2 // showdown.js -- A javascript port of Markdown.
4 // Copyright (c) 2007 John Fraser.
6 // Original Markdown Copyright (c) 2004-2005 John Gruber
7 // <http://daringfireball.net/projects/markdown/>
9 // The full source distribution is at:
15 // <http://www.attacklab.net/>
19 // Wherever possible, Showdown is a straight, line-by-line port
20 // of the Perl version of Markdown.
22 // This is not a normal parser design; it's basically just a
23 // series of string substitutions. It's hard to read and
24 // maintain this way, but keeping Showdown close to the original
25 // design makes it easier to port new features.
27 // More importantly, Showdown behaves like markdown.pl in most
28 // edge cases. So web applications can do client-side preview
29 // in Javascript, and then build identical HTML on the server.
31 // This port needs the new RegExp functionality of ECMA 262,
32 // 3rd Edition (i.e. Javascript 1.5). Most modern web browsers
33 // should do fine. Even with the new regular expression features,
34 // We do a lot of work to emulate Perl's regex functionality.
35 // The tricky changes in this file mostly have the "attacklab:"
36 // label. Major or self-explanatory changes don't.
38 // Smart diff tools like Araxis Merge will be able to match up
39 // this file with markdown.pl in a useful way. A little tweaking
40 // helps: in a copy of markdown.pl, replace "#" with "//" and
41 // replace "$text" with "text". Be sure to ignore whitespace
49 // var text = "Markdown *rocks*.";
51 // var converter = new Attacklab.showdown.converter();
52 // var html = converter.makeHtml(text);
56 // Note: move the sample code to the bottom of this
57 // file before uncommenting it.
62 // Attacklab namespace
64 var Attacklab = Attacklab || {}
69 Attacklab.showdown = Attacklab.showdown || {}
74 // Wraps all "globals" so that the only thing
75 // exposed is makeHtml().
77 Attacklab.showdown.converter = function() {
80 // g_urls and g_titles allow arbitrary user-entered strings as keys. This
81 // caused an exception (and hence stopped the rendering) when the user entered
82 // e.g. [push] or [__proto__]. Adding a prefix to the actual key prevents this
83 // (since no builtin property starts with "s_"). See
84 // http://meta.stackoverflow.com/questions/64655/strange-wmd-bug
85 // (granted, switching from Array() to Object() alone would have left only __proto__
87 var SaveHash = function () {
88 this.set = function (key, value) {
89 this["s_" + key] = value;
91 this.get = function (key) {
92 return this["s_" + key];
100 // Global hashes, used by various utility routines
105 // Used to track when we're inside an ordered or unordered list
106 // (see _ProcessListItems() for details):
107 var g_list_level = 0;
110 this.makeHtml = function(text) {
112 // Main function. The order in which other subs are called here is
113 // essential. Link and image substitutions need to happen before
114 // _EscapeSpecialCharsWithinTagAttributes(), so that any *'s or _'s in the <a>
115 // and <img> tags get encoded.
117 text = html_sanitize(text, function(url) {return url;}, function(id) {return id;});
119 // Clear the global hashes. If we don't clear these, you get conflicts
120 // from other articles when generating a page which contains more than
121 // one article (e.g. an index page that shows the N most recent
123 g_urls = new SaveHash();
124 g_titles = new SaveHash();
125 g_html_blocks = new Array();
127 // attacklab: Replace ~ with ~T
128 // This lets us use tilde as an escape char to avoid md5 hashes
129 // The choice of character is arbitray; anything that isn't
130 // magic in Markdown will work.
131 text = text.replace(/~/g,"~T");
133 // attacklab: Replace $ with ~D
134 // RegExp interprets $ as a special character
135 // when it's in a replacement string
136 text = text.replace(/\$/g,"~D");
138 // Standardize line endings
139 text = text.replace(/\r\n/g,"\n"); // DOS to Unix
140 text = text.replace(/\r/g,"\n"); // Mac to Unix
142 // Make sure text begins and ends with a couple of newlines:
143 text = "\n\n" + text + "\n\n";
145 // Convert all tabs to spaces.
148 // Strip any lines consisting only of spaces and tabs.
149 // This makes subsequent regexen easier to write, because we can
150 // match consecutive blank lines with /\n+/ instead of something
151 // contorted like /[ \t]*\n+/ .
152 text = text.replace(/^[ \t]+$/mg,"");
154 // Turn block-level HTML blocks into hash entries
155 text = _HashHTMLBlocks(text);
157 // Strip link definitions, store in hashes.
