Mercurial > wikked

comparison static/js/pagedown/Markdown.Converter.js @ 60:8250c977bc50

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Moved static files to the root directory.
author Ludovic Chabant <ludovic@chabant.com>
date Tue, 05 Feb 2013 14:49:34 -0800
parents wikked/static/js/pagedown/Markdown.Converter.js@59cad6ce1a1c
children
comparison
equal deleted inserted replaced
59:59ecc742ab8e 60:8250c977bc50
1 var Markdown;
2
3 if (typeof exports === "object" && typeof require === "function") // we're in a CommonJS (e.g. Node.js) module
4 Markdown = exports;
5 else
6 Markdown = {};
7
8 // The following text is included for historical reasons, but should
9 // be taken with a pinch of salt; it's not all true anymore.
10
11 //
12 // Wherever possible, Showdown is a straight, line-by-line port
13 // of the Perl version of Markdown.
14 //
15 // This is not a normal parser design; it's basically just a
16 // series of string substitutions. It's hard to read and
17 // maintain this way, but keeping Showdown close to the original
18 // design makes it easier to port new features.
19 //
20 // More importantly, Showdown behaves like markdown.pl in most
21 // edge cases. So web applications can do client-side preview
22 // in Javascript, and then build identical HTML on the server.
23 //
24 // This port needs the new RegExp functionality of ECMA 262,
25 // 3rd Edition (i.e. Javascript 1.5). Most modern web browsers
26 // should do fine. Even with the new regular expression features,
27 // We do a lot of work to emulate Perl's regex functionality.
28 // The tricky changes in this file mostly have the "attacklab:"
29 // label. Major or self-explanatory changes don't.
30 //
31 // Smart diff tools like Araxis Merge will be able to match up
32 // this file with markdown.pl in a useful way. A little tweaking
33 // helps: in a copy of markdown.pl, replace "#" with "//" and
34 // replace "$text" with "text". Be sure to ignore whitespace
35 // and line endings.
36 //
37
38
39 //
40 // Usage:
41 //
42 // var text = "Markdown *rocks*.";
43 //
44 // var converter = new Markdown.Converter();
45 // var html = converter.makeHtml(text);
46 //
47 // alert(html);
48 //
49 // Note: move the sample code to the bottom of this
50 // file before uncommenting it.
51 //
52
53 (function () {
54
55 function identity(x) { return x; }
56 function returnFalse(x) { return false; }
57
58 function HookCollection() { }
59
60 HookCollection.prototype = {
61
62 chain: function (hookname, func) {
63 var original = this[hookname];
64 if (!original)
65 throw new Error("unknown hook " + hookname);
66
67 if (original === identity)
68 this[hookname] = func;
69 else
70 this[hookname] = function (x) { return func(original(x)); }
71 },
72 set: function (hookname, func) {
73 if (!this[hookname])
74 throw new Error("unknown hook " + hookname);
75 this[hookname] = func;
76 },
77 addNoop: function (hookname) {
78 this[hookname] = identity;
79 },
80 addFalse: function (hookname) {
81 this[hookname] = returnFalse;
82 }
83 };
84
85 Markdown.HookCollection = HookCollection;
86
87 // g_urls and g_titles allow arbitrary user-entered strings as keys. This
88 // caused an exception (and hence stopped the rendering) when the user entered
89 // e.g. [push] or [__proto__]. Adding a prefix to the actual key prevents this
90 // (since no builtin property starts with "s_"). See
91 // http://meta.stackoverflow.com/questions/64655/strange-wmd-bug
92 // (granted, switching from Array() to Object() alone would have left only __proto__
93 // to be a problem)
94 function SaveHash() { }
95 SaveHash.prototype = {
96 set: function (key, value) {
97 this["s_" + key] = value;
98 },
99 get: function (key) {
100 return this["s_" + key];
101 }
102 };
103
104 Markdown.Converter = function () {
105 var pluginHooks = this.hooks = new HookCollection();
106 pluginHooks.addNoop("plainLinkText"); // given a URL that was encountered by itself (without markup), should return the link text that's to be given to this link
107 pluginHooks.addNoop("preConversion"); // called with the orignal text as given to makeHtml. The result of this plugin hook is the actual markdown source that will be cooked
108 pluginHooks.addNoop("postConversion"); // called with the final cooked HTML code. The result of this plugin hook is the actual output of makeHtml
109
110 //
111 // Private state of the converter instance:
112 //
113
114 // Global hashes, used by various utility routines
115 var g_urls;
116 var g_titles;
117 var g_html_blocks;
118
119 // Used to track when we're inside an ordered or unordered list
120 // (see _ProcessListItems() for details):
121 var g_list_level;
122
123 this.makeHtml = function (text) {
124
125 //
126 // Main function. The order in which other subs are called here is
127 // essential. Link and image substitutions need to happen before
128 // _EscapeSpecialCharsWithinTagAttributes(), so that any *'s or _'s in the <a>
129 // and <img> tags get encoded.
130 //
131
132 // This will only happen if makeHtml on the same converter instance is called from a plugin hook.
133 // Don't do that.
134 if (g_urls)
135 throw new Error("Recursive call to converter.makeHtml");
136
137 // Create the private state objects.
138 g_urls = new SaveHash();
139 g_titles = new SaveHash();
140 g_html_blocks = [];
141 g_list_level = 0;
142
143 text = pluginHooks.preConversion(text);
144
145 // attacklab: Replace ~ with ~T
146 // This lets us use tilde as an escape char to avoid md5 hashes
147 // The choice of character is arbitray; anything that isn't
148 // magic in Markdown will work.
149 text = text.replace(/~/g, "~T");
150
151 // attacklab: Replace $ with ~D
152 // RegExp interprets $ as a special character
153 // when it's in a replacement string
154 text = text.replace(/\$/g, "~D");
155
156 // Standardize line endings
157 text = text.replace(/\r\n/g, "\n"); // DOS to Unix
158 text = text.replace(/\r/g, "\n"); // Mac to Unix
159
160 // Make sure text begins and ends with a couple of newlines:
161 text = "\n\n" + text + "\n\n";
162
163 // Convert all tabs to spaces.
