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     // Prevent executing JavaScript from the Anchor href.
 502     text = text.replace(/(<a.*href=[\"|\']javascript\:([^"]+)[\"|\'].*>([^<]+)<\/a>)/g, function() {
 503         return arguments[3];
 504     });
 505
 506         return text;
 507 }
 508
 509 var writeAnchorTag = function(wholeMatch,m1,m2,m3,m4,m5,m6,m7) {
 510         if (m7 == undefined) m7 = "";
 511         var whole_match = m1;
 512         var link_text   = m2;
 513         var link_id      = m3.toLowerCase();
 514         var url         = m4;
 515         var title       = m7;
 516
 517         if (url == "") {
 518                 if (link_id == "") {
 519                         // lower-case and turn embedded newlines into spaces
 520                         link_id = link_text.toLowerCase().replace(/ ?\n/g," ");
 521                 }
 522                 url = "#"+link_id;
 523
 524                 if (g_urls[link_id] != undefined) {
 525                         url = g_urls[link_id];
 526                         if (g_titles[link_id] != undefined) {
 527                                 title = g_titles[link_id];
 528                         }
 529                 }
 530                 else {
 531                         if (whole_match.search(/\(\s*\)$/m)>-1) {
 532                                 // Special case for explicit empty url
 533                                 url = "";
 534                         } else {
 535                                 return whole_match;
 536                         }
 537                 }
 538         }
 539
 540         url = escapeCharacters(url,"*_");
 541         var result = "<a href=\"" + url + "\"";
 542
 543         if (title != "") {
 544                 title = title.replace(/"/g,"&quot;");
 545                 title = escapeCharacters(title,"*_");
 546                 result +=  " title=\"" + title + "\"";
 547         }
 548
 549         result += ">" + link_text + "</a>";
 550
 551         return result;
 552 }
 553
 554
 555 var _DoImages = function(text) {
 556 //
 557 // Turn Markdown image shortcuts into <img> tags.
 558 //
 559
 560         //
 561         // First, handle reference-style labeled images: ![alt text][id]
 562         //
 563
 564         /*
 565                 text = text.replace(/
 566                 (                                               // wrap whole match in $1
 567                         !\[
 568                         (.*?)                           // alt text = $2
 569                         \]
 570
 571                         [ ]?                            // one optional space
 572                         (?:\n[ ]*)?                     // one optional newline followed by spaces
 573
 574                         \[
 575                         (.*?)                           // id = $3
 576                         \]
 577                 )()()()()                               // pad rest of backreferences
 578                 /g,writeImageTag);
 579         */
 580         text = text.replace(/(!\[(.*?)\][ ]?(?:\n[ ]*)?\[(.*?)\])()()()()/g,writeImageTag);
 581
 582         //
 583         // Next, handle inline images:  ![alt text](url "optional title")
 584         // Don't forget: encode * and _
 585
 586         /*
 587                 text = text.replace(/
 588                 (                                               // wrap whole match in $1
 589                         !\[
 590                         (.*?)                           // alt text = $2
 591                         \]
 592                         \s?                                     // One optional whitespace character
 593                         \(                                      // literal paren
 594                         [ \t]*
 595                         ()                                      // no id, so leave $3 empty
 596                         <?(\S+?)>?                      // src url = $4
 597                         [ \t]*
 598                         (                                       // $5
 599                                 (['"])                  // quote char = $6
 600                                 (.*?)                   // title = $7
 601                                 \6                              // matching quote
 602                                 [ \t]*
 603                         )?                                      // title is optional
 604                 \)
 605                 )
 606                 /g,writeImageTag);
 607         */
 608         text = text.replace(/(!\[(.*?)\]\s?\([ \t]*()<?(\S+?)>?[ \t]*((['"])(.*?)\6[ \t]*)?\))/g,writeImageTag);
