showdown.js

//
// showdown.js -- A javascript port of Markdown.
//
// Copyright (c) 2007 John Fraser.
//
// Original Markdown Copyright (c) 2004-2005 John Gruber
//   <http://daringfireball.net/projects/markdown/>
//
// Redistributable under a BSD-style open source license.
// See license.txt for more information.
//
// The full source distribution is at:
//
//				A A L
//				T C A
//				T K B
//
//   <http://www.attacklab.net/>
//

//
// Wherever possible, Showdown is a straight, line-by-line port
// of the Perl version of Markdown.
//
// This is not a normal parser design; it's basically just a
// series of string substitutions.  It's hard to read and
// maintain this way,  but keeping Showdown close to the original
// design makes it easier to port new features.
//
// More importantly, Showdown behaves like markdown.pl in most
// edge cases.  So web applications can do client-side preview
// in Javascript, and then build identical HTML on the server.
//
// This port needs the new RegExp functionality of ECMA 262,
// 3rd Edition (i.e. Javascript 1.5).  Most modern web browsers
// should do fine.  Even with the new regular expression features,
// We do a lot of work to emulate Perl's regex functionality.
// The tricky changes in this file mostly have the "attacklab:"
// label.  Major or self-explanatory changes don't.
//
// Smart diff tools like Araxis Merge will be able to match up
// this file with markdown.pl in a useful way.  A little tweaking
// helps: in a copy of markdown.pl, replace "#" with "//" and
// replace "$text" with "text".  Be sure to ignore whitespace
// and line endings.
//


//
// Showdown usage:
//
//   var text = "Markdown *rocks*.";
//
//   var converter = new Showdown.converter();
//   var html = converter.makeHtml(text);
//
//   alert(html);
//
// Note: move the sample code to the bottom of this
// file before uncommenting it.
//


//
// Showdown namespace
//
var Showdown = {};

//
// converter
//
// Wraps all "globals" so that the only thing
// exposed is makeHtml().
//
Showdown.converter = function() {

//
// Globals:
//

// Global hashes, used by various utility routines
var g_urls;
var g_titles;
var g_html_blocks;

// Used to track when we're inside an ordered or unordered list
// (see _ProcessListItems() for details):
var g_list_level = 0;


this.makeHtml = function(text) {
//
// Main function. The order in which other subs are called here is
// essential. Link and image substitutions need to happen before
// _EscapeSpecialCharsWithinTagAttributes(), so that any *'s or _'s in the <a>
// and <img> tags get encoded.
//

	// Clear the global hashes. If we don't clear these, you get conflicts
	// from other articles when generating a page which contains more than
	// one article (e.g. an index page that shows the N most recent
	// articles):
	g_urls = new Array();
	g_titles = new Array();
	g_html_blocks = new Array();

	// attacklab: Replace ~ with ~T
	// This lets us use tilde as an escape char to avoid md5 hashes
	// The choice of character is arbitray; anything that isn't
    // magic in Markdown will work.
	text = text.replace(/~/g,"~T");

	// attacklab: Replace $ with ~D
	// RegExp interprets $ as a special character
	// when it's in a replacement string
	text = text.replace(/\$/g,"~D");

	// Standardize line endings
	text = text.replace(/\r\n/g,"\n"); // DOS to Unix
	text = text.replace(/\r/g,"\n"); // Mac to Unix

	// Make sure text begins and ends with a couple of newlines:
	text = "\n\n" + text + "\n\n";

	// Convert all tabs to spaces.
	text = _Detab(text);

	// Strip any lines consisting only of spaces and tabs.
	// This makes subsequent regexen easier to write, because we can
	// match consecutive blank lines with /\n+/ instead of something
	// contorted like /[ \t]*\n+/ .
	text = text.replace(/^[ \t]+$/mg,"");

	// Handle github codeblocks prior to running HashHTML so that
	// HTML contained within the codeblock gets escaped propertly
	text = _DoGithubCodeBlocks(text);

	// Turn block-level HTML blocks into hash entries
	text = _HashHTMLBlocks(text);