164 text = _Detab(text);
165
166 // Strip any lines consisting only of spaces and tabs.
167 // This makes subsequent regexen easier to write, because we can
168 // match consecutive blank lines with /\n+/ instead of something
169 // contorted like /[ \t]*\n+/ .
170 text = text.replace(/^[ \t]+$/mg, "");
171
172 // Turn block-level HTML blocks into hash entries
173 text = _HashHTMLBlocks(text);
174
175 // Strip link definitions, store in hashes.
176 text = _StripLinkDefinitions(text);
177
178 text = _RunBlockGamut(text);
179
180 text = _UnescapeSpecialChars(text);
181
182 // attacklab: Restore dollar signs
183 text = text.replace(/~D/g, "$$");
184
185 // attacklab: Restore tildes
186 text = text.replace(/~T/g, "~");
187
188 text = pluginHooks.postConversion(text);
189
190 g_html_blocks = g_titles = g_urls = null;
191
192 return text;
193 };
194
195 function _StripLinkDefinitions(text) {
196 //
197 // Strips link definitions from text, stores the URLs and titles in
198 // hash references.
199 //
200
201 // Link defs are in the form: ^[id]: url "optional title"
202
203 /*
204 text = text.replace(/
205 ^[ ]{0,3}\[(.+)\]: // id = $1 attacklab: g_tab_width - 1
206 [ \t]*
207 \n? // maybe *one* newline
208 [ \t]*
209 <?(\S+?)>? // url = $2
210 (?=\s|$) // lookahead for whitespace instead of the lookbehind removed below
211 [ \t]*
212 \n? // maybe one newline
213 [ \t]*
214 ( // (potential) title = $3
215 (\n*) // any lines skipped = $4 attacklab: lookbehind removed
216 [ \t]+
217 ["(]
218 (.+?) // title = $5
219 [")]
220 [ \t]*
221 )? // title is optional
222 (?:\n+|$)
223 /gm, function(){...});
224 */
225
226 text = text.replace(/^[ ]{0,3}\[(.+)\]:[ \t]*\n?[ \t]*<?(\S+?)>?(?=\s|$)[ \t]*\n?[ \t]*((\n*)["(](.+?)[")][ \t]*)?(?:\n+)/gm,
227 function (wholeMatch, m1, m2, m3, m4, m5) {
228 m1 = m1.toLowerCase();
229 g_urls.set(m1, _EncodeAmpsAndAngles(m2)); // Link IDs are case-insensitive
230 if (m4) {
231 // Oops, found blank lines, so it's not a title.
232 // Put back the parenthetical statement we stole.
233 return m3;
234 } else if (m5) {
235 g_titles.set(m1, m5.replace(/"/g, "&quot;"));
236 }
237
238 // Completely remove the definition from the text
239 return "";
240 }
241 );
242
243 return text;
244 }
245
246 function _HashHTMLBlocks(text) {
247
248 // Hashify HTML blocks:
249 // We only want to do this for block-level HTML tags, such as headers,
250 // lists, and tables. That's because we still want to wrap <p>s around
251 // "paragraphs" that are wrapped in non-block-level tags, such as anchors,
252 // phrase emphasis, and spans. The list of tags we're looking for is
253 // hard-coded:
254 var block_tags_a = "p|div|h[1-6]|blockquote|pre|table|dl|ol|ul|script|noscript|form|fieldset|iframe|math|ins|del"
255 var block_tags_b = "p|div|h[1-6]|blockquote|pre|table|dl|ol|ul|script|noscript|form|fieldset|iframe|math"
256
257 // First, look for nested blocks, e.g.:
258 // <div>
259 // <div>
260 // tags for inner block must be indented.
261 // </div>
262 // </div>
263 //
264 // The outermost tags must start at the left margin for this to match, and
265 // the inner nested divs must be indented.
266 // We need to do this before the next, more liberal match, because the next
267 // match will start at the first `<div>` and stop at the first `</div>`.
268
269 // attacklab: This regex can be expensive when it fails.
270
271 /*
272 text = text.replace(/
273 ( // save in $1
274 ^ // start of line (with /m)
275 <($block_tags_a) // start tag = $2
276 \b // word break
277 // attacklab: hack around khtml/pcre bug...
278 [^\r]*?\n // 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(){...}};
284 */
285 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);
286
287 //
288 // Now match more liberally, simply from `\n<tag>` to `</tag>\n`
289 //
290
291 /*
292 text = text.replace(/
293 ( // save in $1
294 ^ // start of line (with /m)
295 <($block_tags_b) // start tag = $2
296 \b // word break
297 // attacklab: hack around khtml/pcre bug...
298 [^\r]*? // any number of lines, minimally matching
299 .*</\2> // the matching end tag
300 [ \t]* // trailing spaces/tabs
301 (?=\n+) // followed by a newline
302 ) // attacklab: there are sentinel newlines at end of document
303 /gm,function(){...}};
304 */
305 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);
306
307 // Special case just for <hr />. It was easier to make a special case than
308 // to make the other regex more complicated.
309
310 /*
311 text = text.replace(/
312 \n // Starting after a blank line
313 [ ]{0,3}
314 ( // save in $1
315 (<(hr) // start tag = $2
316 \b // word break
317 ([^<>])*?
318 \/?>) // the matching end tag
319 [ \t]*
320 (?=\n{2,}) // followed by a blank line
321 )
322 /g,hashElement);
323 */
324 text = text.replace(/\n[ ]{0,3}((<(hr)\b([^<>])*?\/?>)[ \t]*(?=\n{2,}))/g, hashElement);
325
326 // Special case for standalone HTML comments:
327
328 /*
329 text = text.replace(/
330 \n\n // Starting after a blank line
331 [ ]{0,3} // attacklab: g_tab_width - 1
332 ( // save in $1
333 <!
