// Markdown parser that renders to HTML and ANSI terminal output

// HTML rendering

function escape_html(text)
  text = replace(text, "&", "&")
  text = replace(text, "<", "&lt;")
  text = replace(text, ">", "&gt;")
  return text
end

function escape_markdown(text)
  // Escape characters that could be interpreted as markdown
  text = replace(text, ~\*~, "&#42;")
  text = replace(text, ~\[~, "&#91;")
  text = replace(text, ~\]~, "&#93;")
  text = replace(text, ~\(~, "&#40;")
  text = replace(text, ~\)~, "&#41;")
  return text
end

function addline(text)
  if len(text) > 0 and not contains(text, ~\n\s*\n\s*$~) then
    text = text + "\n"
  end

  return text
end

function parse_inline(text)
  // Process inline code first - escape HTML chars within code blocks
  // This prevents later regexes from matching HTML-like content in code
  text = replace(text, ~`([^`\n]+)`~, function(code_content)
    // Strip backticks from start and end (matched includes them)
    content = substr(code_content, 1, len(code_content) - 2)
    // Escape both HTML and markdown characters to prevent further processing
    escaped = escape_html(escape_markdown(content))
    return "<code>" + escaped + "</code>"
  end, true)

  // Process links and images (to protect them from emphasis parsing)
  text = replace(text, ~\!\[([^\]]*)\]\(([^)]+)\)~, '<img src="$2" alt="$1">')
  text = replace(text, ~\[([^\]]+)\]\(([^)]+)\)~, '<a href="$2">$1</a>')

  // Bold first (before italic to avoid **text** being parsed as *text*)
  text = replace(text, ~\*\*([^\*]+)\*\*~, "<strong>$1</strong>")
  // Italic with asterisk
  text = replace(text, ~\*([^\*]+)\*~, "<em>$1</em>")
  // Italic with underscore - only match when surrounded by spaces or punctuation, not word chars
  text = replace(text, ~(^|\s)_([^_]+)_(\s|$|[^\w])~, "$1<em>$2</em>$3")

  // Unescape escaped characters (remove backslash before special chars)
  text = replace(text, ~\\([\`\*_\[\]\\])~, "$1", true)

  // Unescape markdown characters that were escaped in code blocks
  // Note: must unescape HTML ampersands first since markdown escapes contain &
  text = replace(text, "&amp;#", "&#")
  text = replace(text, "&#91;", "[")
  text = replace(text, "&#93;", "]")
  text = replace(text, "&#40;", "(")
  text = replace(text, "&#41;", ")")

  return text
end

function strip_heading_markers(line)
  if contains(line, ~^###### ~) then
    return substr(line, 7)
  elseif contains(line, ~^##### ~) then
    return substr(line, 6)
  elseif contains(line, ~^#### ~) then
    return substr(line, 5)
  elseif contains(line, ~^### ~) then
    return substr(line, 4)
  elseif contains(line, ~^## ~) then
    return substr(line, 3)
  elseif contains(line, ~^# ~) then
    return substr(line, 2)
  end
  return line
end

function get_heading_level(line)
  if contains(line, ~^###### ~) then
    return 6
  elseif contains(line, ~^##### ~) then
    return 5
  elseif contains(line, ~^#### ~) then
    return 4
  elseif contains(line, ~^### ~) then
    return 3
  elseif contains(line, ~^## ~) then
    return 2
  elseif contains(line, ~^# ~) then
    return 1
  end
  return 0
end

function html(markdown)
  var lines = split(markdown, "\n")
  var output = ""
  var in_code_block = false
  var code_content = ""
  var code_language = "text"
  var in_list = false
  var in_paragraph = false
  var paragraph_content = ""
  var i = 0

  while i < len(lines) do
    var line = lines[i]

    // Handle code blocks
    if contains(line, "```") then
      if in_code_block then
        // Closing code block - unescape escaped chars then escape HTML
        var unescaped = replace(code_content, ~\\([\`\*_\[\]\\])~, "$1", true)
        output = output + "<pre><code class=\"language-" + code_language + "\">" + escape_html(unescaped) + "</code></pre>"
        code_content = ""
        code_language = "text"
        in_code_block = false
      else
        // Opening code block - extract language from fence
        var fence_match = replace(line, ~^```\s*(\w+)?.*$~, "$1", true)
        code_language = fence_match != line ? fence_match : "text"
        if code_language == "" then
          code_language = "text"
        end
        in_code_block = true
      end

    elseif in_code_block then
      // Inside code block - collect content
      code_content = code_content + line + "\n"