	// Strip link definitions, store in hashes.
	text = _StripLinkDefinitions(text);

	text = _RunBlockGamut(text);

	text = _UnescapeSpecialChars(text);

	// attacklab: Restore dollar signs
	text = text.replace(/~D/g,"$$");

	// attacklab: Restore tildes
	text = text.replace(/~T/g,"~");

	return text;
};


var _StripLinkDefinitions = function(text) {
//
// Strips link definitions from text, stores the URLs and titles in
// hash references.
//

	// Link defs are in the form: ^[id]: url "optional title"

	/*
		var text = text.replace(/
				^[ ]{0,3}\[(.+)\]:  // id = $1  attacklab: g_tab_width - 1
				  [ \t]*
				  \n?				// maybe *one* newline
				  [ \t]*
				<?(\S+?)>?			// url = $2
				  [ \t]*
				  \n?				// maybe one newline
				  [ \t]*
				(?:
				  (\n*)				// any lines skipped = $3 attacklab: lookbehind removed
				  ["(]
				  (.+?)				// title = $4
				  [")]
				  [ \t]*
				)?					// title is optional
				(?:\n+|$)
			  /gm,
			  function(){...});
	*/
	var text = text.replace(/^[ ]{0,3}\[(.+)\]:[ \t]*\n?[ \t]*<?(\S+?)>?[ \t]*\n?[ \t]*(?:(\n*)["(](.+?)[")][ \t]*)?(?:\n+|\Z)/gm,
		function (wholeMatch,m1,m2,m3,m4) {
			m1 = m1.toLowerCase();
			g_urls[m1] = _EncodeAmpsAndAngles(m2);  // Link IDs are case-insensitive
			if (m3) {
				// Oops, found blank lines, so it's not a title.
				// Put back the parenthetical statement we stole.
				return m3+m4;
			} else if (m4) {
				g_titles[m1] = m4.replace(/"/g,"&quot;");
			}

			// Completely remove the definition from the text
			return "";
		}
	);

	return text;
}


var _HashHTMLBlocks = function(text) {
	// attacklab: Double up blank lines to reduce lookaround
	text = text.replace(/\n/g,"\n\n");

	// Hashify HTML blocks:
	// We only want to do this for block-level HTML tags, such as headers,
	// lists, and tables. That's because we still want to wrap <p>s around
	// "paragraphs" that are wrapped in non-block-level tags, such as anchors,
	// phrase emphasis, and spans. The list of tags we're looking for is
	// hard-coded:
	var block_tags_a = "p|div|h[1-6]|blockquote|pre|table|dl|ol|ul|script|noscript|form|fieldset|iframe|math|ins|del|style|section|header|footer|nav|article|aside";
	var block_tags_b = "p|div|h[1-6]|blockquote|pre|table|dl|ol|ul|script|noscript|form|fieldset|iframe|math|style|section|header|footer|nav|article|aside";

	// First, look for nested blocks, e.g.:
	//   <div>
	//     <div>
	//     tags for inner block must be indented.
	//     </div>
	//   </div>
	//
	// The outermost tags must start at the left margin for this to match, and
	// the inner nested divs must be indented.
	// We need to do this before the next, more liberal match, because the next
	// match will start at the first `<div>` and stop at the first `</div>`.

	// attacklab: This regex can be expensive when it fails.
	/*
		var text = text.replace(/
		(						// save in $1
			^					// start of line  (with /m)
			<($block_tags_a)	// start tag = $2
			\b					// word break
								// attacklab: hack around khtml/pcre bug...
			[^\r]*?\n			// any number of lines, minimally matching
			</\2>				// the matching end tag
			[ \t]*				// trailing spaces/tabs
			(?=\n+)				// followed by a newline
		)						// attacklab: there are sentinel newlines at end of document
		/gm,function(){...}};
	*/
	text = text.replace(/^(<(p|div|h[1-6]|blockquote|pre|table|dl|ol|ul|script|noscript|form|fieldset|iframe|math|ins|del)\b[^\r]*?\n<\/\2>[ \t]*(?=\n+))/gm,hashElement);