334 (--(?:|(?:[^>-]|-[^>])(?:[^-]|-[^-])*)--) // see http://www.w3.org/TR/html-markup/syntax.html#comments and http://meta.stackoverflow.com/q/95256
335 >
336 [ \t]*
337 (?=\n{2,}) // followed by a blank line
338 )
339 /g,hashElement);
340 */
341 text = text.replace(/\n\n[ ]{0,3}(<!(--(?:|(?:[^>-]|-[^>])(?:[^-]|-[^-])*)--)>[ \t]*(?=\n{2,}))/g, hashElement);
342
343 // PHP and ASP-style processor instructions (<?...?> and <%...%>)
344
345 /*
346 text = text.replace(/
347 (?:
348 \n\n // Starting after a blank line
349 )
350 ( // save in $1
351 [ ]{0,3} // attacklab: g_tab_width - 1
352 (?:
353 <([?%]) // $2
354 [^\r]*?
355 \2>
356 )
357 [ \t]*
358 (?=\n{2,}) // followed by a blank line
359 )
360 /g,hashElement);
361 */
362 text = text.replace(/(?:\n\n)([ ]{0,3}(?:<([?%])[^\r]*?\2>)[ \t]*(?=\n{2,}))/g, hashElement);
363
364 return text;
365 }
366
367 function hashElement(wholeMatch, m1) {
368 var blockText = m1;
369
370 // Undo double lines
371 blockText = blockText.replace(/^\n+/, "");
372
373 // strip trailing blank lines
374 blockText = blockText.replace(/\n+$/g, "");
375
376 // Replace the element text with a marker ("~KxK" where x is its key)
377 blockText = "\n\n~K" + (g_html_blocks.push(blockText) - 1) + "K\n\n";
378
379 return blockText;
380 }
381
382 function _RunBlockGamut(text, doNotUnhash) {
383 //
384 // These are all the transformations that form block-level
385 // tags like paragraphs, headers, and list items.
386 //
387 text = _DoHeaders(text);
388
389 // Do Horizontal Rules:
390 var replacement = "<hr />\n";
391 text = text.replace(/^[ ]{0,2}([ ]?\*[ ]?){3,}[ \t]*$/gm, replacement);
392 text = text.replace(/^[ ]{0,2}([ ]?-[ ]?){3,}[ \t]*$/gm, replacement);
393 text = text.replace(/^[ ]{0,2}([ ]?_[ ]?){3,}[ \t]*$/gm, replacement);
394
395 text = _DoLists(text);
396 text = _DoCodeBlocks(text);
397 text = _DoBlockQuotes(text);
398
399 // We already ran _HashHTMLBlocks() before, in Markdown(), but that
400 // was to escape raw HTML in the original Markdown source. This time,
401 // we're escaping the markup we've just created, so that we don't wrap
402 // <p> tags around block-level tags.
403 text = _HashHTMLBlocks(text);
404 text = _FormParagraphs(text, doNotUnhash);
405
406 return text;
407 }
408
409 function _RunSpanGamut(text) {
410 //
411 // These are all the transformations that occur *within* block-level
412 // tags like paragraphs, headers, and list items.
413 //
414
415 text = _DoCodeSpans(text);
416 text = _EscapeSpecialCharsWithinTagAttributes(text);
417 text = _EncodeBackslashEscapes(text);
418
419 // Process anchor and image tags. Images must come first,
420 // because ![foo][f] looks like an anchor.
421 text = _DoImages(text);
422 text = _DoAnchors(text);
423
424 // Make links out of things like `<http://example.com/>`
425 // Must come after _DoAnchors(), because you can use < and >
426 // delimiters in inline links like [this](<url>).
427 text = _DoAutoLinks(text);
428
429 text = text.replace(/~P/g, "://"); // put in place to prevent autolinking; reset now
430
431 text = _EncodeAmpsAndAngles(text);
432 text = _DoItalicsAndBold(text);
433
434 // Do hard breaks:
435 text = text.replace(/ +\n/g, " <br>\n");
436
437 return text;
438 }
439
440 function _EscapeSpecialCharsWithinTagAttributes(text) {
441 //
442 // Within tags -- meaning between < and > -- encode [\ ` * _] so they
443 // don't conflict with their use in Markdown for code, italics and strong.
444 //
445
446 // Build a regex to find HTML tags and comments. See Friedl's
447 // "Mastering Regular Expressions", 2nd Ed., pp. 200-201.
448
449 // SE: changed the comment part of the regex
450
451 var regex = /(<[a-z\/!$]("[^"]*"|'[^']*'|[^'">])*>|<!(--(?:|(?:[^>-]|-[^>])(?:[^-]|-[^-])*)--)>)/gi;
452
453 text = text.replace(regex, function (wholeMatch) {
454 var tag = wholeMatch.replace(/(.)<\/?code>(?=.)/g, "$1`");
455 tag = escapeCharacters(tag, wholeMatch.charAt(1) == "!" ? "\\`*_/" : "\\`*_"); // also escape slashes in comments to prevent autolinking there -- http://meta.stackoverflow.com/questions/95987
456 return tag;
457 });
458
459 return text;
460 }
461
462 function _DoAnchors(text) {
463 //
464 // Turn Markdown link shortcuts into XHTML <a> tags.