 609
 610         return text;
 611 }
 612
 613 var writeImageTag = function(wholeMatch,m1,m2,m3,m4,m5,m6,m7) {
 614         var whole_match = m1;
 615         var alt_text   = m2;
 616         var link_id      = m3.toLowerCase();
 617         var url         = m4;
 618         var title       = m7;
 619
 620         if (!title) title = "";
 621
 622         if (url == "") {
 623                 if (link_id == "") {
 624                         // lower-case and turn embedded newlines into spaces
 625                         link_id = alt_text.toLowerCase().replace(/ ?\n/g," ");
 626                 }
 627                 url = "#"+link_id;
 628
 629                 if (g_urls[link_id] != undefined) {
 630                         url = g_urls[link_id];
 631                         if (g_titles[link_id] != undefined) {
 632                                 title = g_titles[link_id];
 633                         }
 634                 }
 635                 else {
 636                         return whole_match;
 637                 }
 638         }
 639
 640         alt_text = alt_text.replace(/"/g,"&quot;");
 641         url = escapeCharacters(url,"*_");
 642     if (url.toString().indexOf('http://') != 0 && url.toString().indexOf('https://') != 0) {
 643         url = scriptUrl + url
 644     }
 645         var result = "<img src=\"" + url + "\" alt=\"" + alt_text + "\"";
 646
 647         // attacklab: Markdown.pl adds empty title attributes to images.
 648         // Replicate this bug.
 649
 650         //if (title != "") {
 651                 title = title.replace(/"/g,"&quot;");
 652                 title = escapeCharacters(title,"*_");
 653                 result +=  " title=\"" + title + "\"";
 654         //}
 655
 656         result += " />";
 657
 658         return result;
 659 }
 660
 661
 662 var _DoHeaders = function(text) {
 663
 664         // Setext-style headers:
 665         //      Header 1
 666         //      ========
 667         //
 668         //      Header 2
 669         //      --------
 670         //
 671         text = text.replace(/^(.+)[ \t]*\n=+[ \t]*\n+/gm,
 672                 function(wholeMatch,m1){return hashBlock("<h1>" + _RunSpanGamut(m1) + "</h1>");});
 673
 674         text = text.replace(/^(.+)[ \t]*\n-+[ \t]*\n+/gm,
 675                 function(matchFound,m1){return hashBlock("<h2>" + _RunSpanGamut(m1) + "</h2>");});
 676
 677         // atx-style headers:
 678         //  # Header 1
 679         //  ## Header 2
 680         //  ## Header 2 with closing hashes ##
 681         //  ...
 682         //  ###### Header 6
 683         //
 684
 685         /*
 686                 text = text.replace(/
 687                         ^(\#{1,6})                              // $1 = string of #'s
 688                         [ \t]*
 689                         (.+?)                                   // $2 = Header text
 690                         [ \t]*
 691                         \#*                                             // optional closing #'s (not counted)
 692                         \n+
 693                 /gm, function() {...});
 694         */
 695
 696         text = text.replace(/^(\#{1,6})[ \t]*(.+?)[ \t]*\#*\n+/gm,
 697                 function(wholeMatch,m1,m2) {
 698                         var h_level = m1.length;
 699                         return hashBlock("<h" + h_level + ">" + _RunSpanGamut(m2) + "</h" + h_level + ">");
 700                 });
 701
 702         return text;
 703 }
 704
 705 // This declaration keeps Dojo compressor from outputting garbage:
 706 var _ProcessListItems;
 707
 708 var _DoLists = function(text) {
 709 //
 710 // Form HTML ordered (numbered) and unordered (bulleted) lists.
 711 //
 712
 713         // attacklab: add sentinel to hack around khtml/safari bug:
 714         // http://bugs.webkit.org/show_bug.cgi?id=11231
 715         text += "~0";
 716
 717         // Re-usable pattern to match any entirel ul or ol list:
 718
 719         /*
 720                 var whole_list = /
 721                 (                                                                       // $1 = whole list
 722                         (                                                               // $2
 723                                 [ ]{0,3}                                        // attacklab: g_tab_width - 1
 724                                 ([*+-]|\d+[.])                          // $3 = first list item marker
 725                                 [ \t]+
 726                         )
 727                         [^\r]+?