158 text = _StripLinkDefinitions(text);
160 text = _RunBlockGamut(text);
162 text = _UnescapeSpecialChars(text);
164 // attacklab: Restore dollar signs
165 text = text.replace(/~D/g,"$$");
167 // attacklab: Restore tildes
168 text = text.replace(/~T/g,"~");
170 text = text.replace(/&lt;/g,"<");
171 text = text.replace(/&gt;/g,">");
176 var _StripLinkDefinitions = function(text) {
178 // Strips link definitions from text, stores the URLs and titles in
182 // Link defs are in the form: ^[id]: url "optional title"
185 var text = text.replace(/
186 ^[ ]{0,3}\[(.+)\]: // id = $1 attacklab: g_tab_width - 1
188 \n? // maybe *one* newline
190 <?(\S+?)>? // url = $2
191 (?=\s|$) // lookahead for whitespace instead of the lookbehind removed below
193 \n? // maybe one newline
195 ( // (potential) title = $3
196 (\n*) // any lines skipped = $4 attacklab: lookbehind removed
202 )? // title is optional
207 var text = text.replace(/^[ ]{0,3}\[(.+)\]:[ \t]*\n?[ \t]*<?(\S+?)>?(?=\s|$)[ \t]*\n?[ \t]*((\n*)["(](.+?)[")][ \t]*)?(?:\n+)/gm,
208 function (wholeMatch,m1,m2,m3,m4,m5) {
209 m1 = m1.toLowerCase();
210 g_urls.set(m1, _EncodeAmpsAndAngles(m2)); // Link IDs are case-insensitive
212 // Oops, found blank lines, so it's not a title.
213 // Put back the parenthetical statement we stole.
216 g_titles.set(m1, m5.replace(/"/g,"""));
219 // Completely remove the definition from the text
227 var _HashHTMLBlocks = function(text) {
229 // Hashify HTML blocks:
230 // We only want to do this for block-level HTML tags, such as headers,
231 // lists, and tables. That's because we still want to wrap <p>s around
232 // "paragraphs" that are wrapped in non-block-level tags, such as anchors,
233 // phrase emphasis, and spans. The list of tags we're looking for is
235 var block_tags_a = "p|div|h[1-6]|blockquote|pre|table|dl|ol|ul|script|noscript|form|fieldset|iframe|math|ins|del"
236 var block_tags_b = "p|div|h[1-6]|blockquote|pre|table|dl|ol|ul|script|noscript|form|fieldset|iframe|math"
238 // First, look for nested blocks, e.g.:
241 // tags for inner block must be indented.
245 // The outermost tags must start at the left margin for this to match, and
246 // the inner nested divs must be indented.
247 // We need to do this before the next, more liberal match, because the next
248 // match will start at the first `<div>` and stop at the first `</div>`.
250 // attacklab: This regex can be expensive when it fails.
252 var text = text.replace(/
254 ^ // start of line (with /m)
255 <($block_tags_a) // start tag = $2
257 // attacklab: hack around khtml/pcre bug...
258 [^\r]*?\n // any number of lines, minimally matching
259 </\2> // the matching end tag
260 [ \t]* // trailing spaces/tabs
261 (?=\n+) // followed by a newline
262 ) // attacklab: there are sentinel newlines at end of document
263 /gm,function(){...}};
265 text = text.replace(/^(<(p|div|h[1-6]|blockquote|pre|table|dl|ol|ul|script|noscript|form|fieldset|iframe|math|ins|del)\b[^\r]*?\n<\/\2>[ \t]*(?=\n+))/gm,hashElement);
268 // Now match more liberally, simply from `\n<tag>` to `</tag>\n`
272 var text = text.replace(/
274 ^ // start of line (with /m)
275 <($block_tags_b) // start tag = $2
277 // attacklab: hack around khtml/pcre bug...
278 [^\r]*? // any number of lines, minimally matching
279 .*</\2> // the matching end tag
280 [ \t]* // trailing spaces/tabs
281 (?=\n+) // followed by a newline
282 ) // attacklab: there are sentinel newlines at end of document
283 /gm,function(){...}};
285 text = text.replace(/^(<(p|div|h[1-6]|blockquote|pre|table|dl|ol|ul|script|noscript|form|fieldset|iframe|math)\b[^\r]*?.*<\/\2>[ \t]*(?=\n+)\n)/gm,hashElement);
287 // Special case just for <hr />. It was easier to make a special case than
288 // to make the other regex more complicated.