465 //
466 //
467 // First, handle reference-style links: [link text] [id]
468 //
469
470 /*
471 text = text.replace(/
472 ( // wrap whole match in $1
473 \[
474 (
475 (?:
476 \[[^\]]*\] // allow brackets nested one level
477 |
478 [^\[] // or anything else
479 )*
480 )
481 \]
482
483 [ ]? // one optional space
484 (?:\n[ ]*)? // one optional newline followed by spaces
485
486 \[
487 (.*?) // id = $3
488 \]
489 )
490 ()()()() // pad remaining backreferences
491 /g, writeAnchorTag);
492 */
493 text = text.replace(/(\[((?:\[[^\]]*\]|[^\[\]])*)\][ ]?(?:\n[ ]*)?\[(.*?)\])()()()()/g, writeAnchorTag);
494
495 //
496 // Next, inline-style links: [link text](url "optional title")
497 //
498
499 /*
500 text = text.replace(/
501 ( // wrap whole match in $1
502 \[
503 (
504 (?:
505 \[[^\]]*\] // allow brackets nested one level
506 |
507 [^\[\]] // or anything else
508 )*
509 )
510 \]
511 \( // literal paren
512 [ \t]*
513 () // no id, so leave $3 empty
514 <?( // href = $4
515 (?:
516 \([^)]*\) // allow one level of (correctly nested) parens (think MSDN)
517 |
518 [^()\s]
519 )*?
520 )>?
521 [ \t]*
522 ( // $5
523 (['"]) // quote char = $6
524 (.*?) // Title = $7
525 \6 // matching quote
526 [ \t]* // ignore any spaces/tabs between closing quote and )
527 )? // title is optional
528 \)
529 )
530 /g, writeAnchorTag);
531 */
532
533 text = text.replace(/(\[((?:\[[^\]]*\]|[^\[\]])*)\]\([ \t]*()<?((?:\([^)]*\)|[^()\s])*?)>?[ \t]*((['"])(.*?)\6[ \t]*)?\))/g, writeAnchorTag);
534
535 //
536 // Last, handle reference-style shortcuts: [link text]
537 // These must come last in case you've also got [link test][1]
538 // or [link test](/foo)
539 //
540
541 /*
542 text = text.replace(/
543 ( // wrap whole match in $1
544 \[
545 ([^\[\]]+) // link text = $2; can't contain '[' or ']'
546 \]
547 )
548 ()()()()() // pad rest of backreferences
549 /g, writeAnchorTag);
550 */
551 text = text.replace(/(\[([^\[\]]+)\])()()()()()/g, writeAnchorTag);
552
553 return text;
554 }
555
556 function writeAnchorTag(wholeMatch, m1, m2, m3, m4, m5, m6, m7) {
557 if (m7 == undefined) m7 = "";
558 var whole_match = m1;
559 var link_text = m2.replace(/:\/\//g, "~P"); // to prevent auto-linking withing the link. will be converted back after the auto-linker runs
560 var link_id = m3.toLowerCase();
561 var url = m4;
562 var title = m7;
563
564 if (url == "") {
565 if (link_id == "") {
566 // lower-case and turn embedded newlines into spaces
567 link_id = link_text.toLowerCase().replace(/ ?\n/g, " ");
568 }
569 url = "#" + link_id;
570
571 if (g_urls.get(link_id) != undefined) {
572 url = g_urls.get(link_id);
573 if (g_titles.get(link_id) != undefined) {
574 title = g_titles.get(link_id);
575 }
576 }
577 else {
578 if (whole_match.search(/\(\s*\)$/m) > -1) {
579 // Special case for explicit empty url
580 url = "";
581 } else {
582 return whole_match;
583 }
584 }
585 }
586 url = encodeProblemUrlChars(url);
587 url = escapeCharacters(url, "*_");
588 var result = "<a href=\"" + url + "\"";
589
590 if (title != "") {
591 title = attributeEncode(title);
592 title = escapeCharacters(title, "*_");
593 result += " title=\"" + title + "\"";
594 }
595
596 result += ">" + link_text + "</a>";
597
598 return result;
599 }
600
601 function _DoImages(text) {
602 //
603 // Turn Markdown image shortcuts into <img> tags.
604 //
605
606 //
607 // First, handle reference-style labeled images: ![alt text][id]
608 //
609
610 /*
611 text = text.replace(/
612 ( // wrap whole match in $1
613 !\[
614 (.*?) // alt text = $2
615 \]
616
617 [ ]? // one optional space
618 (?:\n[ ]*)? // one optional newline followed by spaces
619
620 \[
621 (.*?) // id = $3
622 \]
623 )
624 ()()()() // pad rest of backreferences
625 /g, writeImageTag);
626 */
627 text = text.replace(/(!\[(.*?)\][ ]?(?:\n[ ]*)?\[(.*?)\])()()()()/g, writeImageTag);
628
629 //
630 // Next, handle inline images: ![alt text](url "optional title")
631 // Don't forget: encode * and _
632
633 /*
634 text = text.replace(/
635 ( // wrap whole match in $1
636 !\[
637 (.*?) // alt text = $2
638 \]
639 \s? // One optional whitespace character
640 \( // literal paren
641 [ \t]*
642 () // no id, so leave $3 empty
643 <?(\S+?)>? // src url = $4
644 [ \t]*
645 ( // $5
646 (['"]) // quote char = $6
647 (.*?) // title = $7
648 \6 // matching quote
649 [ \t]*
650 )? // title is optional
651 \)
652 )
653 /g, writeImageTag);
654 */
655 text = text.replace(/(!\[(.*?)\]\s?\([ \t]*()<?(\S+?)>?[ \t]*((['"])(.*?)\6[ \t]*)?\))/g, writeImageTag);
656
657 return text;
658 }
659
660 function attributeEncode(text) {
661 // unconditionally replace angle brackets here -- what ends up in an attribute (e.g. alt or title)
662 // never makes sense to have verbatim HTML in it (and the sanitizer would totally break it)
663 return text.replace(/>/g, "&gt;").replace(/</g, "&lt;").replace(/"/g, "&quot;");
664 }
665
666 function writeImageTag(wholeMatch, m1, m2, m3, m4, m5, m6, m7) {
667 var whole_match = m1;
668 var alt_text = m2;
669 var link_id = m3.toLowerCase();
670 var url = m4;
671 var title = m7;
672
673 if (!title) title = "";
674
675 if (url == "") {
676 if (link_id == "") {
677 // lower-case and turn embedded newlines into spaces
678 link_id = alt_text.toLowerCase().replace(/ ?\n/g, " ");
679 }
680 url = "#" + link_id;
681
682 if (g_urls.get(link_id) != undefined) {
683 url = g_urls.get(link_id);
684 if (g_titles.get(link_id) != undefined) {
685 title = g_titles.get(link_id);
686 }
687 }
688 else {
689 return whole_match;
690 }
691 }
692
693 alt_text = escapeCharacters(attributeEncode(alt_text), "*_[]()");
694 url = escapeCharacters(url, "*_");
695 var result = "<img src=\"" + url + "\" alt=\"" + alt_text + "\"";
696
697 // attacklab: Markdown.pl adds empty title attributes to images.