 728                         (                                                               // $4
 729                                 ~0                                                      // sentinel for workaround; should be $
 730                         |
 731                                 \n{2,}
 732                                 (?=\S)
 733                                 (?!                                                     // Negative lookahead for another list item marker
 734                                         [ \t]*
 735                                         (?:[*+-]|\d+[.])[ \t]+
 736                                 )
 737                         )
 738                 )/g
 739         */
 740         var whole_list = /^(([ ]{0,3}([*+-]|\d+[.])[ \t]+)[^\r]+?(~0|\n{2,}(?=\S)(?![ \t]*(?:[*+-]|\d+[.])[ \t]+)))/gm;
 741
 742         if (g_list_level) {
 743                 text = text.replace(whole_list,function(wholeMatch,m1,m2) {
 744                         var list = m1;
 745                         var list_type = (m2.search(/[*+-]/g)>-1) ? "ul" : "ol";
 746
 747                         // Turn double returns into triple returns, so that we can make a
 748                         // paragraph for the last item in a list, if necessary:
 749                         list = list.replace(/\n{2,}/g,"\n\n\n");;
 750                         var result = _ProcessListItems(list);
 751
 752                         // Trim any trailing whitespace, to put the closing `</$list_type>`
 753                         // up on the preceding line, to get it past the current stupid
 754                         // HTML block parser. This is a hack to work around the terrible
 755                         // hack that is the HTML block parser.
 756                         result = result.replace(/\s+$/,"");
 757                         result = "<"+list_type+">" + result + "</"+list_type+">\n";
 758                         return result;
 759                 });
 760         } else {
 761                 whole_list = /(\n\n|^\n?)(([ ]{0,3}([*+-]|\d+[.])[ \t]+)[^\r]+?(~0|\n{2,}(?=\S)(?![ \t]*(?:[*+-]|\d+[.])[ \t]+)))/g;
 762                 text = text.replace(whole_list,function(wholeMatch,m1,m2,m3) {
 763                         var runup = m1;
 764                         var list = m2;
 765
 766                         var list_type = (m3.search(/[*+-]/g)>-1) ? "ul" : "ol";
 767                         // Turn double returns into triple returns, so that we can make a
 768                         // paragraph for the last item in a list, if necessary:
 769                         var list = list.replace(/\n{2,}/g,"\n\n\n");;
 770                         var result = _ProcessListItems(list);
 771                         result = runup + "<"+list_type+">\n" + result + "</"+list_type+">\n";
 772                         return result;
 773                 });
 774         }
 775
 776         // attacklab: strip sentinel
 777         text = text.replace(/~0/,"");
 778
 779         return text;
 780 }
 781
 782 _ProcessListItems = function(list_str) {
 783 //
 784 //  Process the contents of a single ordered or unordered list, splitting it
 785 //  into individual list items.
 786 //
 787         // The $g_list_level global keeps track of when we're inside a list.
 788         // Each time we enter a list, we increment it; when we leave a list,
 789         // we decrement. If it's zero, we're not in a list anymore.
 790         //
 791         // We do this because when we're not inside a list, we want to treat
 792         // something like this:
 793         //
 794         //    I recommend upgrading to version
 795         //    8. Oops, now this line is treated
 796         //    as a sub-list.
 797         //
 798         // As a single paragraph, despite the fact that the second line starts
 799         // with a digit-period-space sequence.
 800         //
 801         // Whereas when we're inside a list (or sub-list), that line will be
 802         // treated as the start of a sub-list. What a kludge, huh? This is
 803         // an aspect of Markdown's syntax that's hard to parse perfectly
 804         // without resorting to mind-reading. Perhaps the solution is to
 805         // change the syntax rules such that sub-lists must start with a
 806         // starting cardinal number; e.g. "1." or "a.".