	//
	// Now match more liberally, simply from `\n<tag>` to `</tag>\n`
	//

	/*
		var text = text.replace(/
		(						// save in $1
			^					// start of line  (with /m)
			<($block_tags_b)	// start tag = $2
			\b					// word break
								// attacklab: hack around khtml/pcre bug...
			[^\r]*?				// any number of lines, minimally matching
			.*</\2>				// the matching end tag
			[ \t]*				// trailing spaces/tabs
			(?=\n+)				// followed by a newline
		)						// attacklab: there are sentinel newlines at end of document
		/gm,function(){...}};
	*/
	text = text.replace(/^(<(p|div|h[1-6]|blockquote|pre|table|dl|ol|ul|script|noscript|form|fieldset|iframe|math|style|section|header|footer|nav|article|aside)\b[^\r]*?.*<\/\2>[ \t]*(?=\n+)\n)/gm,hashElement);

	// Special case just for <hr />. It was easier to make a special case than
	// to make the other regex more complicated.

	/*
		text = text.replace(/
		(						// save in $1
			\n\n				// Starting after a blank line
			[ ]{0,3}
			(<(hr)				// start tag = $2
			\b					// word break
			([^<>])*?			//
			\/?>)				// the matching end tag
			[ \t]*
			(?=\n{2,})			// followed by a blank line
		)
		/g,hashElement);
	*/
	text = text.replace(/(\n[ ]{0,3}(<(hr)\b([^<>])*?\/?>)[ \t]*(?=\n{2,}))/g,hashElement);

	// Special case for standalone HTML comments:

	/*
		text = text.replace(/
		(						// save in $1
			\n\n				// Starting after a blank line
			[ ]{0,3}			// attacklab: g_tab_width - 1
			<!
			(--[^\r]*?--\s*)+
			>
			[ \t]*
			(?=\n{2,})			// followed by a blank line
		)
		/g,hashElement);
	*/
	text = text.replace(/(\n\n[ ]{0,3}<!(--[^\r]*?--\s*)+>[ \t]*(?=\n{2,}))/g,hashElement);

	// PHP and ASP-style processor instructions (<?...?> and <%...%>)

	/*
		text = text.replace(/
		(?:
			\n\n				// Starting after a blank line
		)
		(						// save in $1
			[ ]{0,3}			// attacklab: g_tab_width - 1
			(?:
				<([?%])			// $2
				[^\r]*?
				\2>
			)
			[ \t]*
			(?=\n{2,})			// followed by a blank line
		)
		/g,hashElement);
	*/
	text = text.replace(/(?:\n\n)([ ]{0,3}(?:<([?%])[^\r]*?\2>)[ \t]*(?=\n{2,}))/g,hashElement);

	// attacklab: Undo double lines (see comment at top of this function)
	text = text.replace(/\n\n/g,"\n");
	return text;
}

var hashElement = function(wholeMatch,m1) {
	var blockText = m1;

	// Undo double lines
	blockText = blockText.replace(/\n\n/g,"\n");
	blockText = blockText.replace(/^\n/,"");

	// strip trailing blank lines
	blockText = blockText.replace(/\n+$/g,"");

	// Replace the element text with a marker ("~KxK" where x is its key)
	blockText = "\n\n~K" + (g_html_blocks.push(blockText)-1) + "K\n\n";

	return blockText;
};

var _RunBlockGamut = function(text) {
//
// These are all the transformations that form block-level
// tags like paragraphs, headers, and list items.
//
	text = _DoHeaders(text);

	// Do Horizontal Rules:
	var key = hashBlock("<hr />");
	text = text.replace(/^[ ]{0,2}([ ]?\*[ ]?){3,}[ \t]*$/gm,key);
	text = text.replace(/^[ ]{0,2}([ ]?\-[ ]?){3,}[ \t]*$/gm,key);
	text = text.replace(/^[ ]{0,2}([ ]?\_[ ]?){3,}[ \t]*$/gm,key);

	text = _DoLists(text);
	text = _DoCodeBlocks(text);
	text = _DoBlockQuotes(text);

