forum/skins/default/media/js/wmd/showdown.js

   1 //
   2 // showdown.js -- A javascript port of Markdown.
   3 //
   4 // Copyright (c) 2007 John Fraser.
   5 //
   6 // Original Markdown Copyright (c) 2004-2005 John Gruber
   7 //   <http://daringfireball.net/projects/markdown/>
   8 //
   9 // The full source distribution is at:
  10 //
  11 //                              A A L
  12 //                              T C A
  13 //                              T K B
  14 //
  15 //   <http://www.attacklab.net/>
  16 //
  17
  18 //
  19 // Wherever possible, Showdown is a straight, line-by-line port
  20 // of the Perl version of Markdown.
  21 //
  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.
  26 //
  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.
  30 //
  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.
  37 //
  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
  42 // and line endings.
  43 //
  44
  45
  46 //
  47 // Showdown usage:
  48 //
  49 //   var text = "Markdown *rocks*.";
  50 //
  51 //   var converter = new Attacklab.showdown.converter();
  52 //   var html = converter.makeHtml(text);
  53 //
  54 //   alert(html);
  55 //
  56 // Note: move the sample code to the bottom of this
  57 // file before uncommenting it.
  58 //
  59
  60
  61 //
  62 // Attacklab namespace
  63 //
  64 var Attacklab = Attacklab || {}
  65
  66 //
  67 // Showdown namespace
  68 //
  69 Attacklab.showdown = Attacklab.showdown || {}
  70
  71 //
  72 // converter
  73 //
  74 // Wraps all "globals" so that the only thing
  75 // exposed is makeHtml().
  76 //
  77 Attacklab.showdown.converter = function() {
  78
  79 //
  80 // Globals:
  81 //
  82
  83 // Global hashes, used by various utility routines
  84 var g_urls;
  85 var g_titles;
  86 var g_html_blocks;
  87
  88 // Used to track when we're inside an ordered or unordered list
  89 // (see _ProcessListItems() for details):
  90 var g_list_level = 0;
  91
  92
  93 this.makeHtml = function(text) {
  94 //
  95 // Main function. The order in which other subs are called here is
  96 // essential. Link and image substitutions need to happen before
  97 // _EscapeSpecialCharsWithinTagAttributes(), so that any *'s or _'s in the <a>
  98 // and <img> tags get encoded.
  99 //
 100
 101         // Clear the global hashes. If we don't clear these, you get conflicts
 102         // from other articles when generating a page which contains more than
 103         // one article (e.g. an index page that shows the N most recent
 104         // articles):
 105         g_urls = new Array();
 106         g_titles = new Array();
 107         g_html_blocks = new Array();
 108
 109         // attacklab: Replace ~ with ~T
 110         // This lets us use tilde as an escape char to avoid md5 hashes
 111         // The choice of character is arbitray; anything that isn't
 112     // magic in Markdown will work.
 113         text = text.replace(/~/g,"~T");
 114
 115         // attacklab: Replace $ with ~D
 116         // RegExp interprets $ as a special character
 117         // when it's in a replacement string
 118         text = text.replace(/\$/g,"~D");
 119
 120         // Standardize line endings
 121         text = text.replace(/\r\n/g,"\n"); // DOS to Unix
 122         text = text.replace(/\r/g,"\n"); // Mac to Unix
 123
 124         // Make sure text begins and ends with a couple of newlines:
 125         text = "\n\n" + text + "\n\n";
 126
 127         // Convert all tabs to spaces.
 128         text = _Detab(text);
 129
 130         // Strip any lines consisting only of spaces and tabs.
 131         // This makes subsequent regexen easier to write, because we can
 132         // match consecutive blank lines with /\n+/ instead of something
 133         // contorted like /[ \t]*\n+/ .
 134         text = text.replace(/^[ \t]+$/mg,"");
 135
 136         // Turn block-level HTML blocks into hash entries
 137         text = _HashHTMLBlocks(text);
 138
 139         // Strip link definitions, store in hashes.
 140         text = _StripLinkDefinitions(text);
 141
 142         text = _RunBlockGamut(text);
 143
 144         text = _UnescapeSpecialChars(text);
 145
 146         // attacklab: Restore dollar signs
 147         text = text.replace(/~D/g,"$$");
 148
 149         // attacklab: Restore tildes
 150         text = text.replace(/~T/g,"~");
 151
 152         return text;
 153 }
 154
 155 var _StripLinkDefinitions = function(text) {
 156 //
 157 // Strips link definitions from text, stores the URLs and titles in
 158 // hash references.
 159 //
 160
 161         // Link defs are in the form: ^[id]: url "optional title"
 162
 163         /*
 164                 var text = text.replace(/
 165                                 ^[ ]{0,3}\[(.+)\]:  // id = $1  attacklab: g_tab_width - 1
 166                                   [ \t]*
 167                                   \n?                           // maybe *one* newline
 168                                   [ \t]*
 169                                 <?(\S+?)>?                      // url = $2
 170                                   [ \t]*
 171                                   \n?                           // maybe one newline
 172                                   [ \t]*
 173                                 (?:
 174                                   (\n*)                         // any lines skipped = $3 attacklab: lookbehind removed
 175                                   ["(]
 176                                   (.+?)                         // title = $4
 177                                   [")]
 178                                   [ \t]*
 179                                 )?                                      // title is optional
 180                                 (?:\n+|$)
 181                           /gm,
 182                           function(){...});
 183         */
 184         var text = text.replace(/^[ ]{0,3}\[(.+)\]:[ \t]*\n?[ \t]*<?(\S+?)>?[ \t]*\n?[ \t]*(?:(\n*)["(](.+?)[")][ \t]*)?(?:\n+)/gm,
 185                 function (wholeMatch,m1,m2,m3,m4) {
 186                         m1 = m1.toLowerCase();
 187                         g_urls[m1] = _EncodeAmpsAndAngles(m2);  // Link IDs are case-insensitive
 188                         if (m3) {
 189                                 // Oops, found blank lines, so it's not a title.
 190                                 // Put back the parenthetical statement we stole.