291 text = text.replace(/
292 \n // Starting after a blank line
295 (<(hr) // start tag = $2
298 \/?>) // the matching end tag
300 (?=\n{2,}) // followed by a blank line
304 text = text.replace(/\n[ ]{0,3}((<(hr)\b([^<>])*?\/?>)[ \t]*(?=\n{2,}))/g,hashElement);
306 // Special case for standalone HTML comments:
309 text = text.replace(/
310 \n\n // Starting after a blank line
311 [ ]{0,3} // attacklab: g_tab_width - 1
314 (--(?:|(?:[^>-]|-[^>])(?:[^-]|-[^-])*)--) // see http://www.w3.org/TR/html-markup/syntax.html#comments
317 (?=\n{2,}) // followed by a blank line
321 text = text.replace(/\n\n[ ]{0,3}(<!(--(?:|(?:[^>-]|-[^>])(?:[^-]|-[^-])*)--)>[ \t]*(?=\n{2,}))/g, hashElement);
323 // PHP and ASP-style processor instructions (<?...?> and <%...%>)
326 text = text.replace(/
328 \n\n // Starting after a blank line
331 [ ]{0,3} // attacklab: g_tab_width - 1
338 (?=\n{2,}) // followed by a blank line
342 text = text.replace(/(?:\n\n)([ ]{0,3}(?:<([?%])[^\r]*?\2>)[ \t]*(?=\n{2,}))/g,hashElement);
347 var hashElement = function(wholeMatch,m1) {
351 blockText = blockText.replace(/^\n+/,"");
353 // strip trailing blank lines
354 blockText = blockText.replace(/\n+$/g,"");
356 // Replace the element text with a marker ("~KxK" where x is its key)
357 blockText = "\n\n~K" + (g_html_blocks.push(blockText)-1) + "K\n\n";
362 var _RunBlockGamut = function(text, doNotUnhash) {
364 // These are all the transformations that form block-level
365 // tags like paragraphs, headers, and list items.
367 text = _DoHeaders(text);
369 // Do Horizontal Rules:
370 var key = hashBlock("<hr />");
371 text = text.replace(/^[ ]{0,2}([ ]?\*[ ]?){3,}[ \t]*$/gm,key);
372 text = text.replace(/^[ ]{0,2}([ ]?-[ ]?){3,}[ \t]*$/gm,key);
373 text = text.replace(/^[ ]{0,2}([ ]?_[ ]?){3,}[ \t]*$/gm,key);
375 text = _DoLists(text);
376 text = _DoCodeBlocks(text);
377 text = _DoBlockQuotes(text);
379 // We already ran _HashHTMLBlocks() before, in Markdown(), but that
380 // was to escape raw HTML in the original Markdown source. This time,
381 // we're escaping the markup we've just created, so that we don't wrap
382 // <p> tags around block-level tags.
383 text = _HashHTMLBlocks(text);
384 text = _FormParagraphs(text, doNotUnhash);
390 var _RunSpanGamut = function(text) {
392 // These are all the transformations that occur *within* block-level
393 // tags like paragraphs, headers, and list items.
396 text = _DoCodeSpans(text);
397 text = _EscapeSpecialCharsWithinTagAttributes(text);
398 text = _EncodeBackslashEscapes(text);
400 // Process anchor and image tags. Images must come first,
401 // because ![foo][f] looks like an anchor.
402 text = _DoImages(text);
403 text = _DoAnchors(text);
405 // Make links out of things like `<http://example.com/>`
406 // Must come after _DoAnchors(), because you can use < and >
407 // delimiters in inline links like [this](<url>).
408 text = _DoAutoLinks(text);
409 text = _EncodeAmpsAndAngles(text);
410 text = _DoItalicsAndBold(text);
413 text = text.replace(/ +\n/g," <br />\n");
418 var _EscapeSpecialCharsWithinTagAttributes = function(text) {
420 // Within tags -- meaning between < and > -- encode [\ ` * _] so they
421 // don't conflict with their use in Markdown for code, italics and strong.
424 // Build a regex to find HTML tags and comments. See Friedl's
425 // "Mastering Regular Expressions", 2nd Ed., pp. 200-201.
427 // SE: changed the comment part of the regex
429 var regex = /(<[a-z\/!$]("[^"]*"|'[^']*'|[^'">])*>|<!(--(?:|(?:[^>-]|-[^>])(?:[^-]|-[^-])*)--)>)/gi;
431 text = text.replace(regex, function(wholeMatch) {
432 var tag = wholeMatch.replace(/(.)<\/?code>(?=.)/g,"$1`");
433 tag = escapeCharacters(tag,"\\`*_");
440 var _DoAnchors = function(text) {
442 // Turn Markdown link shortcuts into XHTML <a> tags.