	// We already ran _HashHTMLBlocks() before, in Markdown(), but that
	// was to escape raw HTML in the original Markdown source. This time,
	// we're escaping the markup we've just created, so that we don't wrap
	// <p> tags around block-level tags.
	text = _HashHTMLBlocks(text);
	text = _FormParagraphs(text);

	return text;
};


var _RunSpanGamut = function(text) {
//
// These are all the transformations that occur *within* block-level
// tags like paragraphs, headers, and list items.
//

	text = _DoCodeSpans(text);
	text = _EscapeSpecialCharsWithinTagAttributes(text);
	text = _EncodeBackslashEscapes(text);

	// Process anchor and image tags. Images must come first,
	// because ![foo][f] looks like an anchor.
	text = _DoImages(text);
	text = _DoAnchors(text);

	// Make links out of things like `<http://example.com/>`
	// Must come after _DoAnchors(), because you can use < and >
	// delimiters in inline links like [this](<url>).
	text = _DoAutoLinks(text);
	text = _EncodeAmpsAndAngles(text);
	text = _DoItalicsAndBold(text);

	// Do hard breaks:
	text = text.replace(/  +\n/g," <br />\n");

	return text;
}

var _EscapeSpecialCharsWithinTagAttributes = function(text) {
//
// Within tags -- meaning between < and > -- encode [\ ` * _] so they
// don't conflict with their use in Markdown for code, italics and strong.
//

	// Build a regex to find HTML tags and comments.  See Friedl's
	// "Mastering Regular Expressions", 2nd Ed., pp. 200-201.
	var regex = /(<[a-z\/!$]("[^"]*"|'[^']*'|[^'">])*>|<!(--.*?--\s*)+>)/gi;

	text = text.replace(regex, function(wholeMatch) {
		var tag = wholeMatch.replace(/(.)<\/?code>(?=.)/g,"$1`");
		tag = escapeCharacters(tag,"\\`*_");
		return tag;
	});

	return text;
}

var _DoAnchors = function(text) {
//
// Turn Markdown link shortcuts into XHTML <a> tags.
//
	//
	// First, handle reference-style links: [link text] [id]
	//

	/*
		text = text.replace(/
		(							// wrap whole match in $1
			\[
			(
				(?:
					\[[^\]]*\]		// allow brackets nested one level
					|
					[^\[]			// or anything else
				)*
			)
			\]

			[ ]?					// one optional space
			(?:\n[ ]*)?				// one optional newline followed by spaces

			\[
			(.*?)					// id = $3
			\]
		)()()()()					// pad remaining backreferences
		/g,_DoAnchors_callback);
	*/
	text = text.replace(/(\[((?:\[[^\]]*\]|[^\[\]])*)\][ ]?(?:\n[ ]*)?\[(.*?)\])()()()()/g,writeAnchorTag);

	//
	// Next, inline-style links: [link text](url "optional title")
	//

	/*
		text = text.replace(/
			(						// wrap whole match in $1
				\[
				(
					(?:
						\[[^\]]*\]	// allow brackets nested one level
					|
					[^\[\]]			// or anything else
				)
			)
			\]
			\(						// literal paren
			[ \t]*
			()						// no id, so leave $3 empty
			<?(.*?)>?				// href = $4
			[ \t]*
			(						// $5
				(['"])				// quote char = $6
				(.*?)				// Title = $7
				\6					// matching quote
				[ \t]*				// ignore any spaces/tabs between closing quote and )
			)?						// title is optional
			\)
		)
		/g,writeAnchorTag);
	*/
	text = text.replace(/(\[((?:\[[^\]]*\]|[^\[\]])*)\]\([ \t]*()<?(.*?)>?[ \t]*((['"])(.*?)\6[ \t]*)?\))/g,writeAnchorTag);

	//
	// Last, handle reference-style shortcuts: [link text]
	// These must come last in case you've also got [link test][1]
	// or [link test](/foo)
	//