 807
 808         g_list_level++;
 809
 810         // trim trailing blank lines:
 811         list_str = list_str.replace(/\n{2,}$/,"\n");
 812
 813         // attacklab: add sentinel to emulate \z
 814         list_str += "~0";
 815
 816         /*
 817                 list_str = list_str.replace(/
 818                         (\n)?                                                   // leading line = $1
 819                         (^[ \t]*)                                               // leading whitespace = $2
 820                         ([*+-]|\d+[.]) [ \t]+                   // list marker = $3
 821                         ([^\r]+?                                                // list item text   = $4
 822                         (\n{1,2}))
 823                         (?= \n* (~0 | \2 ([*+-]|\d+[.]) [ \t]+))
 824                 /gm, function(){...});
 825         */
 826         list_str = list_str.replace(/(\n)?(^[ \t]*)([*+-]|\d+[.])[ \t]+([^\r]+?(\n{1,2}))(?=\n*(~0|\2([*+-]|\d+[.])[ \t]+))/gm,
 827                 function(wholeMatch,m1,m2,m3,m4){
 828                         var item = m4;
 829                         var leading_line = m1;
 830                         var leading_space = m2;
 831
 832                         if (leading_line || (item.search(/\n{2,}/)>-1)) {
 833                                 item = _RunBlockGamut(_Outdent(item));
 834                         }
 835                         else {
 836                                 // Recursion for sub-lists:
 837                                 item = _DoLists(_Outdent(item));
 838                                 item = item.replace(/\n$/,""); // chomp(item)
 839                                 item = _RunSpanGamut(item);
 840                         }
 841
 842                         return  "<li>" + item + "</li>\n";
 843                 }
 844         );
 845
 846         // attacklab: strip sentinel
 847         list_str = list_str.replace(/~0/g,"");
 848
 849         g_list_level--;
 850         return list_str;
 851 }
 852
 853
 854 var _DoCodeBlocks = function(text) {
 855 //
 856 //  Process Markdown `<pre><code>` blocks.
 857 //
 858
 859         /*
 860                 text = text.replace(text,
 861                         /(?:\n\n|^)
 862                         (                                                               // $1 = the code block -- one or more lines, starting with a space/tab
 863                                 (?:
 864                                         (?:[ ]{4}|\t)                   // Lines must start with a tab or a tab-width of spaces - attacklab: g_tab_width
 865                                         .*\n+
 866                                 )+
 867                         )
 868                         (\n*[ ]{0,3}[^ \t\n]|(?=~0))    // attacklab: g_tab_width
 869                 /g,function(){...});
 870         */
 871
 872         // attacklab: sentinel workarounds for lack of \A and \Z, safari\khtml bug
 873         text += "~0";
 874
 875         text = text.replace(/(?:\n\n|^)((?:(?:[ ]{4}|\t).*\n+)+)(\n*[ ]{0,3}[^ \t\n]|(?=~0))/g,
 876                 function(wholeMatch,m1,m2) {
 877                         var codeblock = m1;
 878                         var nextChar = m2;
 879
 880                         codeblock = _EncodeCode( _Outdent(codeblock));
 881                         codeblock = _Detab(codeblock);
 882                         codeblock = codeblock.replace(/^\n+/g,""); // trim leading newlines
 883                         codeblock = codeblock.replace(/\n+$/g,""); // trim trailing whitespace
 884
 885                         codeblock = "<pre><code>" + codeblock + "\n</code></pre>";
 886
 887                         return hashBlock(codeblock) + nextChar;
 888                 }
 889         );
 890
 891         // attacklab: strip sentinel
 892         text = text.replace(/~0/,"");
 893
 894         return text;
 895 }
 896
 897 var hashBlock = function(text) {
 898         text = text.replace(/(^\n+|\n+$)/g,"");
 899         return "\n\n~K" + (g_html_blocks.push(text)-1) + "K\n\n";
 900 }
 901
 902
 903 var _DoCodeSpans = function(text) {
 904 //
 905 //   *  Backtick quotes are used for <code></code> spans.