445 // First, handle reference-style links: [link text] [id]
449 text = text.replace(/
450 ( // wrap whole match in $1
454 \[[^\]]*\] // allow brackets nested one level
456 [^\[] // or anything else
461 [ ]? // one optional space
462 (?:\n[ ]*)? // one optional newline followed by spaces
467 )()()()() // pad remaining backreferences
468 /g,_DoAnchors_callback);
470 text = text.replace(/(\[((?:\[[^\]]*\]|[^\[\]])*)\][ ]?(?:\n[ ]*)?\[(.*?)\])()()()()/g,writeAnchorTag);
473 // Next, inline-style links: [link text](url "optional title")
477 text = text.replace(/
478 ( // wrap whole match in $1
482 \[[^\]]*\] // allow brackets nested one level
484 [^\[\]] // or anything else
490 () // no id, so leave $3 empty
493 \([^)]*\) // allow one level of (correctly nested) parens (think MSDN)
500 (['"]) // quote char = $6
503 [ \t]* // ignore any spaces/tabs between closing quote and )
504 )? // title is optional
510 text = text.replace(/(\[((?:\[[^\]]*\]|[^\[\]])*)\]\([ \t]*()<?((?:\([^)]*\)|[^()])*?)>?[ \t]*((['"])(.*?)\6[ \t]*)?\))/g,writeAnchorTag);
513 // Last, handle reference-style shortcuts: [link text]
514 // These must come last in case you've also got [link test][1]
515 // or [link test](/foo)
519 text = text.replace(/
520 ( // wrap whole match in $1
522 ([^\[\]]+) // link text = $2; can't contain '[' or ']'
524 )()()()()() // pad rest of backreferences
527 text = text.replace(/(\[([^\[\]]+)\])()()()()()/g, writeAnchorTag);
532 var writeAnchorTag = function(wholeMatch,m1,m2,m3,m4,m5,m6,m7) {
533 if (m7 == undefined) m7 = "";
534 var whole_match = m1;
536 var link_id = m3.toLowerCase();
542 // lower-case and turn embedded newlines into spaces
543 link_id = link_text.toLowerCase().replace(/ ?\n/g," ");
547 if (g_urls.get(link_id) != undefined) {
548 url = g_urls.get(link_id);
549 if (g_titles.get(link_id) != undefined) {
550 title = g_titles.get(link_id);
554 if (whole_match.search(/\(\s*\)$/m)>-1) {
555 // Special case for explicit empty url
563 url = escapeCharacters(url,"*_");
564 var result = "<a href=\"" + url + "\"";
567 title = title.replace(/"/g,""");
568 title = escapeCharacters(title,"*_");
569 result += " title=\"" + title + "\"";
572 result += ">" + link_text + "</a>";
578 var _DoImages = function(text) {
580 // Turn Markdown image shortcuts into <img> tags.
584 // First, handle reference-style labeled images: ![alt text][id]
588 text = text.replace(/
589 ( // wrap whole match in $1
591 (.*?) // alt text = $2
594 [ ]? // one optional space
595 (?:\n[ ]*)? // one optional newline followed by spaces
600 )()()()() // pad rest of backreferences
603 text = text.replace(/(!\[(.*?)\][ ]?(?:\n[ ]*)?\[(.*?)\])()()()()/g,writeImageTag);
606 // Next, handle inline images: ![alt text](url "optional title")
607 // Don't forget: encode * and _
610 text = text.replace(/
611 ( // wrap whole match in $1
613 (.*?) // alt text = $2
615 \s? // One optional whitespace character
618 () // no id, so leave $3 empty
619 <?(\S+?)>? // src url = $4
622 (['"]) // quote char = $6
626 )? // title is optional
631 text = text.replace(/(!\[(.*?)\]\s?\([ \t]*()<?(\S+?)>?[ \t]*((['"])(.*?)\6[ \t]*)?\))/g,writeImageTag);
636 var writeImageTag = function(wholeMatch,m1,m2,m3,m4,m5,m6,m7) {
637 var whole_match = m1;
639 var link_id = m3.toLowerCase();
643 if (!title) title = "";
647 // lower-case and turn embedded newlines into spaces
648 link_id = alt_text.toLowerCase().replace(/ ?\n/g," ");
652 if (g_urls.get(link_id) != undefined) {
653 url = g_urls.get(link_id);
654 if (g_titles.get(link_id) != undefined) {
655 title = g_titles.get(link_id);
663 alt_text = alt_text.replace(/"/g,""");
664 url = escapeCharacters(url,"*_");
665 var result = "<img src=\"" + url + "\" alt=\"" + alt_text + "\"";
667 // attacklab: Markdown.pl adds empty title attributes to images.
668 // Replicate this bug.