698 // Replicate this bug.
699
700 //if (title != "") {
701 title = attributeEncode(title);
702 title = escapeCharacters(title, "*_");
703 result += " title=\"" + title + "\"";
704 //}
705
706 result += " />";
707
708 return result;
709 }
710
711 function _DoHeaders(text) {
712
713 // Setext-style headers:
714 // Header 1
715 // ========
716 //
717 // Header 2
718 // --------
719 //
720 text = text.replace(/^(.+)[ \t]*\n=+[ \t]*\n+/gm,
721 function (wholeMatch, m1) { return "<h1>" + _RunSpanGamut(m1) + "</h1>\n\n"; }
722 );
723
724 text = text.replace(/^(.+)[ \t]*\n-+[ \t]*\n+/gm,
725 function (matchFound, m1) { return "<h2>" + _RunSpanGamut(m1) + "</h2>\n\n"; }
726 );
727
728 // atx-style headers:
729 // # Header 1
730 // ## Header 2
731 // ## Header 2 with closing hashes ##
732 // ...
733 // ###### Header 6
734 //
735
736 /*
737 text = text.replace(/
738 ^(\#{1,6}) // $1 = string of #'s
739 [ \t]*
740 (.+?) // $2 = Header text
741 [ \t]*
742 \#* // optional closing #'s (not counted)
743 \n+
744 /gm, function() {...});
745 */
746
747 text = text.replace(/^(\#{1,6})[ \t]*(.+?)[ \t]*\#*\n+/gm,
748 function (wholeMatch, m1, m2) {
749 var h_level = m1.length;
750 return "<h" + h_level + ">" + _RunSpanGamut(m2) + "</h" + h_level + ">\n\n";
751 }
752 );
753
754 return text;
755 }
756
757 function _DoLists(text) {
758 //
759 // Form HTML ordered (numbered) and unordered (bulleted) lists.
760 //
761
762 // attacklab: add sentinel to hack around khtml/safari bug:
763 // http://bugs.webkit.org/show_bug.cgi?id=11231
764 text += "~0";
765
766 // Re-usable pattern to match any entirel ul or ol list:
767
768 /*
769 var whole_list = /
770 ( // $1 = whole list
771 ( // $2
772 [ ]{0,3} // attacklab: g_tab_width - 1
773 ([*+-]|\d+[.]) // $3 = first list item marker
774 [ \t]+
775 )
776 [^\r]+?
777 ( // $4
778 ~0 // sentinel for workaround; should be $
779 |
780 \n{2,}
781 (?=\S)
782 (?! // Negative lookahead for another list item marker
783 [ \t]*
784 (?:[*+-]|\d+[.])[ \t]+
785 )
786 )
787 )
788 /g
789 */
790 var whole_list = /^(([ ]{0,3}([*+-]|\d+[.])[ \t]+)[^\r]+?(~0|\n{2,}(?=\S)(?![ \t]*(?:[*+-]|\d+[.])[ \t]+)))/gm;
791
792 if (g_list_level) {
793 text = text.replace(whole_list, function (wholeMatch, m1, m2) {
794 var list = m1;
795 var list_type = (m2.search(/[*+-]/g) > -1) ? "ul" : "ol";
796
797 var result = _ProcessListItems(list, list_type);
798
799 // Trim any trailing whitespace, to put the closing `</$list_type>`
800 // up on the preceding line, to get it past the current stupid
801 // HTML block parser. This is a hack to work around the terrible
802 // hack that is the HTML block parser.
803 result = result.replace(/\s+$/, "");
804 result = "<" + list_type + ">" + result + "</" + list_type + ">\n";
805 return result;
806 });
807 } else {
808 whole_list = /(\n\n|^\n?)(([ ]{0,3}([*+-]|\d+[.])[ \t]+)[^\r]+?(~0|\n{2,}(?=\S)(?![ \t]*(?:[*+-]|\d+[.])[ \t]+)))/g;
809 text = text.replace(whole_list, function (wholeMatch, m1, m2, m3) {
810 var runup = m1;
811 var list = m2;
812
813 var list_type = (m3.search(/[*+-]/g) > -1) ? "ul" : "ol";
814 var result = _ProcessListItems(list, list_type);
815 result = runup + "<" + list_type + ">\n" + result + "</" + list_type + ">\n";
816 return result;
817 });
818 }
819
820 // attacklab: strip sentinel
821 text = text.replace(/~0/, "");
822
823 return text;
824 }
825
826 var _listItemMarkers = { ol: "\\d+[.]", ul: "[*+-]" };
827
828 function _ProcessListItems(list_str, list_type) {
829 //
830 // Process the contents of a single ordered or unordered list, splitting it
831 // into individual list items.
832 //
833 // list_type is either "ul" or "ol".
834
835 // The $g_list_level global keeps track of when we're inside a list.
836 // Each time we enter a list, we increment it; when we leave a list,
837 // we decrement. If it's zero, we're not in a list anymore.
838 //
839 // We do this because when we're not inside a list, we want to treat
840 // something like this:
841 //
842 // I recommend upgrading to version
843 // 8. Oops, now this line is treated
844 // as a sub-list.
845 //
846 // As a single paragraph, despite the fact that the second line starts
847 // with a digit-period-space sequence.