	/*
		text = text.replace(/
		(		 					// wrap whole match in $1
			\[
			([^\[\]]+)				// link text = $2; can't contain '[' or ']'
			\]
		)()()()()()					// pad rest of backreferences
		/g, writeAnchorTag);
	*/
	text = text.replace(/(\[([^\[\]]+)\])()()()()()/g, writeAnchorTag);

	return text;
}

var writeAnchorTag = function(wholeMatch,m1,m2,m3,m4,m5,m6,m7) {
	if (m7 == undefined) m7 = "";
	var whole_match = m1;
	var link_text   = m2;
	var link_id	 = m3.toLowerCase();
	var url		= m4;
	var title	= m7;

	if (url == "") {
		if (link_id == "") {
			// lower-case and turn embedded newlines into spaces
			link_id = link_text.toLowerCase().replace(/ ?\n/g," ");
		}
		url = "#"+link_id;

		if (g_urls[link_id] != undefined) {
			url = g_urls[link_id];
			if (g_titles[link_id] != undefined) {
				title = g_titles[link_id];
			}
		}
		else {
			if (whole_match.search(/\(\s*\)$/m)>-1) {
				// Special case for explicit empty url
				url = "";
			} else {
				return whole_match;
			}
		}
	}

	url = escapeCharacters(url,"*_");
	var result = "<a href=\"" + url + "\"";

	if (title != "") {
		title = title.replace(/"/g,"&quot;");
		title = escapeCharacters(title,"*_");
		result +=  " title=\"" + title + "\"";
	}

	result += ">" + link_text + "</a>";

	return result;
}


var _DoImages = function(text) {
//
// Turn Markdown image shortcuts into <img> tags.
//

	//
	// First, handle reference-style labeled images: ![alt text][id]
	//

	/*
		text = text.replace(/
		(						// wrap whole match in $1
			!\[
			(.*?)				// alt text = $2
			\]

			[ ]?				// one optional space
			(?:\n[ ]*)?			// one optional newline followed by spaces

			\[
			(.*?)				// id = $3
			\]
		)()()()()				// pad rest of backreferences
		/g,writeImageTag);
	*/
	text = text.replace(/(!\[(.*?)\][ ]?(?:\n[ ]*)?\[(.*?)\])()()()()/g,writeImageTag);

	//
	// Next, handle inline images:  ![alt text](url "optional title")
	// Don't forget: encode * and _

	/*
		text = text.replace(/
		(						// wrap whole match in $1
			!\[
			(.*?)				// alt text = $2
			\]
			\s?					// One optional whitespace character
			\(					// literal paren
			[ \t]*
			()					// no id, so leave $3 empty
			<?(\S+?)>?			// src url = $4
			[ \t]*
			(					// $5
				(['"])			// quote char = $6
				(.*?)			// title = $7
				\6				// matching quote
				[ \t]*
			)?					// title is optional
		\)
		)
		/g,writeImageTag);
	*/
	text = text.replace(/(!\[(.*?)\]\s?\([ \t]*()<?(\S+?)>?[ \t]*((['"])(.*?)\6[ \t]*)?\))/g,writeImageTag);

	return text;
}

var writeImageTag = function(wholeMatch,m1,m2,m3,m4,m5,m6,m7) {
	var whole_match = m1;
	var alt_text   = m2;
	var link_id	 = m3.toLowerCase();
	var url		= m4;
	var title	= m7;

	if (!title) title = "";

	if (url == "") {
		if (link_id == "") {
			// lower-case and turn embedded newlines into spaces
			link_id = alt_text.toLowerCase().replace(/ ?\n/g," ");
		}
		url = "#"+link_id;

		if (g_urls[link_id] != undefined) {
			url = g_urls[link_id];
			if (g_titles[link_id] != undefined) {
				title = g_titles[link_id];
			}
		}
		else {
			return whole_match;
		}
	}

	alt_text = alt_text.replace(/"/g,"&quot;");
	url = escapeCharacters(url,"*_");
	var result = "<img src=\"" + url + "\" alt=\"" + alt_text + "\"";

	// attacklab: Markdown.pl adds empty title attributes to images.
	// Replicate this bug.