 191                                 return m3+m4;
 192                         } else if (m4) {
 193                                 g_titles[m1] = m4.replace(/"/g,"&quot;");
 194                         }
 195
 196                         // Completely remove the definition from the text
 197                         return "";
 198                 }
 199         );
 200
 201         return text;
 202 }
 203
 204 var _HashHTMLBlocks = function(text) {
 205         // attacklab: Double up blank lines to reduce lookaround
 206         text = text.replace(/\n/g,"\n\n");
 207
 208         // Hashify HTML blocks:
 209         // We only want to do this for block-level HTML tags, such as headers,
 210         // lists, and tables. That's because we still want to wrap <p>s around
 211         // "paragraphs" that are wrapped in non-block-level tags, such as anchors,
 212         // phrase emphasis, and spans. The list of tags we're looking for is
 213         // hard-coded:
 214         var block_tags_a = "p|div|h[1-6]|blockquote|pre|table|dl|ol|ul|script|noscript|form|fieldset|iframe|math|ins|del"
 215         var block_tags_b = "p|div|h[1-6]|blockquote|pre|table|dl|ol|ul|script|noscript|form|fieldset|iframe|math"
 216
 217         // First, look for nested blocks, e.g.:
 218         //   <div>
 219         //     <div>
 220         //     tags for inner block must be indented.
 221         //     </div>
 222         //   </div>
 223         //
 224         // The outermost tags must start at the left margin for this to match, and
 225         // the inner nested divs must be indented.
 226         // We need to do this before the next, more liberal match, because the next
 227         // match will start at the first `<div>` and stop at the first `</div>`.
 228
 229         // attacklab: This regex can be expensive when it fails.
 230         /*
 231                 var text = text.replace(/
 232                 (                                               // save in $1
 233                         ^                                       // start of line  (with /m)
 234                         <($block_tags_a)        // start tag = $2
 235                         \b                                      // word break
 236                                                                 // attacklab: hack around khtml/pcre bug...
 237                         [^\r]*?\n                       // any number of lines, minimally matching
 238                         </\2>                           // the matching end tag
 239                         [ \t]*                          // trailing spaces/tabs
 240                         (?=\n+)                         // followed by a newline
 241                 )                                               // attacklab: there are sentinel newlines at end of document
 242                 /gm,function(){...}};
 243         */
 244         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);
 245
 246         //
 247         // Now match more liberally, simply from `\n<tag>` to `</tag>\n`
 248         //
 249
 250         /*
 251                 var text = text.replace(/
 252                 (                                               // save in $1
 253                         ^                                       // start of line  (with /m)
 254                         <($block_tags_b)        // start tag = $2
 255                         \b                                      // word break
 256                                                                 // attacklab: hack around khtml/pcre bug...
 257                         [^\r]*?                         // any number of lines, minimally matching
 258                         .*</\2>                         // the matching end tag
 259                         [ \t]*                          // trailing spaces/tabs
 260                         (?=\n+)                         // followed by a newline
 261                 )                                               // attacklab: there are sentinel newlines at end of document
 262                 /gm,function(){...}};
 263         */
 264         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);
 265
 266         // Special case just for <hr />. It was easier to make a special case than
 267         // to make the other regex more complicated.
 268
 269         /*
 270                 text = text.replace(/
 271                 (                                               // save in $1
 272                         \n\n                            // Starting after a blank line
 273                         [ ]{0,3}
 274                         (<(hr)                          // start tag = $2
 275                         \b                                      // word break
 276                         ([^<>])*?                       //
 277                         \/?>)                           // the matching end tag
 278                         [ \t]*
 279                         (?=\n{2,})                      // followed by a blank line
 280                 )
 281                 /g,hashElement);
 282         */
 283         text = text.replace(/(\n[ ]{0,3}(<(hr)\b([^<>])*?\/?>)[ \t]*(?=\n{2,}))/g,hashElement);
 284
 285         // Special case for standalone HTML comments:
 286
 287         /*
 288                 text = text.replace(/
 289                 (                                               // save in $1
 290                         \n\n                            // Starting after a blank line
 291                         [ ]{0,3}                        // attacklab: g_tab_width - 1
 292                         <!
 293                         (--[^\r]*?--\s*)+
 294                         >
 295                         [ \t]*
 296                         (?=\n{2,})                      // followed by a blank line
 297                 )
 298                 /g,hashElement);
 299         */
 300         text = text.replace(/(\n\n[ ]{0,3}<!(--[^\r]*?--\s*)+>[ \t]*(?=\n{2,}))/g,hashElement);
 301
 302         // PHP and ASP-style processor instructions (<?...?> and <%...%>)
 303
 304         /*
 305                 text = text.replace(/
 306                 (?:
 307                         \n\n                            // Starting after a blank line
 308                 )
 309                 (                                               // save in $1
 310                         [ ]{0,3}                        // attacklab: g_tab_width - 1
 311                         (?:
 312                                 <([?%])                 // $2
 313                                 [^\r]*?
 314                                 \2>
 315                         )
 316                         [ \t]*
 317                         (?=\n{2,})                      // followed by a blank line
 318                 )
 319                 /g,hashElement);
 320         */
 321         text = text.replace(/(?:\n\n)([ ]{0,3}(?:<([?%])[^\r]*?\2>)[ \t]*(?=\n{2,}))/g,hashElement);
 322
 323         // attacklab: Undo double lines (see comment at top of this function)
 324         text = text.replace(/\n\n/g,"\n");
 325         return text;
 326 }
 327
 328 var hashElement = function(wholeMatch,m1) {
 329         var blockText = m1;
 330
 331         // Undo double lines
 332         blockText = blockText.replace(/\n\n/g,"\n");
 333         blockText = blockText.replace(/^\n/,"");
 334
 335         // strip trailing blank lines
 336         blockText = blockText.replace(/\n+$/g,"");
 337
 338         // Replace the element text with a marker ("~KxK" where x is its key)
 339         blockText = "\n\n~K" + (g_html_blocks.push(blockText)-1) + "K\n\n";
 340
 341         return blockText;
 342 };
 343
 344 var _RunBlockGamut = function(text) {
 345 //
 346 // These are all the transformations that form block-level
 347 // tags like paragraphs, headers, and list items.