671 title = title.replace(/"/g,""");
672 title = escapeCharacters(title,"*_");
673 result += " title=\"" + title + "\"";
682 var _DoHeaders = function(text) {
684 // Setext-style headers:
691 text = text.replace(/^(.+)[ \t]*\n=+[ \t]*\n+/gm,
692 function(wholeMatch,m1){return "<h1>" + _RunSpanGamut(m1) + "</h1>\n\n";});
694 text = text.replace(/^(.+)[ \t]*\n-+[ \t]*\n+/gm,
695 function(matchFound,m1){return "<h2>" + _RunSpanGamut(m1) + "</h2>\n\n";});
697 // atx-style headers:
700 // ## Header 2 with closing hashes ##
706 text = text.replace(/
707 ^(\#{1,6}) // $1 = string of #'s
709 (.+?) // $2 = Header text
711 \#* // optional closing #'s (not counted)
713 /gm, function() {...});
716 text = text.replace(/^(\#{1,6})[ \t]*(.+?)[ \t]*\#*\n+/gm,
717 function(wholeMatch,m1,m2) {
718 var h_level = m1.length;
719 return "<h" + h_level + ">" + _RunSpanGamut(m2) + "</h" + h_level + ">\n\n";
725 // This declaration keeps Dojo compressor from outputting garbage:
726 var _ProcessListItems;
728 var _DoLists = function(text) {
730 // Form HTML ordered (numbered) and unordered (bulleted) lists.
733 // attacklab: add sentinel to hack around khtml/safari bug:
734 // http://bugs.webkit.org/show_bug.cgi?id=11231
737 // Re-usable pattern to match any entirel ul or ol list:
743 [ ]{0,3} // attacklab: g_tab_width - 1
744 ([*+-]|\d+[.]) // $3 = first list item marker
749 ~0 // sentinel for workaround; should be $
753 (?! // Negative lookahead for another list item marker
755 (?:[*+-]|\d+[.])[ \t]+
760 var whole_list = /^(([ ]{0,3}([*+-]|\d+[.])[ \t]+)[^\r]+?(~0|\n{2,}(?=\S)(?![ \t]*(?:[*+-]|\d+[.])[ \t]+)))/gm;
763 text = text.replace(whole_list,function(wholeMatch,m1,m2) {
765 var list_type = (m2.search(/[*+-]/g)>-1) ? "ul" : "ol";
767 var result = _ProcessListItems(list, list_type);
769 // Trim any trailing whitespace, to put the closing `</$list_type>`
770 // up on the preceding line, to get it past the current stupid
771 // HTML block parser. This is a hack to work around the terrible
772 // hack that is the HTML block parser.
773 result = result.replace(/\s+$/,"");
774 result = "<"+list_type+">" + result + "</"+list_type+">\n";
778 whole_list = /(\n\n|^\n?)(([ ]{0,3}([*+-]|\d+[.])[ \t]+)[^\r]+?(~0|\n{2,}(?=\S)(?![ \t]*(?:[*+-]|\d+[.])[ \t]+)))/g;
779 text = text.replace(whole_list,function(wholeMatch,m1,m2,m3) {
783 var list_type = (m3.search(/[*+-]/g)>-1) ? "ul" : "ol";
784 var result = _ProcessListItems(list, list_type);
785 result = runup + "<"+list_type+">\n" + result + "</"+list_type+">\n";
790 // attacklab: strip sentinel
791 text = text.replace(/~0/,"");
796 var _listItemMarkers = { ol: "\\d+[.]", ul: "[*+-]" };
798 _ProcessListItems = function(list_str, list_type) {
800 // Process the contents of a single ordered or unordered list, splitting it
801 // into individual list items.
803 // list_type is either "ul" or "ol".
805 // The $g_list_level global keeps track of when we're inside a list.
806 // Each time we enter a list, we increment it; when we leave a list,
807 // we decrement. If it's zero, we're not in a list anymore.
809 // We do this because when we're not inside a list, we want to treat
810 // something like this:
812 // I recommend upgrading to version
813 // 8. Oops, now this line is treated
816 // As a single paragraph, despite the fact that the second line starts
817 // with a digit-period-space sequence.
819 // Whereas when we're inside a list (or sub-list), that line will be
820 // treated as the start of a sub-list. What a kludge, huh? This is
821 // an aspect of Markdown's syntax that's hard to parse perfectly
822 // without resorting to mind-reading. Perhaps the solution is to
823 // change the syntax rules such that sub-lists must start with a
824 // starting cardinal number; e.g. "1." or "a.".