	//if (title != "") {
		title = title.replace(/"/g,"&quot;");
		title = escapeCharacters(title,"*_");
		result +=  " title=\"" + title + "\"";
	//}

	result += " />";

	return result;
}


var _DoHeaders = function(text) {

	// Setext-style headers:
	//	Header 1
	//	========
	//
	//	Header 2
	//	--------
	//
	text = text.replace(/^(.+)[ \t]*\n=+[ \t]*\n+/gm,
		function(wholeMatch,m1){return hashBlock('<h1 id="' + headerId(m1) + '">' + _RunSpanGamut(m1) + "</h1>");});

	text = text.replace(/^(.+)[ \t]*\n-+[ \t]*\n+/gm,
		function(matchFound,m1){return hashBlock('<h2 id="' + headerId(m1) + '">' + _RunSpanGamut(m1) + "</h2>");});

	// atx-style headers:
	//  # Header 1
	//  ## Header 2
	//  ## Header 2 with closing hashes ##
	//  ...
	//  ###### Header 6
	//

	/*
		text = text.replace(/
			^(\#{1,6})				// $1 = string of #'s
			[ \t]*
			(.+?)					// $2 = Header text
			[ \t]*
			\#*						// optional closing #'s (not counted)
			\n+
		/gm, function() {...});
	*/

	text = text.replace(/^(\#{1,6})[ \t]*(.+?)[ \t]*\#*\n+/gm,
		function(wholeMatch,m1,m2) {
			var h_level = m1.length;
			return hashBlock("<h" + h_level + ' id="' + headerId(m2) + '">' + _RunSpanGamut(m2) + "</h" + h_level + ">");
		});

	function headerId(m) {
		return m.replace(/[^\w]/g, '').toLowerCase();
	}
	return text;
}

// This declaration keeps Dojo compressor from outputting garbage:
var _ProcessListItems;

var _DoLists = function(text) {
//
// Form HTML ordered (numbered) and unordered (bulleted) lists.
//

	// attacklab: add sentinel to hack around khtml/safari bug:
	// http://bugs.webkit.org/show_bug.cgi?id=11231
	text += "~0";

	// Re-usable pattern to match any entirel ul or ol list:

	/*
		var whole_list = /
		(									// $1 = whole list
			(								// $2
				[ ]{0,3}					// attacklab: g_tab_width - 1
				([*+-]|\d+[.])				// $3 = first list item marker
				[ \t]+
			)
			[^\r]+?
			(								// $4
				~0							// sentinel for workaround; should be $
			|
				\n{2,}
				(?=\S)
				(?!							// Negative lookahead for another list item marker
					[ \t]*
					(?:[*+-]|\d+[.])[ \t]+
				)
			)
		)/g
	*/
	var whole_list = /^(([ ]{0,3}([*+-]|\d+[.])[ \t]+)[^\r]+?(~0|\n{2,}(?=\S)(?![ \t]*(?:[*+-]|\d+[.])[ \t]+)))/gm;

	if (g_list_level) {
		text = text.replace(whole_list,function(wholeMatch,m1,m2) {
			var list = m1;
			var list_type = (m2.search(/[*+-]/g)>-1) ? "ul" : "ol";

			// Turn double returns into triple returns, so that we can make a
			// paragraph for the last item in a list, if necessary:
			list = list.replace(/\n{2,}/g,"\n\n\n");;
			var result = _ProcessListItems(list);