 348 //
 349         text = _DoHeaders(text);
 350
 351         // Do Horizontal Rules:
 352         var key = hashBlock("<hr />");
 353         text = text.replace(/^[ ]{0,2}([ ]?\*[ ]?){3,}[ \t]*$/gm,key);
 354         text = text.replace(/^[ ]{0,2}([ ]?-[ ]?){3,}[ \t]*$/gm,key);
 355         text = text.replace(/^[ ]{0,2}([ ]?_[ ]?){3,}[ \t]*$/gm,key);
 356
 357         text = _DoLists(text);
 358         text = _DoCodeBlocks(text);
 359         text = _DoBlockQuotes(text);
 360
 361         // We already ran _HashHTMLBlocks() before, in Markdown(), but that
 362         // was to escape raw HTML in the original Markdown source. This time,
 363         // we're escaping the markup we've just created, so that we don't wrap
 364         // <p> tags around block-level tags.
 365         text = _HashHTMLBlocks(text);
 366         text = _FormParagraphs(text);
 367
 368         return text;
 369 }
 370
 371
 372 var _RunSpanGamut = function(text) {
 373 //
 374 // These are all the transformations that occur *within* block-level
 375 // tags like paragraphs, headers, and list items.
 376 //
 377
 378         text = _DoCodeSpans(text);
 379         text = _EscapeSpecialCharsWithinTagAttributes(text);
 380         text = _EncodeBackslashEscapes(text);
 381
 382         // Process anchor and image tags. Images must come first,
 383         // because ![foo][f] looks like an anchor.
 384         text = _DoImages(text);
 385         text = _DoAnchors(text);
 386
 387         // Make links out of things like `<http://example.com/>`
 388         // Must come after _DoAnchors(), because you can use < and >
 389         // delimiters in inline links like [this](<url>).
 390         text = _DoAutoLinks(text);
 391         text = _EncodeAmpsAndAngles(text);
 392         text = _DoItalicsAndBold(text);
 393
 394         // Do hard breaks:
 395         text = text.replace(/  +\n/g," <br />\n");
 396
 397         return text;
 398 }
 399
 400 var _EscapeSpecialCharsWithinTagAttributes = function(text) {
 401 //
 402 // Within tags -- meaning between < and > -- encode [\ ` * _] so they
 403 // don't conflict with their use in Markdown for code, italics and strong.
 404 //
 405
 406         // Build a regex to find HTML tags and comments.  See Friedl's
 407         // "Mastering Regular Expressions", 2nd Ed., pp. 200-201.
 408         var regex = /(<[a-z\/!$]("[^"]*"|'[^']*'|[^'">])*>|<!(--.*?--\s*)+>)/gi;
 409
 410         text = text.replace(regex, function(wholeMatch) {
 411                 var tag = wholeMatch.replace(/(.)<\/?code>(?=.)/g,"$1`");
 412                 tag = escapeCharacters(tag,"\\`*_");
 413                 return tag;
 414         });
 415
 416         return text;
 417 }
 418
 419 var _DoAnchors = function(text) {
 420 //
 421 // Turn Markdown link shortcuts into XHTML <a> tags.
 422 //
 423         //
 424         // First, handle reference-style links: [link text] [id]
 425         //
 426
 427         /*
 428                 text = text.replace(/
 429                 (                                                       // wrap whole match in $1
 430                         \[
 431                         (
 432                                 (?:
 433                                         \[[^\]]*\]              // allow brackets nested one level
 434                                         |
 435                                         [^\[]                   // or anything else
 436                                 )*
 437                         )
 438                         \]
 439
 440                         [ ]?                                    // one optional space
 441                         (?:\n[ ]*)?                             // one optional newline followed by spaces
 442
 443                         \[
 444                         (.*?)                                   // id = $3
 445                         \]
 446                 )()()()()                                       // pad remaining backreferences
 447                 /g,_DoAnchors_callback);
 448         */
 449         text = text.replace(/(\[((?:\[[^\]]*\]|[^\[\]])*)\][ ]?(?:\n[ ]*)?\[(.*?)\])()()()()/g,writeAnchorTag);
 450
 451         //
 452         // Next, inline-style links: [link text](url "optional title")
 453         //
 454
 455         /*
 456                 text = text.replace(/
 457                         (                                               // wrap whole match in $1
 458                                 \[
 459                                 (
 460                                         (?:
 461                                                 \[[^\]]*\]      // allow brackets nested one level
 462                                         |
 463                                         [^\[\]]                 // or anything else
 464                                 )
 465                         )
 466                         \]
 467                         \(                                              // literal paren
 468                         [ \t]*
 469                         ()                                              // no id, so leave $3 empty
 470                         <?(.*?)>?                               // href = $4
 471                         [ \t]*
 472                         (                                               // $5
 473                                 (['"])                          // quote char = $6
 474                                 (.*?)                           // Title = $7
 475                                 \6                                      // matching quote
 476                                 [ \t]*                          // ignore any spaces/tabs between closing quote and )
 477                         )?                                              // title is optional
 478                         \)
 479                 )
 480                 /g,writeAnchorTag);
 481         */
 482         text = text.replace(/(\[((?:\[[^\]]*\]|[^\[\]])*)\]\([ \t]*()<?(.*?)>?[ \t]*((['"])(.*?)\6[ \t]*)?\))/g,writeAnchorTag);
 483
 484         //
 485         // Last, handle reference-style shortcuts: [link text]
 486         // These must come last in case you've also got [link test][1]
 487         // or [link test](/foo)
 488         //
 489
 490         /*
 491                 text = text.replace(/
 492                 (                                                       // wrap whole match in $1
 493                         \[
 494                         ([^\[\]]+)                              // link text = $2; can't contain '[' or ']'
 495                         \]
 496                 )()()()()()                                     // pad rest of backreferences
 497                 /g, writeAnchorTag);
 498         */
 499         text = text.replace(/(\[([^\[\]]+)\])()()()()()/g, writeAnchorTag);
 500
 501         return text;
 502 }
 503
 504 var writeAnchorTag = function(wholeMatch,m1,m2,m3,m4,m5,m6,m7) {
 505         if (m7 == undefined) m7 = "";
 506         var whole_match = m1;
 507         var link_text   = m2;
 508         var link_id      = m3.toLowerCase();
 509         var url         = m4;
 510         var title       = m7;
 511
 512         if (url == "") {
 513                 if (link_id == "") {
 514                         // lower-case and turn embedded newlines into spaces
 515                         link_id = link_text.toLowerCase().replace(/ ?\n/g," ");
 516                 }
 517                 url = "#"+link_id;
 518
 519                 if (g_urls[link_id] != undefined) {
 520                         url = g_urls[link_id];
 521                         if (g_titles[link_id] != undefined) {
 522                                 title = g_titles[link_id];
 523                         }
 524                 }
 525                 else {
 526                         if (whole_match.search(/\(\s*\)$/m)>-1) {
 527                                 // Special case for explicit empty url
 528                                 url = "";
 529                         } else {
 530                                 return whole_match;
 531                         }
 532                 }
 533         }
 534
 535         url = escapeCharacters(url,"*_");
 536         var result = "<a href=\"" + url + "\"";
 537
 538         if (title != "") {
 539                 title = title.replace(/"/g,"&quot;");
 540                 title = escapeCharacters(title,"*_");
 541                 result +=  " title=\"" + title + "\"";
 542         }
 543
 544         result += ">" + link_text + "</a>";
 545
 546         return result;
 547 }
 548
 549
 550 var _DoImages = function(text) {
 551 //
 552 // Turn Markdown image shortcuts into <img> tags.
