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 // Redistributable under a BSD-style open source license.
10 // See license.txt for more information.
12 // The full source distribution is at:
18 // <http://www.attacklab.net/>
22 // Wherever possible, Showdown is a straight, line-by-line port
23 // of the Perl version of Markdown.
25 // This is not a normal parser design; it's basically just a
26 // series of string substitutions. It's hard to read and
27 // maintain this way, but keeping Showdown close to the original
28 // design makes it easier to port new features.
30 // More importantly, Showdown behaves like markdown.pl in most
31 // edge cases. So web applications can do client-side preview
32 // in Javascript, and then build identical HTML on the server.
34 // This port needs the new RegExp functionality of ECMA 262,
35 // 3rd Edition (i.e. Javascript 1.5). Most modern web browsers
36 // should do fine. Even with the new regular expression features,
37 // We do a lot of work to emulate Perl's regex functionality.
38 // The tricky changes in this file mostly have the "attacklab:"
39 // label. Major or self-explanatory changes don't.
41 // Smart diff tools like Araxis Merge will be able to match up
42 // this file with markdown.pl in a useful way. A little tweaking
43 // helps: in a copy of markdown.pl, replace "#" with "//" and
44 // replace "$text" with "text". Be sure to ignore whitespace
52 // var text = "Markdown *rocks*.";
54 // var converter = new Showdown.converter();
55 // var html = converter.makeHtml(text);
59 // Note: move the sample code to the bottom of this
60 // file before uncommenting it.
67 Roo.namespace('Roo.ux');
69 Roo.ux.Showdown.makeHTML = function(text) {
70 var c = new Roo.ux.Showdown.converter();
71 return c.makeHtml(text);
76 // Wraps all "globals" so that the only thing
77 // exposed is makeHtml().
79 Roo.ux.Showdown.converter = function() {
85 // Global hashes, used by various utility routines
90 // Used to track when we're inside an ordered or unordered list
91 // (see _ProcessListItems() for details):
95 this.makeHtml = function(text) {
97 // Main function. The order in which other subs are called here is
98 // essential. Link and image substitutions need to happen before
99 // _EscapeSpecialCharsWithinTagAttributes(), so that any *'s or _'s in the <a>
100 // and <img> tags get encoded.
103 // Clear the global hashes. If we don't clear these, you get conflicts
104 // from other articles when generating a page which contains more than
105 // one article (e.g. an index page that shows the N most recent
107 g_urls = new Array();
108 g_titles = new Array();
109 g_html_blocks = new Array();
111 // attacklab: Replace ~ with ~T
112 // This lets us use tilde as an escape char to avoid md5 hashes
113 // The choice of character is arbitray; anything that isn't
114 // magic in Markdown will work.
115 text = text.replace(/~/g,"~T");
117 // attacklab: Replace $ with ~D
118 // RegExp interprets $ as a special character
119 // when it's in a replacement string
120 text = text.replace(/\$/g,"~D");
122 // Standardize line endings
123 text = text.replace(/\r\n/g,"\n"); // DOS to Unix
124 text = text.replace(/\r/g,"\n"); // Mac to Unix
126 // Make sure text begins and ends with a couple of newlines:
127 text = "\n\n" + text + "\n\n";
129 // Convert all tabs to spaces.
132 // Strip any lines consisting only of spaces and tabs.
133 // This makes subsequent regexen easier to write, because we can
134 // match consecutive blank lines with /\n+/ instead of something
135 // contorted like /[ \t]*\n+/ .
136 text = text.replace(/^[ \t]+$/mg,"");
138 // Turn block-level HTML blocks into hash entries
139 text = _HashHTMLBlocks(text);
141 // Strip link definitions, store in hashes.
142 text = _StripLinkDefinitions(text);
144 text = _RunBlockGamut(text);
146 text = _UnescapeSpecialChars(text);
148 // attacklab: Restore dollar signs
149 text = text.replace(/~D/g,"$$");
151 // attacklab: Restore tildes
152 text = text.replace(/~T/g,"~");
158 var _StripLinkDefinitions = function(text) {
160 // Strips link definitions from text, stores the URLs and titles in
164 // Link defs are in the form: ^[id]: url "optional title"
167 var text = text.replace(/
168 ^[ ]{0,3}\[(.+)\]: // id = $1 attacklab: g_tab_width - 1
170 \n? // maybe *one* newline
172 <?(\S+?)>? // url = $2
174 \n? // maybe one newline
177 (\n*) // any lines skipped = $3 attacklab: lookbehind removed
182 )? // title is optional
187 text = text.replace(/^[ ]{0,3}\[(.+)\]:[ \t]*\n?[ \t]*<?(\S+?)>?[ \t]*\n?[ \t]*(?:(\n*)["(](.+?)[")][ \t]*)?(?:\n+|\Z)/gm,
188 function (wholeMatch,m1,m2,m3,m4) {
189 m1 = m1.toLowerCase();
190 g_urls[m1] = _EncodeAmpsAndAngles(m2); // Link IDs are case-insensitive
192 // Oops, found blank lines, so it's not a title.