    // Handle horizontal rules (---, ***, ___)
    elseif contains(line, ~^\s*---\s*$~) or contains(line, ~^\s*\*\*\*\s*$~) or contains(line, ~^\s*___\s*$~) then
      if in_list then
        output = output + "</ul>"
        in_list = false
      end
      output = output + "<hr>"

    // Handle headings
    elseif contains(line, ~^######\s|^#####\s|^####\s|^###\s|^##\s|^#\s~) then
      if in_list then
        output = output + "</ul>"
        in_list = false
      end
      var level = get_heading_level(line)
      var content = strip_heading_markers(line)
      var formatted = parse_inline(content)
      output = output + "<h" + level + ">" + formatted + "</h" + level + ">"

    // Handle blockquotes (> and >>)
    elseif contains(line, ~^>>~) or contains(line, ~^>~) then
      if in_list then
        output = output + "</ul>"
        in_list = false
      end
      // Count the > symbols for nesting level
      var quote_level = 0
      for j = 0, len(line) - 1 do
        if substr(line, j, 1) == ">" then
          quote_level = quote_level + 1
        else
          break
        end
      end
      // Remove the > symbols and leading space
      var content = trim(substr(line, quote_level + 1))
      var formatted = parse_inline(content)
      // Wrap in blockquotes based on level
      for k = 1, quote_level do
        formatted = "<blockquote>" + formatted + "</blockquote>"
      end
      output = output + formatted

    // Handle list items
    elseif contains(line, ~^[-*]\s~) then
      if not in_list then
        output = output + "<ul>"
        in_list = true
      end
      var item_text = trim(substr(line, 2))
      var formatted = parse_inline(item_text)
      output = output + "<li>" + formatted + "</li>"

    // Handle paragraphs - accumulate consecutive lines
    elseif len(trim(line)) > 0 then
      if in_list then
        output = output + "</ul>"
        in_list = false
      end
      if in_paragraph then
        paragraph_content = paragraph_content + " " + line
      else
        in_paragraph = true
        paragraph_content = line
      end

    // Handle blank lines - close paragraph if open
    else
      if in_paragraph then
        output = output + "<p>" + parse_inline(paragraph_content) + "</p>"
        in_paragraph = false
        paragraph_content = ""
      end
      if in_list then
        output = output + "</ul>"
        in_list = false
      end

    end

    i = i + 1
  end

  // Close any open paragraph or list at end
  if in_paragraph then
    output = output + "<p>" + parse_inline(paragraph_content) + "</p>"
  end
  if in_list then
    output = output + "</ul>"
  end

  return output
end

// ANSI rendering

function ansi(markdown, theme)
  if theme == nil then
    theme = default_ansi_theme()
  end

  var lines = split(markdown, "\n")
  var output = ""
  var in_code_block = false
  var code_content = ""
  var i = 0

  while i < len(lines) do
    var line = lines[i]

    // Handle code blocks
    if contains(line, "```") then
      if in_code_block then
        // Closing code block - render with code styling
        output = output + theme.code_start + code_content + theme.reset
        code_content = ""
        in_code_block = false
      else
        // Opening code block
        in_code_block = true
      end

    elseif in_code_block then
      // Inside code block - collect content (with trailing newline for each line)
      code_content = code_content + "  " + line + "\n"

    // Handle horizontal rules
    elseif contains(line, ~^\s*---\s*$~) or contains(line, ~^\s*\*\*\*\s*$~) or contains(line, ~^\s*___\s*$~) then
      output = output + theme.hr + "\n"

    // Handle headings
    elseif contains(line, ~^######\s|^#####\s|^####\s|^###\s|^##\s|^#\s~) then
      var level = get_heading_level(line)
      var content = strip_heading_markers(line)
      var formatted = parse_inline_ansi(content, theme)
      var heading_start = theme["h" + level]
      output = output + heading_start + formatted + theme.reset + "\n"