828 // trim trailing blank lines:
829 list_str = list_str.replace(/\n{2,}$/,"\n");
831 // attacklab: add sentinel to emulate \z
834 // In the original attacklab WMD, list_type was not given to this function, and anything
835 // that matched /[*+-]|\d+[.]/ would just create the next <li>, causing this mismatch:
837 // Markdown rendered by WMD rendered by MarkdownSharp
838 // ------------------------------------------------------------------
839 // 1. first 1. first 1. first
840 // 2. second 2. second 2. second
841 // - third 3. third * third
843 // We changed this to behave identical to MarkdownSharp. This is the constructed RegEx,
844 // with {MARKER} being one of \d+[.] or [*+-], depending on list_type:
846 list_str = list_str.replace(/
847 (^[ \t]*) // leading whitespace = $1
848 ({MARKER}) [ \t]+ // list marker = $2
849 ([^\r]+? // list item text = $3
851 (?= (~0 | \2 ({MARKER}) [ \t]+))
852 /gm, function(){...});
855 var marker = _listItemMarkers[list_type];
856 var re = new RegExp("(^[ \\t]*)(" + marker + ")[ \\t]+([^\\r]+?(\\n+))(?=(~0|\\1(" + marker + ")[ \\t]+))", "gm");
857 var last_item_had_a_double_newline = false;
858 list_str = list_str.replace(re,
859 function(wholeMatch,m1,m2,m3){
861 var leading_space = m1;
862 var ends_with_double_newline = /\n\n$/.test(item);
863 var contains_double_newline = ends_with_double_newline || item.search(/\n{2,}/)>-1;
865 if (contains_double_newline || last_item_had_a_double_newline) {
866 item = _RunBlockGamut(_Outdent(item), /* doNotUnhash = */ true);
869 // Recursion for sub-lists:
870 item = _DoLists(_Outdent(item));
871 item = item.replace(/\n$/,""); // chomp(item)
872 item = _RunSpanGamut(item);
874 last_item_had_a_double_newline = ends_with_double_newline;
875 return "<li>" + item + "</li>\n";
879 // attacklab: strip sentinel
880 list_str = list_str.replace(/~0/g,"");
887 var _DoCodeBlocks = function(text) {
889 // Process Markdown `<pre><code>` blocks.
893 text = text.replace(text,
895 ( // $1 = the code block -- one or more lines, starting with a space/tab
897 (?:[ ]{4}|\t) // Lines must start with a tab or a tab-width of spaces - attacklab: g_tab_width
901 (\n*[ ]{0,3}[^ \t\n]|(?=~0)) // attacklab: g_tab_width
905 // attacklab: sentinel workarounds for lack of \A and \Z, safari\khtml bug
908 text = text.replace(/(?:\n\n|^)((?:(?:[ ]{4}|\t).*\n+)+)(\n*[ ]{0,3}[^ \t\n]|(?=~0))/g,
909 function(wholeMatch,m1,m2) {
913 codeblock = _EncodeCode( _Outdent(codeblock));
914 codeblock = _Detab(codeblock);
915 codeblock = codeblock.replace(/^\n+/g,""); // trim leading newlines
916 codeblock = codeblock.replace(/\n+$/g,""); // trim trailing whitespace
918 codeblock = "<pre><code>" + codeblock + "\n</code></pre>";
920 return "\n\n" + codeblock + "\n\n" + nextChar;
924 // attacklab: strip sentinel
925 text = text.replace(/~0/,"");
930 var hashBlock = function(text) {
931 text = text.replace(/(^\n+|\n+$)/g,"");
932 return "\n\n~K" + (g_html_blocks.push(text)-1) + "K\n\n";
936 var _DoCodeSpans = function(text) {
938 // * Backtick quotes are used for <code></code> spans.
940 // * You can use multiple backticks as the delimiters if you want to
941 // include literal backticks in the code span. So, this input:
943 // Just type ``foo `bar` baz`` at the prompt.
945 // Will translate to:
947 // <p>Just type <code>foo `bar` baz</code> at the prompt.</p>
949 // There's no arbitrary limit to the number of backticks you
950 // can use as delimters. If you need three consecutive backticks
951 // in your code, use four for delimiters, etc.
953 // * You can use spaces to get literal backticks at the edges:
955 // ... type `` `bar` `` ...
959 // ... type <code>`bar`</code> ...
963 text = text.replace(/
964 (^|[^\\]) // Character before opening ` can't be a backslash
965 (`+) // $2 = Opening run of `
966 ( // $3 = The code block
968 [^`] // attacklab: work around lack of lookbehind
970 \2 // Matching closer
972 /gm, function(){...});
975 text = text.replace(/(^|[^\\])(`+)([^\r]*?[^`])\2(?!`)/gm,
976 function(wholeMatch,m1,m2,m3,m4) {
978 c = c.replace(/^([ \t]*)/g,""); // leading whitespace
979 c = c.replace(/[ \t]*$/g,""); // trailing whitespace
981 return m1+"<code>"+c+"</code>";
988 var _EncodeCode = function(text) {
990 // Encode/escape certain characters inside Markdown code runs.
991 // The point is that in code, these characters are literals,
992 // and lose their special Markdown meanings.