193 // Put back the parenthetical statement we stole.
196 g_titles[m1] = m4.replace(/"/g,""");
199 // Completely remove the definition from the text
208 var _HashHTMLBlocks = function(text) {
209 // attacklab: Double up blank lines to reduce lookaround
210 text = text.replace(/\n/g,"\n\n");
212 // Hashify HTML blocks:
213 // We only want to do this for block-level HTML tags, such as headers,
214 // lists, and tables. That's because we still want to wrap <p>s around
215 // "paragraphs" that are wrapped in non-block-level tags, such as anchors,
216 // phrase emphasis, and spans. The list of tags we're looking for is
218 var block_tags_a = "p|div|h[1-6]|blockquote|pre|table|dl|ol|ul|script|noscript|form|fieldset|iframe|math|ins|del"
219 var block_tags_b = "p|div|h[1-6]|blockquote|pre|table|dl|ol|ul|script|noscript|form|fieldset|iframe|math"
221 // First, look for nested blocks, e.g.:
224 // tags for inner block must be indented.
228 // The outermost tags must start at the left margin for this to match, and
229 // the inner nested divs must be indented.
230 // We need to do this before the next, more liberal match, because the next
231 // match will start at the first `<div>` and stop at the first `</div>`.
233 // attacklab: This regex can be expensive when it fails.
235 var text = text.replace(/
237 ^ // start of line (with /m)
238 <($block_tags_a) // start tag = $2
240 // attacklab: hack around khtml/pcre bug...
241 [^\r]*?\n // any number of lines, minimally matching
242 </\2> // the matching end tag
243 [ \t]* // trailing spaces/tabs
244 (?=\n+) // followed by a newline
245 ) // attacklab: there are sentinel newlines at end of document
246 /gm,function(){...}};
248 text = text.replace(/^(<(p|div|h[1-6]|blockquote|pre|table|dl|ol|ul|script|noscript|form|fieldset|iframe|math|ins|del)\b[^\r]*?\n<\/\2>[ \t]*(?=\n+))/gm,hashElement);
251 // Now match more liberally, simply from `\n<tag>` to `</tag>\n`
255 var text = text.replace(/
257 ^ // start of line (with /m)
258 <($block_tags_b) // start tag = $2
260 // attacklab: hack around khtml/pcre bug...
261 [^\r]*? // any number of lines, minimally matching
262 .*</\2> // the matching end tag
263 [ \t]* // trailing spaces/tabs
264 (?=\n+) // followed by a newline
265 ) // attacklab: there are sentinel newlines at end of document
266 /gm,function(){...}};
268 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);
270 // Special case just for <hr />. It was easier to make a special case than
271 // to make the other regex more complicated.
274 text = text.replace(/
276 \n\n // Starting after a blank line
278 (<(hr) // start tag = $2
281 \/?>) // the matching end tag
283 (?=\n{2,}) // followed by a blank line
287 text = text.replace(/(\n[ ]{0,3}(<(hr)\b([^<>])*?\/?>)[ \t]*(?=\n{2,}))/g,hashElement);
289 // Special case for standalone HTML comments:
292 text = text.replace(/
294 \n\n // Starting after a blank line
295 [ ]{0,3} // attacklab: g_tab_width - 1
300 (?=\n{2,}) // followed by a blank line
304 text = text.replace(/(\n\n[ ]{0,3}<!(--[^\r]*?--\s*)+>[ \t]*(?=\n{2,}))/g,hashElement);
306 // PHP and ASP-style processor instructions (<?...?> and <%...%>)
309 text = text.replace(/
311 \n\n // Starting after a blank line
314 [ ]{0,3} // attacklab: g_tab_width - 1
321 (?=\n{2,}) // followed by a blank line
325 text = text.replace(/(?:\n\n)([ ]{0,3}(?:<([?%])[^\r]*?\2>)[ \t]*(?=\n{2,}))/g,hashElement);
327 // attacklab: Undo double lines (see comment at top of this function)
328 text = text.replace(/\n\n/g,"\n");
332 var hashElement = function(wholeMatch,m1) {
336 blockText = blockText.replace(/\n\n/g,"\n");
337 blockText = blockText.replace(/^\n/,"");
339 // strip trailing blank lines
340 blockText = blockText.replace(/\n+$/g,"");
342 // Replace the element text with a marker ("~KxK" where x is its key)
343 blockText = "\n\n~K" + (g_html_blocks.push(blockText)-1) + "K\n\n";
348 var _RunBlockGamut = function(text) {
350 // These are all the transformations that form block-level
351 // tags like paragraphs, headers, and list items.