 906 //
 907 //   *  You can use multiple backticks as the delimiters if you want to
 908 //       include literal backticks in the code span. So, this input:
 909 //
 910 //               Just type ``foo `bar` baz`` at the prompt.
 911 //
 912 //         Will translate to:
 913 //
 914 //               <p>Just type <code>foo `bar` baz</code> at the prompt.</p>
 915 //
 916 //      There's no arbitrary limit to the number of backticks you
 917 //      can use as delimters. If you need three consecutive backticks
 918 //      in your code, use four for delimiters, etc.
 919 //
 920 //  *  You can use spaces to get literal backticks at the edges:
 921 //
 922 //               ... type `` `bar` `` ...
 923 //
 924 //         Turns to:
 925 //
 926 //               ... type <code>`bar`</code> ...
 927 //
 928
 929         /*
 930                 text = text.replace(/
 931                         (^|[^\\])                                       // Character before opening ` can't be a backslash
 932                         (`+)                                            // $2 = Opening run of `
 933                         (                                                       // $3 = The code block
 934                                 [^\r]*?
 935                                 [^`]                                    // attacklab: work around lack of lookbehind
 936                         )
 937                         \2                                                      // Matching closer
 938                         (?!`)
 939                 /gm, function(){...});
 940         */
 941
 942         text = text.replace(/(^|[^\\])(`+)([^\r]*?[^`])\2(?!`)/gm,
 943                 function(wholeMatch,m1,m2,m3,m4) {
 944                         var c = m3;
 945                         c = c.replace(/^([ \t]*)/g,""); // leading whitespace
 946                         c = c.replace(/[ \t]*$/g,"");   // trailing whitespace
 947                         c = _EncodeCode(c);
 948                         return m1+"<code>"+c+"</code>";
 949                 });
 950
 951         return text;
 952 }
 953
 954
 955 var _EncodeCode = function(text) {
 956 //
 957 // Encode/escape certain characters inside Markdown code runs.
 958 // The point is that in code, these characters are literals,
 959 // and lose their special Markdown meanings.
 960 //
 961         // Encode all ampersands; HTML entities are not
 962         // entities within a Markdown code span.
 963         text = text.replace(/&/g,"&amp;");
 964
 965         // Do the angle bracket song and dance:
 966         text = text.replace(/</g,"&lt;");
 967         text = text.replace(/>/g,"&gt;");
 968
 969         // Now, escape characters that are magic in Markdown:
 970         text = escapeCharacters(text,"\*_{}[]\\",false);
 971
 972 // jj the line above breaks this:
 973 //---
 974
 975 //* Item
 976
 977 //   1. Subitem
 978
 979 //            special char: *
 980 //---
 981
 982         return text;
 983 }
 984
 985
 986 var _DoItalicsAndBold = function(text) {
 987
 988         // <strong> must go first:
 989         text = text.replace(/(\*\*|__)(?=\S)([^\r]*?\S[\*_]*)\1/g,
 990                 "<strong>$2</strong>");
 991
 992         text = text.replace(/(\*|_)(?=\S)([^\r]*?\S)\1/g,
 993                 "<em>$2</em>");
 994
 995         return text;
 996 }
 997
 998
 999 var _DoBlockQuotes = function(text) {
1000
1001         /*
1002                 text = text.replace(/
1003                 (                                                               // Wrap whole match in $1
1004                         (
1005                                 ^[ \t]*>[ \t]?                  // '>' at the start of a line
1006                                 .+\n                                    // rest of the first line
1007                                 (.+\n)*                                 // subsequent consecutive lines
1008                                 \n*                                             // blanks
1009                         )+