			// Trim any trailing whitespace, to put the closing `</$list_type>`
			// up on the preceding line, to get it past the current stupid
			// HTML block parser. This is a hack to work around the terrible
			// hack that is the HTML block parser.
			result = result.replace(/\s+$/,"");
			result = "<"+list_type+">" + result + "</"+list_type+">\n";
			return result;
		});
	} else {
		whole_list = /(\n\n|^\n?)(([ ]{0,3}([*+-]|\d+[.])[ \t]+)[^\r]+?(~0|\n{2,}(?=\S)(?![ \t]*(?:[*+-]|\d+[.])[ \t]+)))/g;
		text = text.replace(whole_list,function(wholeMatch,m1,m2,m3) {
			var runup = m1;
			var list = m2;

			var list_type = (m3.search(/[*+-]/g)>-1) ? "ul" : "ol";
			// Turn double returns into triple returns, so that we can make a
			// paragraph for the last item in a list, if necessary:
			var list = list.replace(/\n{2,}/g,"\n\n\n");;
			var result = _ProcessListItems(list);
			result = runup + "<"+list_type+">\n" + result + "</"+list_type+">\n";
			return result;
		});
	}

	// attacklab: strip sentinel
	text = text.replace(/~0/,"");

	return text;
}

_ProcessListItems = function(list_str) {
//
//  Process the contents of a single ordered or unordered list, splitting it
//  into individual list items.
//
	// The $g_list_level global keeps track of when we're inside a list.
	// Each time we enter a list, we increment it; when we leave a list,
	// we decrement. If it's zero, we're not in a list anymore.
	//
	// We do this because when we're not inside a list, we want to treat
	// something like this:
	//
	//    I recommend upgrading to version
	//    8. Oops, now this line is treated
	//    as a sub-list.
	//
	// As a single paragraph, despite the fact that the second line starts
	// with a digit-period-space sequence.
	//
	// Whereas when we're inside a list (or sub-list), that line will be
	// treated as the start of a sub-list. What a kludge, huh? This is
	// an aspect of Markdown's syntax that's hard to parse perfectly
	// without resorting to mind-reading. Perhaps the solution is to
	// change the syntax rules such that sub-lists must start with a
	// starting cardinal number; e.g. "1." or "a.".

	g_list_level++;

	// trim trailing blank lines:
	list_str = list_str.replace(/\n{2,}$/,"\n");

	// attacklab: add sentinel to emulate \z
	list_str += "~0";

	/*
		list_str = list_str.replace(/
			(\n)?							// leading line = $1
			(^[ \t]*)						// leading whitespace = $2
			([*+-]|\d+[.]) [ \t]+			// list marker = $3
			([^\r]+?						// list item text   = $4
			(\n{1,2}))
			(?= \n* (~0 | \2 ([*+-]|\d+[.]) [ \t]+))
		/gm, function(){...});
	*/
	list_str = list_str.replace(/(\n)?(^[ \t]*)([*+-]|\d+[.])[ \t]+([^\r]+?(\n{1,2}))(?=\n*(~0|\2([*+-]|\d+[.])[ \t]+))/gm,
		function(wholeMatch,m1,m2,m3,m4){
			var item = m4;
			var leading_line = m1;
			var leading_space = m2;

			if (leading_line || (item.search(/\n{2,}/)>-1)) {
				item = _RunBlockGamut(_Outdent(item));
			}
			else {
				// Recursion for sub-lists:
				item = _DoLists(_Outdent(item));
				item = item.replace(/\n$/,""); // chomp(item)
				item = _RunSpanGamut(item);
			}

			return  "<li>" + item + "</li>\n";
		}
	);

	// attacklab: strip sentinel
	list_str = list_str.replace(/~0/g,"");

	g_list_level--;
	return list_str;
}


var _DoCodeBlocks = function(text) {
//
//  Process Markdown `<pre><code>` blocks.
//

	/*
		text = text.replace(text,
			/(?:\n\n|^)
			(								// $1 = the code block -- one or more lines, starting with a space/tab
				(?:
					(?:[ ]{4}|\t)			// Lines must start with a tab or a tab-width of spaces - attacklab: g_tab_width
					.*\n+
				)+
			)
			(\n*[ ]{0,3}[^ \t\n]|(?=~0))	// attacklab: g_tab_width
		/g,function(){...});
	*/