 553 //
 554
 555         //
 556         // First, handle reference-style labeled images: ![alt text][id]
 557         //
 558
 559         /*
 560                 text = text.replace(/
 561                 (                                               // wrap whole match in $1
 562                         !\[
 563                         (.*?)                           // alt text = $2
 564                         \]
 565
 566                         [ ]?                            // one optional space
 567                         (?:\n[ ]*)?                     // one optional newline followed by spaces
 568
 569                         \[
 570                         (.*?)                           // id = $3
 571                         \]
 572                 )()()()()                               // pad rest of backreferences
 573                 /g,writeImageTag);
 574         */
 575         text = text.replace(/(!\[(.*?)\][ ]?(?:\n[ ]*)?\[(.*?)\])()()()()/g,writeImageTag);
 576
 577         //
 578         // Next, handle inline images:  ![alt text](url "optional title")
 579         // Don't forget: encode * and _
 580
 581         /*
 582                 text = text.replace(/
 583                 (                                               // wrap whole match in $1
 584                         !\[
 585                         (.*?)                           // alt text = $2
 586                         \]
 587                         \s?                                     // One optional whitespace character
 588                         \(                                      // literal paren
 589                         [ \t]*
 590                         ()                                      // no id, so leave $3 empty
 591                         <?(\S+?)>?                      // src url = $4
 592                         [ \t]*
 593                         (                                       // $5
 594                                 (['"])                  // quote char = $6
 595                                 (.*?)                   // title = $7
 596                                 \6                              // matching quote
 597                                 [ \t]*
 598                         )?                                      // title is optional
 599                 \)
 600                 )
 601                 /g,writeImageTag);
 602         */
 603         text = text.replace(/(!\[(.*?)\]\s?\([ \t]*()<?(\S+?)>?[ \t]*((['"])(.*?)\6[ \t]*)?\))/g,writeImageTag);
 604
 605         return text;
 606 }
 607
 608 var writeImageTag = function(wholeMatch,m1,m2,m3,m4,m5,m6,m7) {
 609         var whole_match = m1;
 610         var alt_text   = m2;
 611         var link_id      = m3.toLowerCase();
 612         var url         = m4;
 613         var title       = m7;
 614
 615         if (!title) title = "";
 616
 617         if (url == "") {
 618                 if (link_id == "") {
 619                         // lower-case and turn embedded newlines into spaces
 620                         link_id = alt_text.toLowerCase().replace(/ ?\n/g," ");
 621                 }
 622                 url = "#"+link_id;
 623
 624                 if (g_urls[link_id] != undefined) {
 625                         url = g_urls[link_id];
 626                         if (g_titles[link_id] != undefined) {
 627                                 title = g_titles[link_id];
 628                         }
 629                 }
 630                 else {
 631                         return whole_match;
 632                 }
 633         }
 634
 635         alt_text = alt_text.replace(/"/g,"&quot;");
 636         url = escapeCharacters(url,"*_");
 637         var result = "<img src=\"" + url + "\" alt=\"" + alt_text + "\"";
 638
 639         // attacklab: Markdown.pl adds empty title attributes to images.
 640         // Replicate this bug.
 641
 642         //if (title != "") {
 643                 title = title.replace(/"/g,"&quot;");
 644                 title = escapeCharacters(title,"*_");
 645                 result +=  " title=\"" + title + "\"";
 646         //}
 647
 648         result += " />";
 649
 650         return result;
 651 }
 652
 653
 654 var _DoHeaders = function(text) {
 655
 656         // Setext-style headers:
 657         //      Header 1
 658         //      ========
 659         //
 660         //      Header 2
 661         //      --------
 662         //
 663         text = text.replace(/^(.+)[ \t]*\n=+[ \t]*\n+/gm,
 664                 function(wholeMatch,m1){return hashBlock("<h1>" + _RunSpanGamut(m1) + "</h1>");});
 665
 666         text = text.replace(/^(.+)[ \t]*\n-+[ \t]*\n+/gm,
 667                 function(matchFound,m1){return hashBlock("<h2>" + _RunSpanGamut(m1) + "</h2>");});
 668
 669         // atx-style headers:
 670         //  # Header 1
 671         //  ## Header 2
 672         //  ## Header 2 with closing hashes ##
 673         //  ...