353 text = _DoHeaders(text);
355 // Do Horizontal Rules:
356 var key = hashBlock("<hr />");
357 text = text.replace(/^[ ]{0,2}([ ]?\*[ ]?){3,}[ \t]*$/gm,key);
358 text = text.replace(/^[ ]{0,2}([ ]?\-[ ]?){3,}[ \t]*$/gm,key);
359 text = text.replace(/^[ ]{0,2}([ ]?\_[ ]?){3,}[ \t]*$/gm,key);
361 text = _DoLists(text);
362 text = _DoCodeBlocks(text);
363 text = _DoBlockQuotes(text);
365 // We already ran _HashHTMLBlocks() before, in Markdown(), but that
366 // was to escape raw HTML in the original Markdown source. This time,
367 // we're escaping the markup we've just created, so that we don't wrap
368 // <p> tags around block-level tags.
369 text = _HashHTMLBlocks(text);
370 text = _FormParagraphs(text);
376 var _RunSpanGamut = function(text) {
378 // These are all the transformations that occur *within* block-level
379 // tags like paragraphs, headers, and list items.
382 text = _DoCodeSpans(text);
383 text = _EscapeSpecialCharsWithinTagAttributes(text);
384 text = _EncodeBackslashEscapes(text);
386 // Process anchor and image tags. Images must come first,
387 // because ![foo][f] looks like an anchor.
388 text = _DoImages(text);
389 text = _DoAnchors(text);
391 // Make links out of things like `<http://example.com/>`
392 // Must come after _DoAnchors(), because you can use < and >
393 // delimiters in inline links like [this](<url>).
394 text = _DoAutoLinks(text);
395 text = _EncodeAmpsAndAngles(text);
396 text = _DoItalicsAndBold(text);
399 text = text.replace(/ +\n/g," <br />\n");
404 var _EscapeSpecialCharsWithinTagAttributes = function(text) {
406 // Within tags -- meaning between < and > -- encode [\ ` * _] so they
407 // don't conflict with their use in Markdown for code, italics and strong.
410 // Build a regex to find HTML tags and comments. See Friedl's
411 // "Mastering Regular Expressions", 2nd Ed., pp. 200-201.
412 var regex = /(<[a-z\/!$]("[^"]*"|'[^']*'|[^'">])*>|<!(--.*?--\s*)+>)/gi;
414 text = text.replace(regex, function(wholeMatch) {
415 var tag = wholeMatch.replace(/(.)<\/?code>(?=.)/g,"$1`");
416 tag = escapeCharacters(tag,"\\`*_");
423 var _DoAnchors = function(text) {
425 // Turn Markdown link shortcuts into XHTML <a> tags.