994 // Encode all ampersands; HTML entities are not
995 // entities within a Markdown code span.
996 text = text.replace(/&/g,"&");
998 // Do the angle bracket song and dance:
999 text = text.replace(/</g,"<");
1000 text = text.replace(/>/g,">");
1002 // Now, escape characters that are magic in Markdown:
1003 text = escapeCharacters(text,"\*_{}[]\\",false);
1005 // jj the line above breaks this:
1019 var _DoItalicsAndBold = function(text) {
1021 // <strong> must go first:
1022 text = text.replace(/(\*\*|__)(?=\S)([^\r]*?\S[\*_]*)\1/g,
1023 "<strong>$2</strong>");
1025 text = text.replace(/(\*|_)(?=\S)([^\r]*?\S)\1/g,
1032 var _DoBlockQuotes = function(text) {
1035 text = text.replace(/
1036 ( // Wrap whole match in $1
1038 ^[ \t]*>[ \t]? // '>' at the start of a line
1039 .+\n // rest of the first line
1040 (.+\n)* // subsequent consecutive lines
1044 /gm, function(){...});
1047 text = text.replace(/((^[ \t]*>[ \t]?.+\n(.+\n)*\n*)+)/gm,
1048 function(wholeMatch,m1) {
1051 // attacklab: hack around Konqueror 3.5.4 bug:
1052 // "----------bug".replace(/^-/g,"") == "bug"
1054 bq = bq.replace(/^[ \t]*>[ \t]?/gm,"~0"); // trim one level of quoting
1056 // attacklab: clean up hack
1057 bq = bq.replace(/~0/g,"");
1059 bq = bq.replace(/^[ \t]+$/gm,""); // trim whitespace-only lines
1060 bq = _RunBlockGamut(bq); // recurse
1062 bq = bq.replace(/(^|\n)/g,"$1 ");
1063 // These leading spaces screw with <pre> content, so we need to fix that:
1065 /(\s*<pre>[^\r]+?<\/pre>)/gm,
1066 function(wholeMatch,m1) {
1068 // attacklab: hack around Konqueror 3.5.4 bug:
1069 pre = pre.replace(/^ /mg,"~0");
1070 pre = pre.replace(/~0/g,"");
1074 return hashBlock("<blockquote>\n" + bq + "\n</blockquote>");
1080 var _FormParagraphs = function(text, doNotUnhash) {
1083 // $text - string to process with html <p> tags
1086 // Strip leading and trailing lines:
1087 text = text.replace(/^\n+/g,"");
1088 text = text.replace(/\n+$/g,"");
1090 var grafs = text.split(/\n{2,}/g);
1091 var grafsOut = new Array();
1096 var end = grafs.length;
1097 for (var i=0; i<end; i++) {
1100 // if this is an HTML marker, copy it
1101 if (str.search(/~K(\d+)K/g) >= 0) {
1104 else if (str.search(/\S/) >= 0) {
1105 str = _RunSpanGamut(str);
1106 str = str.replace(/^([ \t]*)/g,"<p>");
1113 // Unhashify HTML blocks
1116 end = grafsOut.length;
1117 for (var i=0; i<end; i++) {
1118 // if this is a marker for an html block...
1119 while (grafsOut[i].search(/~K(\d+)K/) >= 0) {
1120 var blockText = g_html_blocks[RegExp.$1];
1121 blockText = blockText.replace(/\$/g,"$$$$"); // Escape any dollar signs
1122 grafsOut[i] = grafsOut[i].replace(/~K\d+K/,blockText);
1126 return grafsOut.join("\n\n");
1130 var _EncodeAmpsAndAngles = function(text) {
1131 // Smart processing for ampersands and angle brackets that need to be encoded.
1133 // Ampersand-encoding based entirely on Nat Irons's Amputator MT plugin:
1134 // http://bumppo.net/projects/amputator/
1135 text = text.replace(/&(?!#?[xX]?(?:[0-9a-fA-F]+|\w+);)/g,"&");
1138 text = text.replace(/<(?![a-z\/?\$!])/gi,"<");
1144 var _EncodeBackslashEscapes = function(text) {
1146 // Parameter: String.
1147 // Returns: The string, with after processing the following backslash
1148 // escape sequences.
1151 // attacklab: The polite way to do this is with the new
1152 // escapeCharacters() function:
1154 // text = escapeCharacters(text,"\\",true);
1155 // text = escapeCharacters(text,"`*_{}[]()>#+-.!",true);
1157 // ...but we're sidestepping its use of the (slow) RegExp constructor
1158 // as an optimization for Firefox. This function gets called a LOT.