1010                 )
1011                 /gm, function(){...});
1012         */
1013
1014         text = text.replace(/((^[ \t]*>[ \t]?.+\n(.+\n)*\n*)+)/gm,
1015                 function(wholeMatch,m1) {
1016                         var bq = m1;
1017
1018                         // attacklab: hack around Konqueror 3.5.4 bug:
1019                         // "----------bug".replace(/^-/g,"") == "bug"
1020
1021                         bq = bq.replace(/^[ \t]*>[ \t]?/gm,"~0");       // trim one level of quoting
1022
1023                         // attacklab: clean up hack
1024                         bq = bq.replace(/~0/g,"");
1025
1026                         bq = bq.replace(/^[ \t]+$/gm,"");               // trim whitespace-only lines
1027                         bq = _RunBlockGamut(bq);                                // recurse
1028
1029                         bq = bq.replace(/(^|\n)/g,"$1  ");
1030                         // These leading spaces screw with <pre> content, so we need to fix that:
1031                         bq = bq.replace(
1032                                         /(\s*<pre>[^\r]+?<\/pre>)/gm,
1033                                 function(wholeMatch,m1) {
1034                                         var pre = m1;
1035                                         // attacklab: hack around Konqueror 3.5.4 bug:
1036                                         pre = pre.replace(/^  /mg,"~0");
1037                                         pre = pre.replace(/~0/g,"");
1038                                         return pre;
1039                                 });
1040
1041                         return hashBlock("<blockquote>\n" + bq + "\n</blockquote>");
1042                 });
1043         return text;
1044 }
1045
1046
1047 var _FormParagraphs = function(text) {
1048 //
1049 //  Params:
1050 //    $text - string to process with html <p> tags
1051 //
1052
1053         // Strip leading and trailing lines:
1054         text = text.replace(/^\n+/g,"");
1055         text = text.replace(/\n+$/g,"");
1056
1057         var grafs = text.split(/\n{2,}/g);
1058         var grafsOut = new Array();
1059
1060         //
1061         // Wrap <p> tags.
1062         //
1063         var end = grafs.length;
1064         for (var i=0; i<end; i++) {
1065                 var str = grafs[i];
1066
1067                 // if this is an HTML marker, copy it
1068                 if (str.search(/~K(\d+)K/g) >= 0) {
1069                         grafsOut.push(str);
1070                 }
1071                 else if (str.search(/\S/) >= 0) {
1072                         str = _RunSpanGamut(str);
1073                         str = str.replace(/^([ \t]*)/g,"<p>");
1074                         str += "</p>"
1075                         grafsOut.push(str);
1076                 }
1077
1078         }
1079
1080         //
1081         // Unhashify HTML blocks
1082         //
1083         end = grafsOut.length;
1084         for (var i=0; i<end; i++) {
1085                 // if this is a marker for an html block...
1086                 while (grafsOut[i].search(/~K(\d+)K/) >= 0) {
1087                         var blockText = g_html_blocks[RegExp.$1];
1088                         blockText = blockText.replace(/\$/g,"$$$$"); // Escape any dollar signs
1089                         grafsOut[i] = grafsOut[i].replace(/~K\d+K/,blockText);
1090                 }
1091         }
1092
1093         return grafsOut.join("\n\n");
1094 }
1095
1096
1097 var _EncodeAmpsAndAngles = function(text) {
1098 // Smart processing for ampersands and angle brackets that need to be encoded.
1099
1100         // Ampersand-encoding based entirely on Nat Irons's Amputator MT plugin:
1101         //   http://bumppo.net/projects/amputator/
1102         text = text.replace(/&(?!#?[xX]?(?:[0-9a-fA-F]+|\w+);)/g,"&amp;");
1103
1104         // Encode naked <'s
1105         text = text.replace(/<(?![a-z\/?\$!])/gi,"&lt;");
1106
1107         return text;
1108 }
1109
1110
1111 var _EncodeBackslashEscapes = function(text) {
1112 //
1113 //   Parameter:  String.