428 // First, handle reference-style links: [link text] [id]
432 text = text.replace(/
433 ( // wrap whole match in $1
437 \[[^\]]*\] // allow brackets nested one level
439 [^\[] // or anything else
444 [ ]? // one optional space
445 (?:\n[ ]*)? // one optional newline followed by spaces
450 )()()()() // pad remaining backreferences
451 /g,_DoAnchors_callback);
453 text = text.replace(/(\[((?:\[[^\]]*\]|[^\[\]])*)\][ ]?(?:\n[ ]*)?\[(.*?)\])()()()()/g,writeAnchorTag);
456 // Next, inline-style links: [link text](url "optional title")
460 text = text.replace(/
461 ( // wrap whole match in $1
465 \[[^\]]*\] // allow brackets nested one level
467 [^\[\]] // or anything else
473 () // no id, so leave $3 empty
474 <?(.*?)>? // href = $4
477 (['"]) // quote char = $6
480 [ \t]* // ignore any spaces/tabs between closing quote and )
481 )? // title is optional
486 text = text.replace(/(\[((?:\[[^\]]*\]|[^\[\]])*)\]\([ \t]*()<?(.*?)>?[ \t]*((['"])(.*?)\6[ \t]*)?\))/g,writeAnchorTag);
489 // Last, handle reference-style shortcuts: [link text]
490 // These must come last in case you've also got [link test][1]
491 // or [link test](/foo)
495 text = text.replace(/
496 ( // wrap whole match in $1
498 ([^\[\]]+) // link text = $2; can't contain '[' or ']'
500 )()()()()() // pad rest of backreferences
503 text = text.replace(/(\[([^\[\]]+)\])()()()()()/g, writeAnchorTag);
508 var writeAnchorTag = function(wholeMatch,m1,m2,m3,m4,m5,m6,m7) {
509 if (m7 == undefined) m7 = "";
510 var whole_match = m1;
512 var link_id = m3.toLowerCase();
518 // lower-case and turn embedded newlines into spaces
519 link_id = link_text.toLowerCase().replace(/ ?\n/g," ");
523 if (g_urls[link_id] != undefined) {
524 url = g_urls[link_id];
525 if (g_titles[link_id] != undefined) {
526 title = g_titles[link_id];
530 if (whole_match.search(/\(\s*\)$/m)>-1) {
531 // Special case for explicit empty url
539 url = escapeCharacters(url,"*_");
540 var result = "<a href=\"" + url + "\"";
543 title = title.replace(/"/g,""");
544 title = escapeCharacters(title,"*_");
545 result += " title=\"" + title + "\"";
548 result += ">" + link_text + "</a>";
554 var _DoImages = function(text) {
556 // Turn Markdown image shortcuts into <img> tags.
560 // First, handle reference-style labeled images: ![alt text][id]
564 text = text.replace(/
565 ( // wrap whole match in $1
567 (.*?) // alt text = $2
570 [ ]? // one optional space
571 (?:\n[ ]*)? // one optional newline followed by spaces
576 )()()()() // pad rest of backreferences
579 text = text.replace(/(!\[(.*?)\][ ]?(?:\n[ ]*)?\[(.*?)\])()()()()/g,writeImageTag);
582 // Next, handle inline images: ![alt text](url "optional title")
583 // Don't forget: encode * and _
586 text = text.replace(/
587 ( // wrap whole match in $1
589 (.*?) // alt text = $2
591 \s? // One optional whitespace character
594 () // no id, so leave $3 empty
595 <?(\S+?)>? // src url = $4
598 (['"]) // quote char = $6
602 )? // title is optional
607 text = text.replace(/(!\[(.*?)\]\s?\([ \t]*()<?(\S+?)>?[ \t]*((['"])(.*?)\6[ \t]*)?\))/g,writeImageTag);
612 var writeImageTag = function(wholeMatch,m1,m2,m3,m4,m5,m6,m7) {
613 var whole_match = m1;
615 var link_id = m3.toLowerCase();
619 if (!title) title = "";
623 // lower-case and turn embedded newlines into spaces
624 link_id = alt_text.toLowerCase().replace(/ ?\n/g," ");
628 if (g_urls[link_id] != undefined) {
629 url = g_urls[link_id];
630 if (g_titles[link_id] != undefined) {
631 title = g_titles[link_id];
639 alt_text = alt_text.replace(/"/g,""");
640 url = escapeCharacters(url,"*_");
641 var result = "<img src=\"" + url + "\" alt=\"" + alt_text + "\"";
643 // attacklab: Markdown.pl adds empty title attributes to images.
644 // Replicate this bug.
647 title = title.replace(/"/g,""");
648 title = escapeCharacters(title,"*_");
649 result += " title=\"" + title + "\"";
658 var _DoHeaders = function(text) {
660 // Setext-style headers:
667 text = text.replace(/^(.+)[ \t]*\n=+[ \t]*\n+/gm,
668 function(wholeMatch,m1){return hashBlock('<h1 id="' + headerId(m1) + '">' + _RunSpanGamut(m1) + "</h1>");});
670 text = text.replace(/^(.+)[ \t]*\n-+[ \t]*\n+/gm,
671 function(matchFound,m1){return hashBlock('<h2 id="' + headerId(m1) + '">' + _RunSpanGamut(m1) + "</h2>");});
673 // atx-style headers:
676 // ## Header 2 with closing hashes ##
682 text = text.replace(/
683 ^(\#{1,6}) // $1 = string of #'s
685 (.+?) // $2 = Header text
687 \#* // optional closing #'s (not counted)
689 /gm, function() {...});
692 text = text.replace(/^(\#{1,6})[ \t]*(.+?)[ \t]*\#*\n+/gm,
693 function(wholeMatch,m1,m2) {
694 var h_level = m1.length;
695 return hashBlock("<h" + h_level + ' id="' + headerId(m2) + '">' + _RunSpanGamut(m2) + "</h" + h_level + ">");
698 function headerId(m) {
699 return m.replace(/[^\w]/g, '').toLowerCase();
704 // This declaration keeps Dojo compressor from outputting garbage:
705 var _ProcessListItems;
707 var _DoLists = function(text) {
709 // Form HTML ordered (numbered) and unordered (bulleted) lists.
