" y="-9" dy="0em" transform="translate(498.95 48)">100,000</text></g><g class="tick"><path stroke="currentColor" d="M567.44 48v-6"/><text fill="currentColor" y="-9" dy="0em" transform="translate(567.44 48)">120,000</text></g><g class="tick"><path stroke="currentColor" d="M635.93 48v-6"/><text fill="currentColor" y="-9" dy="0em" transform="translate(635.93 48)">140,000</text></g><g class="tick"><path stroke="currentColor" d="M704.42 48v-6"/><text fill="currentColor" y="-9" dy="0em" transform="translate(704.42 48)">160,000</text></g><g class="tick"><path stroke="currentColor" d="M772.909 48v-6"/><text fill="currentColor" y="-9" dy="0em" transform="translate(772.909 48)">180,000</text></g></g></svg>
+<svg xmlns="http://www.w3.org/2000/svg" width="840" height="300" font-family="-apple-system,BlinkMacSystemFont,&quot;Segoe UI&quot;,Helvetica,Arial,sans-serif" font-size="16"><g text-anchor="end" font-weight="500"><text alignment-baseline="central" x="156" y="68" dx="-10">markdown-wasm</text><text alignment-baseline="central" x="156" y="104" dx="-10">remarkable</text><text alignment-baseline="central" x="156" y="140" dx="-10">markdown-it</text><text alignment-baseline="central" x="156" y="176" dx="-10">marked</text><text alignment-baseline="central" x="156" y="212" dx="-10">commonmark</text><text alignment-baseline="central" x="156" y="248" dx="-10">showdown</text></g><path fill="#4e79a7" d="M156 52h668v32H156z"/><path fill="#f28e2c" d="M156 88h453.666v32H156z"/><path fill="#e15759" d="M156 124h300.527v32H156z"/><path fill="#76b7b2" d="M156 160h259.363v32H156z"/><path fill="#59a14f" d="M156 196h210.906v32H156z"/><path fill="#edc949" d="M156 232h30.517v32H156z"/><g fill="#fff" text-anchor="end" font-weight="500"><text x="824" y="68" alignment-baseline="central" dx="-5">136,888</text><text x="609.666" y="104" alignment-baseline="central" dx="-5">92,966</text><text x="456.527" y="140" alignment-baseline="central" dx="-5">61,584</text><text x="415.363" y="176" alignment-baseline="central" dx="-5">53,149</text><text x="366.906" y="212" alignment-baseline="central" dx="-5">43,219</text><text x="186.517" y="248" alignment-baseline="central" dx="5" fill="#000" text-anchor="start">6,254</text></g><g fill="none" font-size="12.8" text-anchor="middle" opacity=".5"><g class="tick"><path stroke="currentColor" d="M156.5 48v-6"/><text fill="currentColor" y="-9" dy="0em" transform="translate(156.5 48)">0</text></g><g class="tick"><path stroke="currentColor" d="M254.098 48v-6"/><text fill="currentColor" y="-9" dy="0em" transform="translate(254.098 48)">20,000</text></g><g class="tick"><path stroke="currentColor" d="M351.697 48v-6"/><text fill="currentColor" y="-9" dy="0em" transform="translate(351.697 48)">40,000</text></g><g class="tick"><path stroke="currentColor" d="M449.295 48v-6"/><text fill="currentColor" y="-9" dy="0em" transform="translate(449.295 48)">60,000</text></g><g class="tick"><path stroke="currentColor" d="M546.893 48v-6"/><text fill="currentColor" y="-9" dy="0em" transform="translate(546.893 48)">80,000</text></g><g class="tick"><path stroke="currentColor" d="M644.491 48v-6"/><text fill="currentColor" y="-9" dy="0em" transform="translate(644.491 48)">100,000</text></g><g class="tick"><path stroke="currentColor" d="M742.09 48v-6"/><text fill="currentColor" y="-9" dy="0em" transform="translate(742.09 48)">120,000</text></g></g></svg>

test/benchmark/results/minmax-parse-time.svg