1114 //   Returns:   The string, with after processing the following backslash
1115 //                         escape sequences.
1116 //
1117
1118         // attacklab: The polite way to do this is with the new
1119         // escapeCharacters() function:
1120         //
1121         //      text = escapeCharacters(text,"\\",true);
1122         //      text = escapeCharacters(text,"`*_{}[]()>#+-.!",true);
1123         //
1124         // ...but we're sidestepping its use of the (slow) RegExp constructor
1125         // as an optimization for Firefox.  This function gets called a LOT.
1126
1127         text = text.replace(/\\(\\)/g,escapeCharacters_callback);
1128         text = text.replace(/\\([`*_{}\[\]()>#+-.!])/g,escapeCharacters_callback);
1129         return text;
1130 }
1131
1132
1133 var _DoAutoLinks = function(text) {
1134
1135         text = text.replace(/<((https?|ftp|dict):[^'">\s]+)>/gi,"<a href=\"$1\">$1</a>");
1136
1137         // Email addresses: <address@domain.foo>
1138
1139         /*
1140                 text = text.replace(/
1141                         <
1142                         (?:mailto:)?
1143                         (
1144                                 [-.\w]+
1145                                 \@
1146                                 [-a-z0-9]+(\.[-a-z0-9]+)*\.[a-z]+
1147                         )
1148                         >
1149                 /gi, _DoAutoLinks_callback());
1150         */
1151         text = text.replace(/<(?:mailto:)?([-.\w]+\@[-a-z0-9]+(\.[-a-z0-9]+)*\.[a-z]+)>/gi,
1152                 function(wholeMatch,m1) {
1153                         return _EncodeEmailAddress( _UnescapeSpecialChars(m1) );
1154                 }
1155         );
1156
1157         return text;
1158 }
1159
1160
1161 var _EncodeEmailAddress = function(addr) {
1162 //
1163 //  Input: an email address, e.g. "foo@example.com"
1164 //
1165 //  Output: the email address as a mailto link, with each character
1166 //      of the address encoded as either a decimal or hex entity, in
1167 //      the hopes of foiling most address harvesting spam bots. E.g.:
1168 //
1169 //      <a href="&#x6D;&#97;&#105;&#108;&#x74;&#111;:&#102;&#111;&#111;&#64;&#101;
1170 //         x&#x61;&#109;&#x70;&#108;&#x65;&#x2E;&#99;&#111;&#109;">&#102;&#111;&#111;
1171 //         &#64;&#101;x&#x61;&#109;&#x70;&#108;&#x65;&#x2E;&#99;&#111;&#109;</a>
1172 //
1173 //  Based on a filter by Matthew Wickline, posted to the BBEdit-Talk
1174 //  mailing list: <http://tinyurl.com/yu7ue>
1175 //
1176
1177         // attacklab: why can't javascript speak hex?
1178         function char2hex(ch) {
1179                 var hexDigits = '0123456789ABCDEF';
1180                 var dec = ch.charCodeAt(0);
1181                 return(hexDigits.charAt(dec>>4) + hexDigits.charAt(dec&15));
1182         }
1183
1184         var encode = [
1185                 function(ch){return "&#"+ch.charCodeAt(0)+";";},
1186                 function(ch){return "&#x"+char2hex(ch)+";";},
1187                 function(ch){return ch;}
1188         ];
1189
1190         addr = "mailto:" + addr;
1191
1192         addr = addr.replace(/./g, function(ch) {
1193                 if (ch == "@") {
1194                         // this *must* be encoded. I insist.