    // Handle blockquotes
    elseif contains(line, ~^>>~) or contains(line, ~^>~) then
      var quote_level = 0
      for j = 0, len(line) - 1 do
        if substr(line, j, 1) == ">" then
          quote_level = quote_level + 1
        else
          break
        end
      end
      var content = trim(substr(line, quote_level + 1))
      var formatted = parse_inline_ansi(content, theme)
      output = output + theme.blockquote + formatted + theme.reset + "\n"

    // Handle list items (preserve source indentation, replace bullets)
    elseif contains(line, ~^\s*[-*]\s~) then
      // Find the position of the dash or asterisk
      var bullet_pos = 0
      for j = 0, len(line) - 1 do
        if substr(line, j, 1) == "-" or substr(line, j, 1) == "*" then
          bullet_pos = j
          break
        end
      end

      // Extract leading whitespace
      var leading_spaces = substr(line, 0, bullet_pos)

      // Extract item text (skip bullet and following space)
      var item_start = bullet_pos + 2
      var item_text = trim(substr(line, item_start))
      var formatted = parse_inline_ansi(item_text, theme)

      output = output + leading_spaces + theme.list_item + "• " + formatted + theme.reset + "\n"

    // Handle paragraphs
    elseif len(trim(line)) > 0 then
      var formatted = parse_inline_ansi(line, theme)
      output = output + formatted + "\n"

    // Handle blank lines - preserve spacing from source
    else
      if len(output) > 0 then
        output = output + "\n"
      end
    end

    i = i + 1
  end

  return output
end

function parse_inline_ansi(text, theme)
  // Handle links first (before code/bold/italic to avoid conflicts)
  // Matches [text](url) and applies color while keeping markdown syntax
  text = replace(text, ~\[([^\]]+)\]\(([^)]+)\)~, function(match)
    return theme.link + match + theme.reset
  end, true)

  // Handle inline code
  text = replace(text, ~`([^`\n]+)`~, function(code_content)
    // Strip backticks from start and end (0-based indexing)
    content = substr(code_content, 1, len(code_content) - 2)
    return theme.code_inline + content + theme.reset
  end, true)

  // Handle bold (before italic to avoid conflicts)
  text = replace(text, ~\*\*([^\*]+)\*\*~, function(match)
    // Strip ** from both sides: start at position 2, length len-4 (0-based)
    content = substr(match, 2, len(match) - 4)
    return theme.bold + content + theme.reset
  end, true)

  // Handle italic with asterisk (single *)
  text = replace(text, ~\*([^\*]+)\*~, function(match)
    // Strip * from both sides: start at position 1, length len-2 (0-based)
    content = substr(match, 1, len(match) - 2)
    return theme.italic + content + theme.reset
  end, true)

  return text
end

function default_ansi_theme()
  // Return colorless theme if -no-color flag is set
  if sys("-no-color") then
    return default_text_theme()
  end

  ansi = require("ansi")
  // Build raw ANSI codes that can be combined
  return {
    h1 = ansi.combine(fg="yellow", bold=true, underline=true, bright=true),
    h2 = ansi.combine(fg="white", bold=true, underline=true, bright=true),
    h3 = ansi.combine(fg="white", underline=true),
    h4 = ansi.combine(fg="white", underline=true),
    h5 = ansi.combine(fg="white", underline=true),
    h6 = ansi.combine(fg="white", underline=true),
    code_start = ansi.combine(fg="green", bright=true),
    code_inline = ansi.combine(fg="green"),
    blockquote = ansi.combine(fg="gray"),
    list_item = "",
    bold = ansi.combine(bold=true, bright=true),
    italic = ansi.combine(italic=true),
    link = ansi.combine(fg="blue", bright=true, bold=true, underline=true),
    hr = ansi.combine(fg="gray"),
    reset = ansi.clear
  }
end

function default_text_theme()
  // Colorless theme - all empty strings, no formatting codes
  return {
    h1 = "",
    h2 = "",
    h3 = "",
    h4 = "",
    h5 = "",
    h6 = "",
    code_start = "",
    code_inline = "",
    blockquote = "",
    list_item = "",
    bold = "",
    italic = "",
    link = "",
    hr = "",
    reset = ""
  }
end

function text(markdown)
  // Render markdown to plain text without ANSI color codes
  return ansi(markdown, default_text_theme())
end

return {
  html = html,
  ansi = ansi,
  text = text
}