848 //
849 // Whereas when we're inside a list (or sub-list), that line will be
850 // treated as the start of a sub-list. What a kludge, huh? This is
851 // an aspect of Markdown's syntax that's hard to parse perfectly
852 // without resorting to mind-reading. Perhaps the solution is to
853 // change the syntax rules such that sub-lists must start with a
854 // starting cardinal number; e.g. "1." or "a.".
855
856 g_list_level++;
857
858 // trim trailing blank lines:
859 list_str = list_str.replace(/\n{2,}$/, "\n");
860
861 // attacklab: add sentinel to emulate \z
862 list_str += "~0";
863
864 // In the original attacklab showdown, list_type was not given to this function, and anything
865 // that matched /[*+-]|\d+[.]/ would just create the next <li>, causing this mismatch:
866 //
867 // Markdown rendered by WMD rendered by MarkdownSharp
868 // ------------------------------------------------------------------
869 // 1. first 1. first 1. first
870 // 2. second 2. second 2. second
871 // - third 3. third * third
872 //
873 // We changed this to behave identical to MarkdownSharp. This is the constructed RegEx,
874 // with {MARKER} being one of \d+[.] or [*+-], depending on list_type:
875
876 /*
877 list_str = list_str.replace(/
878 (^[ \t]*) // leading whitespace = $1
879 ({MARKER}) [ \t]+ // list marker = $2
880 ([^\r]+? // list item text = $3
881 (\n+)
882 )
883 (?=
884 (~0 | \2 ({MARKER}) [ \t]+)
885 )
886 /gm, function(){...});
887 */
888
889 var marker = _listItemMarkers[list_type];
890 var re = new RegExp("(^[ \\t]*)(" + marker + ")[ \\t]+([^\\r]+?(\\n+))(?=(~0|\\1(" + marker + ")[ \\t]+))", "gm");
891 var last_item_had_a_double_newline = false;
892 list_str = list_str.replace(re,
893 function (wholeMatch, m1, m2, m3) {
894 var item = m3;
895 var leading_space = m1;
896 var ends_with_double_newline = /\n\n$/.test(item);
897 var contains_double_newline = ends_with_double_newline || item.search(/\n{2,}/) > -1;
898
899 if (contains_double_newline || last_item_had_a_double_newline) {
900 item = _RunBlockGamut(_Outdent(item), /* doNotUnhash = */true);
901 }
902 else {
903 // Recursion for sub-lists:
904 item = _DoLists(_Outdent(item));
905 item = item.replace(/\n$/, ""); // chomp(item)
906 item = _RunSpanGamut(item);
907 }
908 last_item_had_a_double_newline = ends_with_double_newline;
909 return "<li>" + item + "</li>\n";
910 }
911 );
912
913 // attacklab: strip sentinel
914 list_str = list_str.replace(/~0/g, "");
915
916 g_list_level--;
917 return list_str;
918 }
919
920 function _DoCodeBlocks(text) {
921 //
922 // Process Markdown `<pre><code>` blocks.
923 //
924
925 /*
926 text = text.replace(/
927 (?:\n\n|^)
928 ( // $1 = the code block -- one or more lines, starting with a space/tab
929 (?:
930 (?:[ ]{4}|\t) // Lines must start with a tab or a tab-width of spaces - attacklab: g_tab_width
931 .*\n+
932 )+
933 )
934 (\n*[ ]{0,3}[^ \t\n]|(?=~0)) // attacklab: g_tab_width
935 /g ,function(){...});
936 */
937
938 // attacklab: sentinel workarounds for lack of \A and \Z, safari\khtml bug
939 text += "~0";
940
941 text = text.replace(/(?:\n\n|^)((?:(?:[ ]{4}|\t).*\n+)+)(\n*[ ]{0,3}[^ \t\n]|(?=~0))/g,
942 function (wholeMatch, m1, m2) {
943 var codeblock = m1;
944 var nextChar = m2;
945
946 codeblock = _EncodeCode(_Outdent(codeblock));
947 codeblock = _Detab(codeblock);
948 codeblock = codeblock.replace(/^\n+/g, ""); // trim leading newlines
949 codeblock = codeblock.replace(/\n+$/g, ""); // trim trailing whitespace
950
951 codeblock = "<pre><code>" + codeblock + "\n</code></pre>";
952
953 return "\n\n" + codeblock + "\n\n" + nextChar;
954 }
955 );
956
957 // attacklab: strip sentinel
958 text = text.replace(/~0/, "");
959
960 return text;
961 }
962
963 function hashBlock(text) {
964 text = text.replace(/(^\n+|\n+$)/g, "");
965 return "\n\n~K" + (g_html_blocks.push(text) - 1) + "K\n\n";
966 }
967
968 function _DoCodeSpans(text) {
969 //
970 // * Backtick quotes are used for <code></code> spans.
971 //
972 // * You can use multiple backticks as the delimiters if you want to
973 // include literal backticks in the code span. So, this input:
974 //
975 // Just type ``foo `bar` baz`` at the prompt.
976 //
977 // Will translate to:
978 //
979 // <p>Just type <code>foo `bar` baz</code> at the prompt.</p>
980 //
981 // There's no arbitrary limit to the number of backticks you
982 // can use as delimters. If you need three consecutive backticks
983 // in your code, use four for delimiters, etc.
984 //
985 // * You can use spaces to get literal backticks at the edges:
986 //
987 // ... type `` `bar` `` ...
988 //
989 // Turns to:
990 //
991 // ... type <code>`bar`</code> ...
992 //
993
994 /*
995 text = text.replace(/
996 (^|[^\\]) // Character before opening ` can't be a backslash
997 (`+) // $2 = Opening run of `
998 ( // $3 = The code block
999 [^\r]*?