 674         //  ###### Header 6
 675         //
 676
 677         /*
 678                 text = text.replace(/
 679                         ^(\#{1,6})                              // $1 = string of #'s
 680                         [ \t]*
 681                         (.+?)                                   // $2 = Header text
 682                         [ \t]*
 683                         \#*                                             // optional closing #'s (not counted)
 684                         \n+
 685                 /gm, function() {...});
 686         */
 687
 688         text = text.replace(/^(\#{1,6})[ \t]*(.+?)[ \t]*\#*\n+/gm,
 689                 function(wholeMatch,m1,m2) {
 690                         var h_level = m1.length;
 691                         return hashBlock("<h" + h_level + ">" + _RunSpanGamut(m2) + "</h" + h_level + ">");
 692                 });
 693
 694         return text;
 695 }
 696
 697 // This declaration keeps Dojo compressor from outputting garbage:
 698 var _ProcessListItems;
 699
 700 var _DoLists = function(text) {
 701 //
 702 // Form HTML ordered (numbered) and unordered (bulleted) lists.
 703 //
 704
 705         // attacklab: add sentinel to hack around khtml/safari bug:
 706         // http://bugs.webkit.org/show_bug.cgi?id=11231
 707         text += "~0";
 708
 709         // Re-usable pattern to match any entirel ul or ol list:
 710
 711         /*
 712                 var whole_list = /
 713                 (                                                                       // $1 = whole list
 714                         (                                                               // $2
 715                                 [ ]{0,3}                                        // attacklab: g_tab_width - 1
 716                                 ([*+-]|\d+[.])                          // $3 = first list item marker
 717                                 [ \t]+
 718                         )
 719                         [^\r]+?
 720                         (                                                               // $4
 721                                 ~0                                                      // sentinel for workaround; should be $
 722                         |
 723                                 \n{2,}
 724                                 (?=\S)
 725                                 (?!                                                     // Negative lookahead for another list item marker
 726                                         [ \t]*
 727                                         (?:[*+-]|\d+[.])[ \t]+
 728                                 )
 729                         )
 730                 )/g
 731         */
 732         var whole_list = /^(([ ]{0,3}([*+-]|\d+[.])[ \t]+)[^\r]+?(~0|\n{2,}(?=\S)(?![ \t]*(?:[*+-]|\d+[.])[ \t]+)))/gm;
 733
 734         if (g_list_level) {
 735                 text = text.replace(whole_list,function(wholeMatch,m1,m2) {
 736                         var list = m1;
 737                         var list_type = (m2.search(/[*+-]/g)>-1) ? "ul" : "ol";
 738
 739                         // Turn double returns into triple returns, so that we can make a
 740                         // paragraph for the last item in a list, if necessary:
 741                         list = list.replace(/\n{2,}/g,"\n\n\n");;
 742                         var result = _ProcessListItems(list);
 743
 744                         // Trim any trailing whitespace, to put the closing `</$list_type>`
 745                         // up on the preceding line, to get it past the current stupid
 746                         // HTML block parser. This is a hack to work around the terrible
 747                         // hack that is the HTML block parser.
 748                         result = result.replace(/\s+$/,"");
 749                         result = "<"+list_type+">" + result + "</"+list_type+">\n";
 750                         return result;
 751                 });
 752         } else {
 753                 whole_list = /(\n\n|^\n?)(([ ]{0,3}([*+-]|\d+[.])[ \t]+)[^\r]+?(~0|\n{2,}(?=\S)(?![ \t]*(?:[*+-]|\d+[.])[ \t]+)))/g;
 754                 text = text.replace(whole_list,function(wholeMatch,m1,m2,m3) {
 755                         var runup = m1;
 756                         var list = m2;
 757
 758                         var list_type = (m3.search(/[*+-]/g)>-1) ? "ul" : "ol";
 759                         // Turn double returns into triple returns, so that we can make a
 760                         // paragraph for the last item in a list, if necessary:
 761                         var list = list.replace(/\n{2,}/g,"\n\n\n");;
 762                         var result = _ProcessListItems(list);
 763                         result = runup + "<"+list_type+">\n" + result + "</"+list_type+">\n";
 764                         return result;
 765                 });
 766         }
 767
 768         // attacklab: strip sentinel
 769         text = text.replace(/~0/,"");
 770
 771         return text;
 772 }
 773
 774 _ProcessListItems = function(list_str) {
 775 //
 776 //  Process the contents of a single ordered or unordered list, splitting it
 777 //  into individual list items.
 778 //
 779         // The $g_list_level global keeps track of when we're inside a list.
 780         // Each time we enter a list, we increment it; when we leave a list,
 781         // we decrement. If it's zero, we're not in a list anymore.
 782         //
 783         // We do this because when we're not inside a list, we want to treat
 784         // something like this:
 785         //
 786         //    I recommend upgrading to version
 787         //    8. Oops, now this line is treated
 788         //    as a sub-list.
 789         //
 790         // As a single paragraph, despite the fact that the second line starts
 791         // with a digit-period-space sequence.
 792         //
 793         // Whereas when we're inside a list (or sub-list), that line will be
 794         // treated as the start of a sub-list. What a kludge, huh? This is
 795         // an aspect of Markdown's syntax that's hard to parse perfectly
 796         // without resorting to mind-reading. Perhaps the solution is to
 797         // change the syntax rules such that sub-lists must start with a
 798         // starting cardinal number; e.g. "1." or "a.".