712 // attacklab: add sentinel to hack around khtml/safari bug:
713 // http://bugs.webkit.org/show_bug.cgi?id=11231
716 // Re-usable pattern to match any entirel ul or ol list:
722 [ ]{0,3} // attacklab: g_tab_width - 1
723 ([*+-]|\d+[.]) // $3 = first list item marker
728 ~0 // sentinel for workaround; should be $
732 (?! // Negative lookahead for another list item marker
734 (?:[*+-]|\d+[.])[ \t]+
739 var whole_list = /^(([ ]{0,3}([*+-]|\d+[.])[ \t]+)[^\r]+?(~0|\n{2,}(?=\S)(?![ \t]*(?:[*+-]|\d+[.])[ \t]+)))/gm;
742 text = text.replace(whole_list,function(wholeMatch,m1,m2) {
744 var list_type = (m2.search(/[*+-]/g)>-1) ? "ul" : "ol";
746 // Turn double returns into triple returns, so that we can make a
747 // paragraph for the last item in a list, if necessary:
748 list = list.replace(/\n{2,}/g,"\n\n\n");;
749 var result = _ProcessListItems(list);
751 // Trim any trailing whitespace, to put the closing `</$list_type>`
752 // up on the preceding line, to get it past the current stupid
753 // HTML block parser. This is a hack to work around the terrible
754 // hack that is the HTML block parser.
755 result = result.replace(/\s+$/,"");
756 result = "<"+list_type+">" + result + "</"+list_type+">\n";
760 whole_list = /(\n\n|^\n?)(([ ]{0,3}([*+-]|\d+[.])[ \t]+)[^\r]+?(~0|\n{2,}(?=\S)(?![ \t]*(?:[*+-]|\d+[.])[ \t]+)))/g;
761 text = text.replace(whole_list,function(wholeMatch,m1,m2,m3) {
765 var list_type = (m3.search(/[*+-]/g)>-1) ? "ul" : "ol";
766 // Turn double returns into triple returns, so that we can make a
767 // paragraph for the last item in a list, if necessary:
768 var list = list.replace(/\n{2,}/g,"\n\n\n");;
769 var result = _ProcessListItems(list);
770 result = runup + "<"+list_type+">\n" + result + "</"+list_type+">\n";
775 // attacklab: strip sentinel
776 text = text.replace(/~0/,"");
781 _ProcessListItems = function(list_str) {
783 // Process the contents of a single ordered or unordered list, splitting it
784 // into individual list items.
786 // The $g_list_level global keeps track of when we're inside a list.
787 // Each time we enter a list, we increment it; when we leave a list,
788 // we decrement. If it's zero, we're not in a list anymore.
790 // We do this because when we're not inside a list, we want to treat
791 // something like this:
793 // I recommend upgrading to version
794 // 8. Oops, now this line is treated
797 // As a single paragraph, despite the fact that the second line starts
798 // with a digit-period-space sequence.
800 // Whereas when we're inside a list (or sub-list), that line will be
801 // treated as the start of a sub-list. What a kludge, huh? This is
802 // an aspect of Markdown's syntax that's hard to parse perfectly
803 // without resorting to mind-reading. Perhaps the solution is to
804 // change the syntax rules such that sub-lists must start with a
805 // starting cardinal number; e.g. "1." or "a.".