1000 [^`] // attacklab: work around lack of lookbehind
1001 )
1002 \2 // Matching closer
1003 (?!`)
1004 /gm, function(){...});
1005 */
1006
1007 text = text.replace(/(^|[^\\])(`+)([^\r]*?[^`])\2(?!`)/gm,
1008 function (wholeMatch, m1, m2, m3, m4) {
1009 var c = m3;
1010 c = c.replace(/^([ \t]*)/g, ""); // leading whitespace
1011 c = c.replace(/[ \t]*$/g, ""); // trailing whitespace
1012 c = _EncodeCode(c);
1013 c = c.replace(/:\/\//g, "~P"); // to prevent auto-linking. Not necessary in code *blocks*, but in code spans. Will be converted back after the auto-linker runs.
1014 return m1 + "<code>" + c + "</code>";
1015 }
1016 );
1017
1018 return text;
1019 }
1020
1021 function _EncodeCode(text) {
1022 //
1023 // Encode/escape certain characters inside Markdown code runs.
1024 // The point is that in code, these characters are literals,
1025 // and lose their special Markdown meanings.
1026 //
1027 // Encode all ampersands; HTML entities are not
1028 // entities within a Markdown code span.
1029 text = text.replace(/&/g, "&amp;");
1030
1031 // Do the angle bracket song and dance:
1032 text = text.replace(/</g, "&lt;");
1033 text = text.replace(/>/g, "&gt;");
1034
1035 // Now, escape characters that are magic in Markdown:
1036 text = escapeCharacters(text, "\*_{}[]\\", false);
1037
1038 // jj the line above breaks this:
1039 //---
1040
1041 //* Item
1042
1043 // 1. Subitem
1044
1045 // special char: *
1046 //---
1047
1048 return text;
1049 }
1050
1051 function _DoItalicsAndBold(text) {
1052
1053 // <strong> must go first:
1054 text = text.replace(/([\W_]|^)(\*\*|__)(?=\S)([^\r]*?\S[\*_]*)\2([\W_]|$)/g,
1055 "$1<strong>$3</strong>$4");
1056
1057 text = text.replace(/([\W_]|^)(\*|_)(?=\S)([^\r\*_]*?\S)\2([\W_]|$)/g,
1058 "$1<em>$3</em>$4");
1059
1060 return text;
1061 }
1062
1063 function _DoBlockQuotes(text) {
1064
1065 /*
1066 text = text.replace(/
1067 ( // Wrap whole match in $1
1068 (
1069 ^[ \t]*>[ \t]? // '>' at the start of a line
1070 .+\n // rest of the first line
1071 (.+\n)* // subsequent consecutive lines
1072 \n* // blanks
1073 )+
1074 )
1075 /gm, function(){...});
1076 */
1077
1078 text = text.replace(/((^[ \t]*>[ \t]?.+\n(.+\n)*\n*)+)/gm,
1079 function (wholeMatch, m1) {
1080 var bq = m1;
1081
1082 // attacklab: hack around Konqueror 3.5.4 bug:
1083 // "----------bug".replace(/^-/g,"") == "bug"
1084
1085 bq = bq.replace(/^[ \t]*>[ \t]?/gm, "~0"); // trim one level of quoting
1086
1087 // attacklab: clean up hack
1088 bq = bq.replace(/~0/g, "");
1089
1090 bq = bq.replace(/^[ \t]+$/gm, ""); // trim whitespace-only lines
1091 bq = _RunBlockGamut(bq); // recurse
1092
1093 bq = bq.replace(/(^|\n)/g, "$1 ");
1094 // These leading spaces screw with <pre> content, so we need to fix that:
1095 bq = bq.replace(
1096 /(\s*<pre>[^\r]+?<\/pre>)/gm,
1097 function (wholeMatch, m1) {
1098 var pre = m1;
1099 // attacklab: hack around Konqueror 3.5.4 bug:
1100 pre = pre.replace(/^ /mg, "~0");
1101 pre = pre.replace(/~0/g, "");
1102 return pre;
1103 });
1104
1105 return hashBlock("<blockquote>\n" + bq + "\n</blockquote>");
1106 }
1107 );
1108 return text;
1109 }
1110
1111 function _FormParagraphs(text, doNotUnhash) {
1112 //
1113 // Params:
1114 // $text - string to process with html <p> tags
1115 //
1116
1117 // Strip leading and trailing lines:
1118 text = text.replace(/^\n+/g, "");
1119 text = text.replace(/\n+$/g, "");
1120
1121 var grafs = text.split(/\n{2,}/g);
1122 var grafsOut = [];
1123
1124 var markerRe = /~K(\d+)K/;
1125
1126 //
1127 // Wrap <p> tags.
1128 //
1129 var end = grafs.length;
1130 for (var i = 0; i < end; i++) {
1131 var str = grafs[i];
1132
1133 // if this is an HTML marker, copy it
1134 if (markerRe.test(str)) {
1135 grafsOut.push(str);
1136 }
1137 else if (/\S/.test(str)) {
1138 str = _RunSpanGamut(str);
1139 str = str.replace(/^([ \t]*)/g, "<p>");
1140 str += "</p>"
1141 grafsOut.push(str);
1142 }
1143
1144 }
1145 //
1146 // Unhashify HTML blocks
1147 //
1148 if (!doNotUnhash) {
1149 end = grafsOut.length;
1150 for (var i = 0; i < end; i++) {
1151 var foundAny = true;
1152 while (foundAny) { // we may need several runs, since the data may be nested
1153 foundAny = false;
1154 grafsOut[i] = grafsOut[i].replace(/~K(\d+)K/g, function (wholeMatch, id) {
1155 foundAny = true;
1156 return g_html_blocks[id];
1157 });
1158 }
1159 }
1160 }
1161 return grafsOut.join("\n\n");
1162 }
1163
1164 function _EncodeAmpsAndAngles(text) {
1165 // Smart processing for ampersands and angle brackets that need to be encoded.