 799
 800         g_list_level++;
 801
 802         // trim trailing blank lines:
 803         list_str = list_str.replace(/\n{2,}$/,"\n");
 804
 805         // attacklab: add sentinel to emulate \z
 806         list_str += "~0";
 807
 808         /*
 809                 list_str = list_str.replace(/
 810                         (\n)?                                                   // leading line = $1
 811                         (^[ \t]*)                                               // leading whitespace = $2
 812                         ([*+-]|\d+[.]) [ \t]+                   // list marker = $3
 813                         ([^\r]+?                                                // list item text   = $4
 814                         (\n{1,2}))
 815                         (?= \n* (~0 | \2 ([*+-]|\d+[.]) [ \t]+))
 816                 /gm, function(){...});
 817         */
 818         list_str = list_str.replace(/(\n)?(^[ \t]*)([*+-]|\d+[.])[ \t]+([^\r]+?(\n{1,2}))(?=\n*(~0|\2([*+-]|\d+[.])[ \t]+))/gm,
 819                 function(wholeMatch,m1,m2,m3,m4){
 820                         var item = m4;
 821                         var leading_line = m1;
 822                         var leading_space = m2;
 823
 824                         if (leading_line || (item.search(/\n{2,}/)>-1)) {
 825                                 item = _RunBlockGamut(_Outdent(item));
 826                         }
 827                         else {
 828                                 // Recursion for sub-lists:
 829                                 item = _DoLists(_Outdent(item));
 830                                 item = item.replace(/\n$/,""); // chomp(item)
 831                                 item = _RunSpanGamut(item);
 832                         }
 833
 834                         return  "<li>" + item + "</li>\n";
 835                 }
 836         );
 837
 838         // attacklab: strip sentinel
 839         list_str = list_str.replace(/~0/g,"");
 840
 841         g_list_level--;
 842         return list_str;
 843 }
 844
 845
 846 var _DoCodeBlocks = function(text) {
 847 //
 848 //  Process Markdown `<pre><code>` blocks.
 849 //
 850
 851         /*
 852                 text = text.replace(text,
 853                         /(?:\n\n|^)
 854                         (                                                               // $1 = the code block -- one or more lines, starting with a space/tab
 855                                 (?:
 856                                         (?:[ ]{4}|\t)                   // Lines must start with a tab or a tab-width of spaces - attacklab: g_tab_width
 857                                         .*\n+
 858                                 )+
 859                         )
 860                         (\n*[ ]{0,3}[^ \t\n]|(?=~0))    // attacklab: g_tab_width
 861                 /g,function(){...});
 862         */
 863
 864         // attacklab: sentinel workarounds for lack of \A and \Z, safari\khtml bug
 865         text += "~0";
 866
 867         text = text.replace(/(?:\n\n|^)((?:(?:[ ]{4}|\t).*\n+)+)(\n*[ ]{0,3}[^ \t\n]|(?=~0))/g,
 868                 function(wholeMatch,m1,m2) {
 869                         var codeblock = m1;
 870                         var nextChar = m2;
 871
 872                         codeblock = _EncodeCode( _Outdent(codeblock));
 873                         codeblock = _Detab(codeblock);
 874                         codeblock = codeblock.replace(/^\n+/g,""); // trim leading newlines
 875                         codeblock = codeblock.replace(/\n+$/g,""); // trim trailing whitespace
 876
 877                         codeblock = "<pre><code>" + codeblock + "\n</code></pre>";
 878
 879                         return hashBlock(codeblock) + nextChar;
 880                 }
 881         );
 882
 883         // attacklab: strip sentinel
 884         text = text.replace(/~0/,"");
 885
 886         return text;
 887 }
 888
 889 var hashBlock = function(text) {
 890         text = text.replace(/(^\n+|\n+$)/g,"");
 891         return "\n\n~K" + (g_html_blocks.push(text)-1) + "K\n\n";
 892 }
 893
 894
 895 var _DoCodeSpans = function(text) {
 896 //
 897 //   *  Backtick quotes are used for <code></code> spans.
 898 //
 899 //   *  You can use multiple backticks as the delimiters if you want to
 900 //       include literal backticks in the code span. So, this input:
 901 //
 902 //               Just type ``foo `bar` baz`` at the prompt.
 903 //
 904 //         Will translate to:
 905 //
 906 //               <p>Just type <code>foo `bar` baz</code> at the prompt.</p>
 907 //
 908 //      There's no arbitrary limit to the number of backticks you
 909 //      can use as delimters. If you need three consecutive backticks
 910 //      in your code, use four for delimiters, etc.
 911 //
 912 //  *  You can use spaces to get literal backticks at the edges:
 913 //
 914 //               ... type `` `bar` `` ...
 915 //
 916 //         Turns to:
 917 //
 918 //               ... type <code>`bar`</code> ...
 919 //
 920
 921         /*
 922                 text = text.replace(/
 923                         (^|[^\\])                                       // Character before opening ` can't be a backslash
 924                         (`+)                                            // $2 = Opening run of `
 925                         (                                                       // $3 = The code block
 926                                 [^\r]*?
 927                                 [^`]                                    // attacklab: work around lack of lookbehind
 928                         )
 929                         \2                                                      // Matching closer
 930                         (?!`)
 931                 /gm, function(){...});
 932         */
 933
 934         text = text.replace(/(^|[^\\])(`+)([^\r]*?[^`])\2(?!`)/gm,
 935                 function(wholeMatch,m1,m2,m3,m4) {
 936                         var c = m3;
 937                         c = c.replace(/^([ \t]*)/g,""); // leading whitespace
 938                         c = c.replace(/[ \t]*$/g,"");   // trailing whitespace
 939                         c = _EncodeCode(c);
 940                         return m1+"<code>"+c+"</code>";
 941                 });
 942
 943         return text;
 944 }
 945
 946
 947 var _EncodeCode = function(text) {
 948 //
 949 // Encode/escape certain characters inside Markdown code runs.
 950 // The point is that in code, these characters are literals,
 951 // and lose their special Markdown meanings.
 952 //
 953         // Encode all ampersands; HTML entities are not
 954         // entities within a Markdown code span.