1160 text = text.replace(/\\(\\)/g,escapeCharacters_callback);
1161 text = text.replace(/\\([`*_{}\[\]()>#+-.!])/g,escapeCharacters_callback);
1166 var _DoAutoLinks = function(text) {
1168 text = text.replace(/<((https?|ftp|dict):[^'">\s]+)>/gi,"<a href=\"$1\">$1</a>");
1170 // Email addresses: <address@domain.foo>
1173 text = text.replace(/
1179 [-a-z0-9]+(\.[-a-z0-9]+)*\.[a-z]+
1182 /gi, _DoAutoLinks_callback());
1184 text = text.replace(/<(?:mailto:)?([-.\w]+\@[-a-z0-9]+(\.[-a-z0-9]+)*\.[a-z]+)>/gi,
1185 function(wholeMatch,m1) {
1186 return _EncodeEmailAddress( _UnescapeSpecialChars(m1) );
1194 var _EncodeEmailAddress = function(addr) {
1196 // Input: an email address, e.g. "foo@example.com"
1198 // Output: the email address as a mailto link, with each character
1199 // of the address encoded as either a decimal or hex entity, in
1200 // the hopes of foiling most address harvesting spam bots. E.g.:
1202 // <a href="mailto:foo@e
1203 // xample.com">foo
1204 // @example.com</a>
1206 // Based on a filter by Matthew Wickline, posted to the BBEdit-Talk
1207 // mailing list: <http://tinyurl.com/yu7ue>
1210 // attacklab: why can't javascript speak hex?
1211 function char2hex(ch) {
1212 var hexDigits = '0123456789ABCDEF';
1213 var dec = ch.charCodeAt(0);
1214 return(hexDigits.charAt(dec>>4) + hexDigits.charAt(dec&15));
1218 function(ch){return "&#"+ch.charCodeAt(0)+";";},
1219 function(ch){return "&#x"+char2hex(ch)+";";},
1220 function(ch){return ch;}
1223 addr = "mailto:" + addr;
1225 addr = addr.replace(/./g, function(ch) {
1227 // this *must* be encoded. I insist.
1228 ch = encode[Math.floor(Math.random()*2)](ch);
1229 } else if (ch !=":") {
1230 // leave ':' alone (to spot mailto: later)
1231 var r = Math.random();
1232 // roughly 10% raw, 45% hex, 45% dec
1234 r > .9 ? encode[2](ch) :
1235 r > .45 ? encode[1](ch) :
1242 addr = "<a href=\"" + addr + "\">" + addr + "</a>";
1243 addr = addr.replace(/">.+:/g,"\">"); // strip the mailto: from the visible part
1249 var _UnescapeSpecialChars = function(text) {
1251 // Swap back in all the special characters we've hidden.
1253 text = text.replace(/~E(\d+)E/g,
1254 function(wholeMatch,m1) {
1255 var charCodeToReplace = parseInt(m1);
1256 return String.fromCharCode(charCodeToReplace);
1263 var _Outdent = function(text) {
1265 // Remove one level of line-leading tabs or spaces
1268 // attacklab: hack around Konqueror 3.5.4 bug:
1269 // "----------bug".replace(/^-/g,"") == "bug"
1271 text = text.replace(/^(\t|[ ]{1,4})/gm,"~0"); // attacklab: g_tab_width
1273 // attacklab: clean up hack
1274 text = text.replace(/~0/g,"")
1279 var _Detab = function (text) {
1280 if (!/\t/.test(text))
1283 var spaces = [" ", " ", " ", " "],
1287 return text.replace(/[\n\t]/g, function (match, offset) {
1288 if (match === "\n") {
1292 v = (offset - skew) % 4;
1299 // attacklab: Utility functions
1303 var escapeCharacters = function(text, charsToEscape, afterBackslash) {
1304 // First we have to escape the escape characters so that
1305 // we can build a character class out of them
1306 var regexString = "([" + charsToEscape.replace(/([\[\]\\])/g,"\\$1") + "])";
1308 if (afterBackslash) {
1309 regexString = "\\\\" + regexString;
1312 var regex = new RegExp(regexString,"g");
1313 text = text.replace(regex,escapeCharacters_callback);
1319 var escapeCharacters_callback = function(wholeMatch,m1) {
1320 var charCodeToEscape = m1.charCodeAt(0);
1321 return "~E"+charCodeToEscape+"E";
1324 } // end of Attacklab.showdown.converter
1327 // Version 0.9 used the Showdown namespace instead of Attacklab.showdown
1328 // The old namespace is deprecated, but we'll support it for now:
1329 var Showdown = Attacklab.showdown;
1331 // If anyone's interested, tell the world that this file's been loaded
1332 if (Attacklab.fileLoaded) {
1333 Attacklab.fileLoaded("showdown.js");