	// attacklab: sentinel workarounds for lack of \A and \Z, safari\khtml bug
	text += "~0";

	text = text.replace(/(?:\n\n|^)((?:(?:[ ]{4}|\t).*\n+)+)(\n*[ ]{0,3}[^ \t\n]|(?=~0))/g,
		function(wholeMatch,m1,m2) {
			var codeblock = m1;
			var nextChar = m2;

			codeblock = _EncodeCode( _Outdent(codeblock));
			codeblock = _Detab(codeblock);
			codeblock = codeblock.replace(/^\n+/g,""); // trim leading newlines
			codeblock = codeblock.replace(/\n+$/g,""); // trim trailing whitespace

			codeblock = "<pre><code>" + codeblock + "\n</code></pre>";

			return hashBlock(codeblock) + nextChar;
		}
	);

	// attacklab: strip sentinel
	text = text.replace(/~0/,"");

	return text;
};

var _DoGithubCodeBlocks = function(text) {
//
//  Process Github-style code blocks
//  Example:
//  ```ruby
//  def hello_world(x)
//    puts "Hello, #{x}"
//  end
//  ```
//


	// attacklab: sentinel workarounds for lack of \A and \Z, safari\khtml bug
	text += "~0";

	text = text.replace(/(?:^|\n)```(.*)\n([\s\S]*?)\n```/g,
		function(wholeMatch,m1,m2) {
			var language = m1;
			var codeblock = m2;

			codeblock = _EncodeCode(codeblock);
			codeblock = _Detab(codeblock);
			codeblock = codeblock.replace(/^\n+/g,""); // trim leading newlines
			codeblock = codeblock.replace(/\n+$/g,""); // trim trailing whitespace

			codeblock = "<pre><code" + (language ? " class=\"" + language + '"' : "") + ">" + codeblock + "\n</code></pre>";

			return hashBlock(codeblock);
		}
	);

	// attacklab: strip sentinel
	text = text.replace(/~0/,"");

	return text;
}

var hashBlock = function(text) {
	text = text.replace(/(^\n+|\n+$)/g,"");
	return "\n\n~K" + (g_html_blocks.push(text)-1) + "K\n\n";
}

var _DoCodeSpans = function(text) {
//
//   *  Backtick quotes are used for <code></code> spans.
//
//   *  You can use multiple backticks as the delimiters if you want to
//	 include literal backticks in the code span. So, this input:
//
//		 Just type ``foo `bar` baz`` at the prompt.
//
//	   Will translate to:
//
//		 <p>Just type <code>foo `bar` baz</code> at the prompt.</p>
//
//	There's no arbitrary limit to the number of backticks you
//	can use as delimters. If you need three consecutive backticks
//	in your code, use four for delimiters, etc.
//
//  *  You can use spaces to get literal backticks at the edges:
//
//		 ... type `` `bar` `` ...
//
//	   Turns to:
//
//		 ... type <code>`bar`</code> ...
//

	/*
		text = text.replace(/
			(^|[^\\])					// Character before opening ` can't be a backslash
			(`+)						// $2 = Opening run of `
			(							// $3 = The code block
				[^\r]*?
				[^`]					// attacklab: work around lack of lookbehind
			)
			\2							// Matching closer
			(?!`)
		/gm, function(){...});
	*/

	text = text.replace(/(^|[^\\])(`+)([^\r]*?[^`])\2(?!`)/gm,
		function(wholeMatch,m1,m2,m3,m4) {
			var c = m3;
			c = c.replace(/^([ \t]*)/g,"");	// leading whitespace
			c = c.replace(/[ \t]*$/g,"");	// trailing whitespace
			c = _EncodeCode(c);
			return m1+"<code>"+c+"</code>";
		});

	return text;
}

var _EncodeCode = function(text) {
//
// Encode/escape certain characters inside Markdown code runs.
// The point is that in code, these characters are literals,
// and lose their special Markdown meanings.
//
	// Encode all ampersands; HTML entities are not
	// entities within a Markdown code span.
	text = text.replace(/&/g,"&amp;");

	// Do the angle bracket song and dance:
	text = text.replace(/</g,"&lt;");