809 // trim trailing blank lines:
810 list_str = list_str.replace(/\n{2,}$/,"\n");
812 // attacklab: add sentinel to emulate \z
816 list_str = list_str.replace(/
817 (\n)? // leading line = $1
818 (^[ \t]*) // leading whitespace = $2
819 ([*+-]|\d+[.]) [ \t]+ // list marker = $3
820 ([^\r]+? // list item text = $4
822 (?= \n* (~0 | \2 ([*+-]|\d+[.]) [ \t]+))
823 /gm, function(){...});
825 list_str = list_str.replace(/(\n)?(^[ \t]*)([*+-]|\d+[.])[ \t]+([^\r]+?(\n{1,2}))(?=\n*(~0|\2([*+-]|\d+[.])[ \t]+))/gm,
826 function(wholeMatch,m1,m2,m3,m4){
828 var leading_line = m1;
829 var leading_space = m2;
831 if (leading_line || (item.search(/\n{2,}/)>-1)) {
832 item = _RunBlockGamut(_Outdent(item));
835 // Recursion for sub-lists:
836 item = _DoLists(_Outdent(item));
837 item = item.replace(/\n$/,""); // chomp(item)
838 item = _RunSpanGamut(item);
841 return "<li>" + item + "</li>\n";
845 // attacklab: strip sentinel
846 list_str = list_str.replace(/~0/g,"");
853 var _DoCodeBlocks = function(text) {
855 // Process Markdown `<pre><code>` blocks.
859 text = text.replace(text,
861 ( // $1 = the code block -- one or more lines, starting with a space/tab
863 (?:[ ]{4}|\t) // Lines must start with a tab or a tab-width of spaces - attacklab: g_tab_width
867 (\n*[ ]{0,3}[^ \t\n]|(?=~0)) // attacklab: g_tab_width
871 // attacklab: sentinel workarounds for lack of \A and \Z, safari\khtml bug
874 text = text.replace(/(?:\n\n|^)((?:(?:[ ]{4}|\t).*\n+)+)(\n*[ ]{0,3}[^ \t\n]|(?=~0))/g,
875 function(wholeMatch,m1,m2) {
879 codeblock = _EncodeCode( _Outdent(codeblock));
880 codeblock = _Detab(codeblock);
881 codeblock = codeblock.replace(/^\n+/g,""); // trim leading newlines
882 codeblock = codeblock.replace(/\n+$/g,""); // trim trailing whitespace
884 codeblock = "<pre><code>" + codeblock + "\n</code></pre>";
886 return hashBlock(codeblock) + nextChar;
890 // attacklab: strip sentinel
891 text = text.replace(/~0/,"");
896 var hashBlock = function(text) {
897 text = text.replace(/(^\n+|\n+$)/g,"");
898 return "\n\n~K" + (g_html_blocks.push(text)-1) + "K\n\n";
902 var _DoCodeSpans = function(text) {
904 // * Backtick quotes are used for <code></code> spans.
906 // * You can use multiple backticks as the delimiters if you want to
907 // include literal backticks in the code span. So, this input:
909 // Just type ``foo `bar` baz`` at the prompt.
911 // Will translate to:
913 // <p>Just type <code>foo `bar` baz</code> at the prompt.</p>
915 // There's no arbitrary limit to the number of backticks you
916 // can use as delimters. If you need three consecutive backticks
917 // in your code, use four for delimiters, etc.
919 // * You can use spaces to get literal backticks at the edges:
921 // ... type `` `bar` `` ...
925 // ... type <code>`bar`</code> ...
929 text = text.replace(/
930 (^|[^\\]) // Character before opening ` can't be a backslash
931 (`+) // $2 = Opening run of `
932 ( // $3 = The code block
934 [^`] // attacklab: work around lack of lookbehind
936 \2 // Matching closer
938 /gm, function(){...});
941 text = text.replace(/(^|[^\\])(`+)([^\r]*?[^`])\2(?!`)/gm,
942 function(wholeMatch,m1,m2,m3,m4) {
944 c = c.replace(/^([ \t]*)/g,""); // leading whitespace
945 c = c.replace(/[ \t]*$/g,""); // trailing whitespace
947 return m1+"<code>"+c+"</code>";
954 var _EncodeCode = function(text) {
956 // Encode/escape certain characters inside Markdown code runs.
957 // The point is that in code, these characters are literals,
958 // and lose their special Markdown meanings.
960 // Encode all ampersands; HTML entities are not
961 // entities within a Markdown code span.