 955         text = text.replace(/&/g,"&amp;");
 956
 957         // Do the angle bracket song and dance:
 958         text = text.replace(/</g,"&lt;");
 959         text = text.replace(/>/g,"&gt;");
 960
 961         // Now, escape characters that are magic in Markdown:
 962         text = escapeCharacters(text,"\*_{}[]\\",false);
 963
 964 // jj the line above breaks this:
 965 //---
 966
 967 //* Item
 968
 969 //   1. Subitem
 970
 971 //            special char: *
 972 //---
 973
 974         return text;
 975 }
 976
 977
 978 var _DoItalicsAndBold = function(text) {
 979
 980         // <strong> must go first:
 981         text = text.replace(/(\*\*|__)(?=\S)([^\r]*?\S[\*_]*)\1/g,
 982                 "<strong>$2</strong>");
 983
 984         text = text.replace(/(\*|_)(?=\S)([^\r]*?\S)\1/g,
 985                 "<em>$2</em>");
 986
 987         return text;
 988 }
 989
 990
 991 var _DoBlockQuotes = function(text) {
 992
 993         /*
 994                 text = text.replace(/
 995                 (                                                               // Wrap whole match in $1
 996                         (
 997                                 ^[ \t]*>[ \t]?                  // '>' at the start of a line
 998                                 .+\n                                    // rest of the first line
 999                                 (.+\n)*                                 // subsequent consecutive lines
1000                                 \n*                                             // blanks
1001                         )+
1002                 )
1003                 /gm, function(){...});
1004         */
1005
1006         text = text.replace(/((^[ \t]*>[ \t]?.+\n(.+\n)*\n*)+)/gm,
1007                 function(wholeMatch,m1) {
1008                         var bq = m1;
1009
1010                         // attacklab: hack around Konqueror 3.5.4 bug:
1011                         // "----------bug".replace(/^-/g,"") == "bug"
1012
1013                         bq = bq.replace(/^[ \t]*>[ \t]?/gm,"~0");       // trim one level of quoting
1014
1015                         // attacklab: clean up hack
1016                         bq = bq.replace(/~0/g,"");
1017
1018                         bq = bq.replace(/^[ \t]+$/gm,"");               // trim whitespace-only lines
1019                         bq = _RunBlockGamut(bq);                                // recurse
1020
1021                         bq = bq.replace(/(^|\n)/g,"$1  ");
1022                         // These leading spaces screw with <pre> content, so we need to fix that:
1023                         bq = bq.replace(
1024                                         /(\s*<pre>[^\r]+?<\/pre>)/gm,
1025                                 function(wholeMatch,m1) {
1026                                         var pre = m1;
1027                                         // attacklab: hack around Konqueror 3.5.4 bug:
1028                                         pre = pre.replace(/^  /mg,"~0");
1029                                         pre = pre.replace(/~0/g,"");
1030                                         return pre;
1031                                 });
1032
1033                         return hashBlock("<blockquote>\n" + bq + "\n</blockquote>");
1034                 });
1035         return text;
1036 }
1037
1038
1039 var _FormParagraphs = function(text) {
1040 //
1041 //  Params:
1042 //    $text - string to process with html <p> tags
1043 //
1044
1045         // Strip leading and trailing lines:
1046         text = text.replace(/^\n+/g,"");
1047         text = text.replace(/\n+$/g,"");
1048
1049         var grafs = text.split(/\n{2,}/g);
1050         var grafsOut = new Array();
1051
1052         //
1053         // Wrap <p> tags.
1054         //
1055         var end = grafs.length;
1056         for (var i=0; i<end; i++) {
1057                 var str = grafs[i];
1058
1059                 // if this is an HTML marker, copy it
1060                 if (str.search(/~K(\d+)K/g) >= 0) {
1061                         grafsOut.push(str);
1062                 }
1063                 else if (str.search(/\S/) >= 0) {
1064                         str = _RunSpanGamut(str);
1065                         str = str.replace(/^([ \t]*)/g,"<p>");
1066                         str += "</p>"
1067                         grafsOut.push(str);
1068                 }
1069
1070         }
1071
1072         //
1073         // Unhashify HTML blocks
1074         //
1075         end = grafsOut.length;
1076         for (var i=0; i<end; i++) {
1077                 // if this is a marker for an html block...
1078                 while (grafsOut[i].search(/~K(\d+)K/) >= 0) {
1079                         var blockText = g_html_blocks[RegExp.$1];
1080                         blockText = blockText.replace(/\$/g,"$$$$"); // Escape any dollar signs
1081                         grafsOut[i] = grafsOut[i].replace(/~K\d+K/,blockText);
1082                 }
1083         }
1084
1085         return grafsOut.join("\n\n");
1086 }
1087
1088
1089 var _EncodeAmpsAndAngles = function(text) {
1090 // Smart processing for ampersands and angle brackets that need to be encoded.
1091
1092         // Ampersand-encoding based entirely on Nat Irons's Amputator MT plugin:
1093         //   http://bumppo.net/projects/amputator/
1094         text = text.replace(/&(?!#?[xX]?(?:[0-9a-fA-F]+|\w+);)/g,"&amp;");
1095
1096         // Encode naked <'s
1097         text = text.replace(/<(?![a-z\/?\$!])/gi,"&lt;");
1098
1099         return text;
1100 }
1101
1102
1103 var _EncodeBackslashEscapes = function(text) {
1104 //
1105 //   Parameter:  String.
1106 //   Returns:   The string, with after processing the following backslash
1107 //                         escape sequences.
1108 //
1109
1110         // attacklab: The polite way to do this is with the new
1111         // escapeCharacters() function:
1112         //
1113         //      text = escapeCharacters(text,"\\",true);
1114         //      text = escapeCharacters(text,"`*_{}[]()>#+-.!",true);
1115         //
1116         // ...but we're sidestepping its use of the (slow) RegExp constructor
1117         // as an optimization for Firefox.  This function gets called a LOT.
1118
1119         text = text.replace(/\\(\\)/g,escapeCharacters_callback);
1120         text = text.replace(/\\([`*_{}\[\]()>#+-.!])/g,escapeCharacters_callback);
1121         return text;
1122 }
1123
1124
1125 var _DoAutoLinks = function(text) {
1126
1127         text = text.replace(/<((https?|ftp|dict):[^'">\s]+)>/gi,"<a href=\"$1\">$1</a>");
1128
1129         // Email addresses: <address@domain.foo>
1130
1131         /*
1132                 text = text.replace(/
1133                         <
1134                         (?:mailto:)?