@@ -1 +1 @@
-<svg xmlns="http://www.w3.org/2000/svg" width="840" height="264" font-family="-apple-system,BlinkMacSystemFont,&quot;Segoe UI&quot;,Helvetica,Arial,sans-serif" font-size="16"><path fill="#4e79a7" d="M21 52h486v32H21z"/><path fill="#e15759" d="M16 88h576v32H16z"/><path fill="#f28e2c" d="M21 124h573v32H21z"/><path fill="#76b7b2" d="M70 160h541v32H70z"/><path fill="#59a14f" d="M227 196h597v32H227z"/><g fill="#fff" font-weight="500" text-anchor="middle"><text x="264" y="68" alignment-baseline="central">markdown-wasm</text><text x="304" y="104" alignment-baseline="central">marked</text><text x="307.5" y="140" alignment-baseline="central">markdown-it</text><text x="340.5" y="176" alignment-baseline="central">commonmark</text><text x="525.5" y="212" alignment-baseline="central">showdown</text></g><g fill="#fff" font-weight="500"><text x="21" y="68" alignment-baseline="central" dx="5">2.0us</text><text x="16" y="104" alignment-baseline="central" dx="5">1.8us</text><text x="21" y="140" alignment-baseline="central" dx="5">2.0us</text><text x="70" y="176" alignment-baseline="central" dx="5">4.0us</text><text x="227" y="212" alignment-baseline="central" dx="5">40.8us</text></g><g fill="#fff" font-weight="500" text-anchor="end"><text x="507" y="68" alignment-baseline="central" dx="-5">2.5ms</text><text x="592" y="104" alignment-baseline="central" dx="-5">8.6ms</text><text x="594" y="140" alignment-baseline="central" dx="-5">8.9ms</text><text x="611" y="176" alignment-baseline="central" dx="-5">11.4ms</text><text x="824" y="212" alignment-baseline="central" dx="-5">262.8ms</text></g><g fill="none" font-size="12.8" text-anchor="middle" opacity=".5"><path stroke="currentColor" d="M22.5 48v-6M50.5 48v-6M69.5 48v-6M85.5 48v-6M97.5 48v-6M108.5 48v-6M117.5 48v-6M125.5 48v-6"/><g class="tick"><path stroke="currentColor" d="M132.5 48v-6"/><text fill="currentColor" y="-9" dy="0em" transform="translate(132.5 48)">10.0us</text></g><path stroke="currentColor" d="M179.5 48v-6M207.5 48v-6M226.5 48v-6M241.5 48v-6M254.5 48v-6M264.5 48v-6M273.5 48v-6M281.5 48v-6"/><g class="tick"><path stroke="currentColor" d="M288.5 48v-6"/><text fill="currentColor" y="-9" dy="0em" transform="translate(288.5 48)">100.0us</text></g><path stroke="currentColor" d="M336.5 48v-6M363.5 48v-6M383.5 48v-6M398.5 48v-6M410.5 48v-6M421.5 48v-6M430.5 48v-6M438.5 48v-6"/><g class="tick"><path stroke="currentColor" d="M445.5 48v-6"/><text fill="currentColor" y="-9" dy="0em" transform="translate(445.5 48)">1000.0us</text></g><path stroke="currentColor" d="M492.5 48v-6M520.5 48v-6M539.5 48v-6M555.5 48v-6M567.5 48v-6M577.5 48v-6M586.5 48v-6M594.5 48v-6"/><g class="tick"><path stroke="currentColor" d="M602.5 48v-6"/><text fill="currentColor" y="-9" dy="0em" transform="translate(602.5 48)">10.0ms</text></g><path stroke="currentColor" d="M649.5 48v-6M676.5 48v-6M696.5 48v-6M711.5 48v-6M724.5 48v-6M734.5 48v-6M743.5 48v-6M751.5 48v-6"/><g class="tick"><path stroke="currentColor" d="M758.5 48v-6"/><text fill="currentColor" y="-9" dy="0em" transform="translate(758.5 48)">100.0ms</text></g><path stroke="currentColor" d="M805.5 48v-6"/></g></svg>
