# frozen_string_literal: true

# WikiClient — fetches the Vodafone Wiki.js content via its GraphQL API for the RAG index.
# JWT from env (WIKI_JWT). Strips HTML to plain text. Excludes pure data-dump pages (inventory /
# location tables) that are noise for call summaries. Standard library only; transport injectable for tests.

require "json"
require "net/http"
require "uri"

module Helpdesk
  class WikiClient
    DEFAULT_BASE = "https://wikivf.max.support"
    # exclude data dumps: path keywords + an oversize guard (the standards pages ~145K chars stay;
    # the 511K PSU-inventory and 203K location-table pages go).
    EXCLUDE_PATHS = [/inventura/i, /invent/i, /lokalit/i].freeze
    MAX_PAGE_CHARS = 160_000

    def initialize(base_url: DEFAULT_BASE, jwt: ENV["WIKI_JWT"], transport: nil)
      @base = base_url.chomp("/")
      @jwt = jwt
      @transport = transport # ->(query_string) { response_body_json } — for tests
    end

    # → [{ id:, path:, title:, updated_at: }]
    def list
      d = gql("{ pages { list { id path title updatedAt } } }")
      d.dig("data", "pages", "list").map { |p| { id: p["id"], path: p["path"], title: p["title"], updated_at: p["updatedAt"] } }
    end

    # → plain text of one page (HTML stripped), or nil if forbidden/missing.
    def content(id)
      d = gql("{ pages { single(id:#{id}){ content } } }")
      html = d.dig("data", "pages", "single", "content")
      html && html_to_text(html)
    end

    # The knowledge pages to index: [{ title:, path:, text: }], excluding data dumps + empties.
    def knowledge_pages(exclude_paths: EXCLUDE_PATHS, max_chars: MAX_PAGE_CHARS)
      list.reject { |p| exclude_paths.any? { |re| p[:path] =~ re } }.filter_map do |p|
        text = content(p[:id])
        next if text.nil? || text.strip.empty? || text.length > max_chars
        { title: p[:title], path: p[:path], text: text }
      end
    end

    private

    def gql(query)
      raw = @transport ? @transport.call(query) : post(query)
      JSON.parse(raw)
    end

    def post(query)
      uri = URI("#{@base}/graphql")
      http = Net::HTTP.new(uri.host, uri.port)
      http.use_ssl = uri.scheme == "https"
      http.open_timeout = 15
      http.read_timeout = 60
      req = Net::HTTP::Post.new(uri)
      req["Authorization"] = "Bearer #{@jwt}"
      req["Content-Type"] = "application/json"
      req.body = JSON.generate(query: query)
      http.request(req).body
    end

    # minimal HTML → text: drop script/style, tags → spaces, decode a few entities, collapse whitespace.
    def html_to_text(html)
      t = html.dup
      t.gsub!(%r{<(script|style)[^>]*>.*?</\1>}mi, " ")
      t.gsub!(/<[^>]+>/, " ")
      { "&nbsp;" => " ", "&amp;" => "&", "&lt;" => "<", "&gt;" => ">", "&quot;" => '"', "&#39;" => "'" }
        .each { |k, v| t.gsub!(k, v) }
      t.gsub(/\s+/, " ").strip
    end
  end
end