962 text = text.replace(/&/g,"&");
964 // Do the angle bracket song and dance:
965 text = text.replace(/</g,"<");
966 text = text.replace(/>/g,">");
968 // Now, escape characters that are magic in Markdown:
969 text = escapeCharacters(text,"\*_{}[]\\",false);
971 // jj the line above breaks this:
985 var _DoItalicsAndBold = function(text) {
987 // <strong> must go first:
988 text = text.replace(/(\*\*|__)(?=\S)([^\r]*?\S[*_]*)\1/g,
989 "<strong>$2</strong>");
991 text = text.replace(/(\*|_)(?=\S)([^\r]*?\S)\1/g,
998 var _DoBlockQuotes = function(text) {
1001 text = text.replace(/
1002 ( // Wrap whole match in $1
1004 ^[ \t]*>[ \t]? // '>' at the start of a line
1005 .+\n // rest of the first line
1006 (.+\n)* // subsequent consecutive lines
1010 /gm, function(){...});
1013 text = text.replace(/((^[ \t]*>[ \t]?.+\n(.+\n)*\n*)+)/gm,
1014 function(wholeMatch,m1) {
1017 // attacklab: hack around Konqueror 3.5.4 bug:
1018 // "----------bug".replace(/^-/g,"") == "bug"
1020 bq = bq.replace(/^[ \t]*>[ \t]?/gm,"~0"); // trim one level of quoting
1022 // attacklab: clean up hack
1023 bq = bq.replace(/~0/g,"");
1025 bq = bq.replace(/^[ \t]+$/gm,""); // trim whitespace-only lines
1026 bq = _RunBlockGamut(bq); // recurse
1028 bq = bq.replace(/(^|\n)/g,"$1 ");
1029 // These leading spaces screw with <pre> content, so we need to fix that:
1031 /(\s*<pre>[^\r]+?<\/pre>)/gm,
1032 function(wholeMatch,m1) {
1034 // attacklab: hack around Konqueror 3.5.4 bug:
1035 pre = pre.replace(/^ /mg,"~0");
1036 pre = pre.replace(/~0/g,"");
1040 return hashBlock("<blockquote>\n" + bq + "\n</blockquote>");
1046 var _FormParagraphs = function(text) {
1049 // $text - string to process with html <p> tags
1052 // Strip leading and trailing lines:
1053 text = text.replace(/^\n+/g,"");
1054 text = text.replace(/\n+$/g,"");
1056 var grafs = text.split(/\n{2,}/g);
1057 var grafsOut = new Array();
1062 var end = grafs.length;
1063 for (var i=0; i<end; i++) {
1066 // if this is an HTML marker, copy it
1067 if (str.search(/~K(\d+)K/g) >= 0) {
1070 else if (str.search(/\S/) >= 0) {
1071 str = _RunSpanGamut(str);
1072 str = str.replace(/^([ \t]*)/g,"<p>");
1080 // Unhashify HTML blocks
1082 end = grafsOut.length;
1083 for (var i=0; i<end; i++) {
1084 // if this is a marker for an html block...
1085 while (grafsOut[i].search(/~K(\d+)K/) >= 0) {
1086 var blockText = g_html_blocks[RegExp.$1];
1087 blockText = blockText.replace(/\$/g,"$$$$"); // Escape any dollar signs
1088 grafsOut[i] = grafsOut[i].replace(/~K\d+K/,blockText);
1092 return grafsOut.join("\n\n");
1096 var _EncodeAmpsAndAngles = function(text) {
1097 // Smart processing for ampersands and angle brackets that need to be encoded.
1099 // Ampersand-encoding based entirely on Nat Irons's Amputator MT plugin:
1100 // http://bumppo.net/projects/amputator/
1101 text = text.replace(/&(?!#?[xX]?(?:[0-9a-fA-F]+|\w+);)/g,"&");
1104 text = text.replace(/<(?![a-z\/?\$!])/gi,"<");
1110 var _EncodeBackslashEscapes = function(text) {
1112 // Parameter: String.
1113 // Returns: The string, with after processing the following backslash
1114 // escape sequences.
1117 // attacklab: The polite way to do this is with the new
1118 // escapeCharacters() function:
1120 // text = escapeCharacters(text,"\\",true);
1121 // text = escapeCharacters(text,"`*_{}[]()>#+-.!",true);
1123 // ...but we're sidestepping its use of the (slow) RegExp constructor
1124 // as an optimization for Firefox. This function gets called a LOT.