1166
1167 // Ampersand-encoding based entirely on Nat Irons's Amputator MT plugin:
1168 // http://bumppo.net/projects/amputator/
1169 text = text.replace(/&(?!#?[xX]?(?:[0-9a-fA-F]+|\w+);)/g, "&amp;");
1170
1171 // Encode naked <'s
1172 text = text.replace(/<(?![a-z\/?\$!])/gi, "&lt;");
1173
1174 return text;
1175 }
1176
1177 function _EncodeBackslashEscapes(text) {
1178 //
1179 // Parameter: String.
1180 // Returns: The string, with after processing the following backslash
1181 // escape sequences.
1182 //
1183
1184 // attacklab: The polite way to do this is with the new
1185 // escapeCharacters() function:
1186 //
1187 // text = escapeCharacters(text,"\\",true);
1188 // text = escapeCharacters(text,"`*_{}[]()>#+-.!",true);
1189 //
1190 // ...but we're sidestepping its use of the (slow) RegExp constructor
1191 // as an optimization for Firefox. This function gets called a LOT.
1192
1193 text = text.replace(/\\(\\)/g, escapeCharacters_callback);
1194 text = text.replace(/\\([`*_{}\[\]()>#+-.!])/g, escapeCharacters_callback);
1195 return text;
1196 }
1197
1198 function _DoAutoLinks(text) {
1199
1200 // note that at this point, all other URL in the text are already hyperlinked as <a href=""></a>
1201 // *except* for the <http://www.foo.com> case
1202
1203 // automatically add < and > around unadorned raw hyperlinks
1204 // must be preceded by space/BOF and followed by non-word/EOF character
1205 text = text.replace(/(^|\s)(https?|ftp)(:\/\/[-A-Z0-9+&@#\/%?=~_|\[\]\(\)!:,\.;]*[-A-Z0-9+&@#\/%=~_|\[\]])($|\W)/gi, "$1<$2$3>$4");
1206
1207 // autolink anything like <http://example.com>
1208
1209 var replacer = function (wholematch, m1) { return "<a href=\"" + m1 + "\">" + pluginHooks.plainLinkText(m1) + "</a>"; }
1210 text = text.replace(/<((https?|ftp):[^'">\s]+)>/gi, replacer);
1211
1212 // Email addresses: <address@domain.foo>
1213 /*
1214 text = text.replace(/
1215 <
1216 (?:mailto:)?
1217 (
1218 [-.\w]+
1219 \@
1220 [-a-z0-9]+(\.[-a-z0-9]+)*\.[a-z]+
1221 )
1222 >
1223 /gi, _DoAutoLinks_callback());
1224 */
1225
1226 /* disabling email autolinking, since we don't do that on the server, either
1227 text = text.replace(/<(?:mailto:)?([-.\w]+\@[-a-z0-9]+(\.[-a-z0-9]+)*\.[a-z]+)>/gi,
1228 function(wholeMatch,m1) {
1229 return _EncodeEmailAddress( _UnescapeSpecialChars(m1) );
1230 }
1231 );
1232 */
1233 return text;
1234 }
1235
1236 function _UnescapeSpecialChars(text) {
1237 //
1238 // Swap back in all the special characters we've hidden.
1239 //
1240 text = text.replace(/~E(\d+)E/g,
1241 function (wholeMatch, m1) {
1242 var charCodeToReplace = parseInt(m1);
1243 return String.fromCharCode(charCodeToReplace);
1244 }
1245 );
1246 return text;
1247 }
1248
1249 function _Outdent(text) {
1250 //
1251 // Remove one level of line-leading tabs or spaces
1252 //
1253
1254 // attacklab: hack around Konqueror 3.5.4 bug:
1255 // "----------bug".replace(/^-/g,"") == "bug"
1256
1257 text = text.replace(/^(\t|[ ]{1,4})/gm, "~0"); // attacklab: g_tab_width
1258
1259 // attacklab: clean up hack
1260 text = text.replace(/~0/g, "")
1261
1262 return text;
1263 }
1264
1265 function _Detab(text) {
1266 if (!/\t/.test(text))
1267 return text;
1268
1269 var spaces = [" ", " ", " ", " "],
1270 skew = 0,
1271 v;
1272
1273 return text.replace(/[\n\t]/g, function (match, offset) {
1274 if (match === "\n") {
1275 skew = offset + 1;
1276 return match;
1277 }
1278 v = (offset - skew) % 4;
1279 skew = offset + 1;
1280 return spaces[v];
1281 });
1282 }
1283
1284 //
1285 // attacklab: Utility functions
1286 //
1287
1288 var _problemUrlChars = /(?:["'*()[\]:]|~D)/g;
1289
1290 // hex-encodes some unusual "problem" chars in URLs to avoid URL detection problems
1291 function encodeProblemUrlChars(url) {
1292 if (!url)
1293 return "";
1294
1295 var len = url.length;
1296
1297 return url.replace(_problemUrlChars, function (match, offset) {
1298 if (match == "~D") // escape for dollar
1299 return "%24";
1300 if (match == ":") {
1301 if (offset == len - 1 || /[0-9\/]/.test(url.charAt(offset + 1)))
1302 return ":"
1303 }
1304 return "%" + match.charCodeAt(0).toString(16);
1305 });
1306 }
1307
1308
1309 function escapeCharacters(text, charsToEscape, afterBackslash) {
1310 // First we have to escape the escape characters so that
1311 // we can build a character class out of them
1312 var regexString = "([" + charsToEscape.replace(/([\[\]\\])/g, "\\$1") + "])";
1313
1314 if (afterBackslash) {
1315 regexString = "\\\\" + regexString;
1316 }
1317
1318 var regex = new RegExp(regexString, "g");
1319 text = text.replace(regex, escapeCharacters_callback);
1320
1321 return text;
1322 }
1323
1324
1325 function escapeCharacters_callback(wholeMatch, m1) {
1326 var charCodeToEscape = m1.charCodeAt(0);
1327 return "~E" + charCodeToEscape + "E";
1328 }
1329
1330 }; // end of the Markdown.Converter constructor
1331
1332 })();