1135                         (
1136                                 [-.\w]+
1137                                 \@
1138                                 [-a-z0-9]+(\.[-a-z0-9]+)*\.[a-z]+
1139                         )
1140                         >
1141                 /gi, _DoAutoLinks_callback());
1142         */
1143         text = text.replace(/<(?:mailto:)?([-.\w]+\@[-a-z0-9]+(\.[-a-z0-9]+)*\.[a-z]+)>/gi,
1144                 function(wholeMatch,m1) {
1145                         return _EncodeEmailAddress( _UnescapeSpecialChars(m1) );
1146                 }
1147         );
1148
1149         return text;
1150 }
1151
1152
1153 var _EncodeEmailAddress = function(addr) {
1154 //
1155 //  Input: an email address, e.g. "foo@example.com"
1156 //
1157 //  Output: the email address as a mailto link, with each character
1158 //      of the address encoded as either a decimal or hex entity, in
1159 //      the hopes of foiling most address harvesting spam bots. E.g.:
1160 //
1161 //      <a href="&#x6D;&#97;&#105;&#108;&#x74;&#111;:&#102;&#111;&#111;&#64;&#101;
1162 //         x&#x61;&#109;&#x70;&#108;&#x65;&#x2E;&#99;&#111;&#109;">&#102;&#111;&#111;
1163 //         &#64;&#101;x&#x61;&#109;&#x70;&#108;&#x65;&#x2E;&#99;&#111;&#109;</a>
1164 //
1165 //  Based on a filter by Matthew Wickline, posted to the BBEdit-Talk
1166 //  mailing list: <http://tinyurl.com/yu7ue>
1167 //
1168
1169         // attacklab: why can't javascript speak hex?
1170         function char2hex(ch) {
1171                 var hexDigits = '0123456789ABCDEF';
1172                 var dec = ch.charCodeAt(0);
1173                 return(hexDigits.charAt(dec>>4) + hexDigits.charAt(dec&15));
1174         }
1175
1176         var encode = [
1177                 function(ch){return "&#"+ch.charCodeAt(0)+";";},
1178                 function(ch){return "&#x"+char2hex(ch)+";";},
1179                 function(ch){return ch;}
1180         ];
1181
1182         addr = "mailto:" + addr;
1183
1184         addr = addr.replace(/./g, function(ch) {
1185                 if (ch == "@") {
1186                         // this *must* be encoded. I insist.
1187                         ch = encode[Math.floor(Math.random()*2)](ch);
1188                 } else if (ch !=":") {
1189                         // leave ':' alone (to spot mailto: later)
1190                         var r = Math.random();
1191                         // roughly 10% raw, 45% hex, 45% dec
1192                         ch =  (
1193                                         r > .9  ?       encode[2](ch)   :
1194                                         r > .45 ?       encode[1](ch)   :
1195                                                                 encode[0](ch)
1196                                 );
1197                 }
1198                 return ch;
1199         });
1200
1201         addr = "<a href=\"" + addr + "\">" + addr + "</a>";
1202         addr = addr.replace(/">.+:/g,"\">"); // strip the mailto: from the visible part
1203
1204         return addr;
1205 }
1206
1207
1208 var _UnescapeSpecialChars = function(text) {
1209 //
1210 // Swap back in all the special characters we've hidden.
1211 //
1212         text = text.replace(/~E(\d+)E/g,
1213                 function(wholeMatch,m1) {
1214                         var charCodeToReplace = parseInt(m1);
1215                         return String.fromCharCode(charCodeToReplace);
1216                 }
1217         );
1218         return text;
1219 }
1220
1221
1222 var _Outdent = function(text) {
1223 //
1224 // Remove one level of line-leading tabs or spaces
1225 //
1226
1227         // attacklab: hack around Konqueror 3.5.4 bug:
1228         // "----------bug".replace(/^-/g,"") == "bug"
1229
1230         text = text.replace(/^(\t|[ ]{1,4})/gm,"~0"); // attacklab: g_tab_width
1231
1232         // attacklab: clean up hack
1233         text = text.replace(/~0/g,"")
1234
1235         return text;
1236 }
1237
1238 var _Detab = function(text) {
1239 // attacklab: Detab's completely rewritten for speed.
1240 // In perl we could fix it by anchoring the regexp with \G.
1241 // In javascript we're less fortunate.
1242
1243         // expand first n-1 tabs
1244         text = text.replace(/\t(?=\t)/g,"    "); // attacklab: g_tab_width
1245
1246         // replace the nth with two sentinels
1247         text = text.replace(/\t/g,"~A~B");
1248
1249         // use the sentinel to anchor our regex so it doesn't explode
1250         text = text.replace(/~B(.+?)~A/g,
1251                 function(wholeMatch,m1,m2) {
1252                         var leadingText = m1;
1253                         var numSpaces = 4 - leadingText.length % 4;  // attacklab: g_tab_width
1254
1255                         // there *must* be a better way to do this:
1256                         for (var i=0; i<numSpaces; i++) leadingText+=" ";
1257
1258                         return leadingText;
1259                 }
1260         );
1261
1262         // clean up sentinels
1263         text = text.replace(/~A/g,"    ");  // attacklab: g_tab_width
1264         text = text.replace(/~B/g,"");
1265
1266         return text;
1267 }
1268
1269
1270 //
1271 //  attacklab: Utility functions
1272 //
1273
1274
1275 var escapeCharacters = function(text, charsToEscape, afterBackslash) {
1276         // First we have to escape the escape characters so that
1277         // we can build a character class out of them
1278         var regexString = "([" + charsToEscape.replace(/([\[\]\\])/g,"\\$1") + "])";
1279
1280         if (afterBackslash) {
1281                 regexString = "\\\\" + regexString;
1282         }
1283
1284         var regex = new RegExp(regexString,"g");
1285         text = text.replace(regex,escapeCharacters_callback);
1286
1287         return text;
1288 }
1289
1290
1291 var escapeCharacters_callback = function(wholeMatch,m1) {
1292         var charCodeToEscape = m1.charCodeAt(0);
1293         return "~E"+charCodeToEscape+"E";
1294 }
1295
1296 } // end of Attacklab.showdown.converter
1297
1298
1299 // Version 0.9 used the Showdown namespace instead of Attacklab.showdown
1300 // The old namespace is deprecated, but we'll support it for now:
1301 var Showdown = Attacklab.showdown;
1302
1303 // If anyone's interested, tell the world that this file's been loaded
1304 if (Attacklab.fileLoaded) {
1305         Attacklab.fileLoaded("showdown.js");
1306 }