1126 text = text.replace(/\\(\\)/g,escapeCharacters_callback);
1127 text = text.replace(/\\([`*_{}\[\]()>#+-.!])/g,escapeCharacters_callback);
1132 var _DoAutoLinks = function(text) {
1134 text = text.replace(/<((https?|ftp|dict):[^'">\s]+)>/gi,"<a href=\"$1\">$1</a>");
1136 // Email addresses: <address@domain.foo>
1139 text = text.replace(/
1145 [-a-z0-9]+(\.[-a-z0-9]+)*\.[a-z]+
1148 /gi, _DoAutoLinks_callback());
1150 text = text.replace(/<(?:mailto:)?([-.\w]+\@[-a-z0-9]+(\.[-a-z0-9]+)*\.[a-z]+)>/gi,
1151 function(wholeMatch,m1) {
1152 return _EncodeEmailAddress( _UnescapeSpecialChars(m1) );
1160 var _EncodeEmailAddress = function(addr) {
1162 // Input: an email address, e.g. "foo@example.com"
1164 // Output: the email address as a mailto link, with each character
1165 // of the address encoded as either a decimal or hex entity, in
1166 // the hopes of foiling most address harvesting spam bots. E.g.:
1168 // <a href="mailto:foo@e
1169 // xample.com">foo
1170 // @example.com</a>
1172 // Based on a filter by Matthew Wickline, posted to the BBEdit-Talk
1173 // mailing list: <http://tinyurl.com/yu7ue>
1176 // attacklab: why can't javascript speak hex?
1177 function char2hex(ch) {
1178 var hexDigits = '0123456789ABCDEF';
1179 var dec = ch.charCodeAt(0);
1180 return(hexDigits.charAt(dec>>4) + hexDigits.charAt(dec&15));
1184 function(ch){return "&#"+ch.charCodeAt(0)+";";},
1185 function(ch){return "&#x"+char2hex(ch)+";";},
1186 function(ch){return ch;}
1189 addr = "mailto:" + addr;
1191 addr = addr.replace(/./g, function(ch) {
1193 // this *must* be encoded. I insist.
1194 ch = encode[Math.floor(Math.random()*2)](ch);
1195 } else if (ch !=":") {
1196 // leave ':' alone (to spot mailto: later)
1197 var r = Math.random();
1198 // roughly 10% raw, 45% hex, 45% dec
1200 r > .9 ? encode[2](ch) :
1201 r > .45 ? encode[1](ch) :
1208 addr = "<a href=\"" + addr + "\">" + addr + "</a>";
1209 addr = addr.replace(/">.+:/g,"\">"); // strip the mailto: from the visible part
1215 var _UnescapeSpecialChars = function(text) {
1217 // Swap back in all the special characters we've hidden.
1219 text = text.replace(/~E(\d+)E/g,
1220 function(wholeMatch,m1) {
1221 var charCodeToReplace = parseInt(m1);
1222 return String.fromCharCode(charCodeToReplace);
1229 var _Outdent = function(text) {
1231 // Remove one level of line-leading tabs or spaces
1234 // attacklab: hack around Konqueror 3.5.4 bug:
1235 // "----------bug".replace(/^-/g,"") == "bug"
1237 text = text.replace(/^(\t|[ ]{1,4})/gm,"~0"); // attacklab: g_tab_width
1239 // attacklab: clean up hack
1240 text = text.replace(/~0/g,"")
1245 var _Detab = function(text) {
1246 // attacklab: Detab's completely rewritten for speed.
1247 // In perl we could fix it by anchoring the regexp with \G.
1248 // In javascript we're less fortunate.
1250 // expand first n-1 tabs
1251 text = text.replace(/\t(?=\t)/g," "); // attacklab: g_tab_width
1253 // replace the nth with two sentinels
1254 text = text.replace(/\t/g,"~A~B");
1256 // use the sentinel to anchor our regex so it doesn't explode
1257 text = text.replace(/~B(.+?)~A/g,
1258 function(wholeMatch,m1,m2) {
1259 var leadingText = m1;
1260 var numSpaces = 4 - leadingText.length % 4; // attacklab: g_tab_width
1262 // there *must* be a better way to do this:
1263 for (var i=0; i<numSpaces; i++) leadingText+=" ";
1269 // clean up sentinels
1270 text = text.replace(/~A/g," "); // attacklab: g_tab_width
1271 text = text.replace(/~B/g,"");
1278 // attacklab: Utility functions
1282 var escapeCharacters = function(text, charsToEscape, afterBackslash) {
1283 // First we have to escape the escape characters so that
1284 // we can build a character class out of them
1285 var regexString = "([" + charsToEscape.replace(/([\[\]\\])/g,"\\$1") + "])";
1287 if (afterBackslash) {
1288 regexString = "\\\\" + regexString;
1291 var regex = new RegExp(regexString,"g");
1292 text = text.replace(regex,escapeCharacters_callback);
1298 var escapeCharacters_callback = function(wholeMatch,m1) {
1299 var charCodeToEscape = m1.charCodeAt(0);
1300 return "~E"+charCodeToEscape+"E";
1303 } // end of Showdown.converter
1306 //if (typeof exports != 'undefined') exports.Showdown = Showdown;