+<svg xmlns="http://www.w3.org/2000/svg" width="840" height="300" font-family="-apple-system,BlinkMacSystemFont,&quot;Segoe UI&quot;,Helvetica,Arial,sans-serif" font-size="16"><path fill="#4e79a7" d="M45 52h483v32H45z"/><path fill="#f28e2c" d="M16 88h582v32H16z"/><path fill="#76b7b2" d="M38 124h584v32H38z"/><path fill="#e15759" d="M47 160h578v32H47z"/><path fill="#59a14f" d="M96 196h553v32H96z"/><path fill="#edc949" d="M258 232h566v32H258z"/><g fill="#fff" font-weight="500" text-anchor="middle"><text x="286.5" y="68" alignment-baseline="central">markdown-wasm</text><text x="307" y="104" alignment-baseline="central">remarkable</text><text x="330" y="140" alignment-baseline="central">marked</text><text x="336" y="176" alignment-baseline="central">markdown-it</text><text x="372.5" y="212" alignment-baseline="central">commonmark</text><text x="541" y="248" alignment-baseline="central">showdown</text></g><g fill="#fff" font-weight="500"><text x="45" y="68" alignment-baseline="central" dx="5">2.7us</text><text x="16" y="104" alignment-baseline="central" dx="5">1.8us</text><text x="38" y="140" alignment-baseline="central" dx="5">2.5us</text><text x="47" y="176" alignment-baseline="central" dx="5">2.8us</text><text x="96" y="212" alignment-baseline="central" dx="5">5.7us</text><text x="258" y="248" alignment-baseline="central" dx="5">59.6us</text></g><g fill="#fff" font-weight="500" text-anchor="end"><text x="528" y="68" alignment-baseline="central" dx="-5">3.1ms</text><text x="598" y="104" alignment-baseline="central" dx="-5">8.4ms</text><text x="622" y="140" alignment-baseline="central" dx="-5">12.0ms</text><text x="625" y="176" alignment-baseline="central" dx="-5">12.5ms</text><text x="649" y="212" alignment-baseline="central" dx="-5">17.9ms</text><text x="824" y="248" alignment-baseline="central" dx="-5">226.4ms</text></g><g fill="none" font-size="12.8" text-anchor="middle" opacity=".5"><path stroke="currentColor" d="M24.5 48v-6M52.5 48v-6M72.5 48v-6M87.5 48v-6M100.5 48v-6M111.5 48v-6M120.5 48v-6M128.5 48v-6"/><g class="tick"><path stroke="currentColor" d="M135.5 48v-6"/><text fill="currentColor" y="-9" dy="0em" transform="translate(135.5 48)">10.0us</text></g><path stroke="currentColor" d="M183.5 48v-6M211.5 48v-6M230.5 48v-6M246.5 48v-6M258.5 48v-6M269.5 48v-6M278.5 48v-6M286.5 48v-6"/><g class="tick"><path stroke="currentColor" d="M293.5 48v-6"/><text fill="currentColor" y="-9" dy="0em" transform="translate(293.5 48)">100.0us</text></g><path stroke="currentColor" d="M341.5 48v-6M369.5 48v-6M388.5 48v-6M404.5 48v-6M416.5 48v-6M427.5 48v-6M436.5 48v-6M444.5 48v-6"/><g class="tick"><path stroke="currentColor" d="M451.5 48v-6"/><text fill="currentColor" y="-9" dy="0em" transform="translate(451.5 48)">1000.0us</text></g><path stroke="currentColor" d="M499.5 48v-6M527.5 48v-6M547.5 48v-6M562.5 48v-6M575.5 48v-6M585.5 48v-6M594.5 48v-6M602.5 48v-6"/><g class="tick"><path stroke="currentColor" d="M610.5 48v-6"/><text fill="currentColor" y="-9" dy="0em" transform="translate(610.5 48)">10.0ms</text></g><path stroke="currentColor" d="M657.5 48v-6M685.5 48v-6M705.5 48v-6M720.5 48v-6M733.5 48v-6M743.5 48v-6M753.5 48v-6M761.5 48v-6"/><g class="tick"><path stroke="currentColor" d="M768.5 48v-6"/><text fill="currentColor" y="-9" dy="0em" transform="translate(768.5 48)">100.0ms</text></g><path stroke="currentColor" d="M815.5 48v-6"/></g></svg>

wasmc.js

@@ -23,6 +23,10 @@ const m = {
   ] : [
     // release flags
   ]),
+  lflags: [
+    // force inclusion of addFunction & removeFunction in release builds (emcc bug?)
+    "-s","EXPORTED_RUNTIME_METHODS=addFunction,removeFunction",
+  ],
   constants: {
     VERSION: package.version,
   },