1195                         ch = encode[Math.floor(Math.random()*2)](ch);
1196                 } else if (ch !=":") {
1197                         // leave ':' alone (to spot mailto: later)
1198                         var r = Math.random();
1199                         // roughly 10% raw, 45% hex, 45% dec
1200                         ch =  (
1201                                         r > .9  ?       encode[2](ch)   :
1202                                         r > .45 ?       encode[1](ch)   :
1203                                                                 encode[0](ch)
1204                                 );
1205                 }
1206                 return ch;
1207         });
1208
1209         addr = "<a href=\"" + addr + "\">" + addr + "</a>";
1210         addr = addr.replace(/">.+:/g,"\">"); // strip the mailto: from the visible part
1211
1212         return addr;
1213 }
1214
1215
1216 var _UnescapeSpecialChars = function(text) {
1217 //
1218 // Swap back in all the special characters we've hidden.
1219 //
1220         text = text.replace(/~E(\d+)E/g,
1221                 function(wholeMatch,m1) {
1222                         var charCodeToReplace = parseInt(m1);
1223                         return String.fromCharCode(charCodeToReplace);
1224                 }
1225         );
1226         return text;
1227 }
1228
1229
1230 var _Outdent = function(text) {
1231 //
1232 // Remove one level of line-leading tabs or spaces
1233 //
1234
1235         // attacklab: hack around Konqueror 3.5.4 bug:
1236         // "----------bug".replace(/^-/g,"") == "bug"
1237
1238         text = text.replace(/^(\t|[ ]{1,4})/gm,"~0"); // attacklab: g_tab_width
1239
1240         // attacklab: clean up hack
1241         text = text.replace(/~0/g,"")
1242
1243         return text;
1244 }
1245
1246 var _Detab = function(text) {
1247 // attacklab: Detab's completely rewritten for speed.
1248 // In perl we could fix it by anchoring the regexp with \G.
1249 // In javascript we're less fortunate.
1250
1251         // expand first n-1 tabs
1252         text = text.replace(/\t(?=\t)/g,"    "); // attacklab: g_tab_width
1253
1254         // replace the nth with two sentinels
1255         text = text.replace(/\t/g,"~A~B");
1256
1257         // use the sentinel to anchor our regex so it doesn't explode
1258         text = text.replace(/~B(.+?)~A/g,
1259                 function(wholeMatch,m1,m2) {
1260                         var leadingText = m1;
1261                         var numSpaces = 4 - leadingText.length % 4;  // attacklab: g_tab_width
1262
1263                         // there *must* be a better way to do this:
1264                         for (var i=0; i<numSpaces; i++) leadingText+=" ";
1265
1266                         return leadingText;
1267                 }
1268         );
1269
1270         // clean up sentinels
1271         text = text.replace(/~A/g,"    ");  // attacklab: g_tab_width
1272         text = text.replace(/~B/g,"");
1273
1274         return text;
1275 }
1276
1277
1278 //
1279 //  attacklab: Utility functions
1280 //
1281
1282
1283 var escapeCharacters = function(text, charsToEscape, afterBackslash) {
1284         // First we have to escape the escape characters so that
1285         // we can build a character class out of them
1286         var regexString = "([" + charsToEscape.replace(/([\[\]\\])/g,"\\$1") + "])";
1287
1288         if (afterBackslash) {
1289                 regexString = "\\\\" + regexString;
1290         }
1291
1292         var regex = new RegExp(regexString,"g");
1293         text = text.replace(regex,escapeCharacters_callback);
1294
1295         return text;
1296 }
1297
1298
1299 var escapeCharacters_callback = function(wholeMatch,m1) {
1300         var charCodeToEscape = m1.charCodeAt(0);
1301         return "~E"+charCodeToEscape+"E";
1302 }
1303
1304 } // end of Attacklab.showdown.converter
1305
1306
1307 // Version 0.9 used the Showdown namespace instead of Attacklab.showdown
1308 // The old namespace is deprecated, but we'll support it for now:
1309 var Showdown = Attacklab.showdown;
1310
1311 // If anyone's interested, tell the world that this file's been loaded
1312 if (Attacklab.fileLoaded) {
1313         Attacklab.fileLoaded("showdown.js");
1314 }