includes/Parser.php

   1 <?php
   2 /**
   3  * File for Parser and related classes
   4  *
   5  * @package MediaWiki
   6  * @subpackage Parser
   7  */
   8
   9 /** */
  10 require_once( 'Sanitizer.php' );
  11
  12 /**
  13  * Update this version number when the ParserOutput format
  14  * changes in an incompatible way, so the parser cache
  15  * can automatically discard old data.
  16  */
  17 define( 'MW_PARSER_VERSION', '1.5.0' );
  18
  19 /**
  20  * Variable substitution O(N^2) attack
  21  *
  22  * Without countermeasures, it would be possible to attack the parser by saving
  23  * a page filled with a large number of inclusions of large pages. The size of
  24  * the generated page would be proportional to the square of the input size.
  25  * Hence, we limit the number of inclusions of any given page, thus bringing any
  26  * attack back to O(N).
  27  */
  28
  29 define( 'MAX_INCLUDE_REPEAT', 100 );
  30 define( 'MAX_INCLUDE_SIZE', 1000000 ); // 1 Million
  31
  32 define( 'RLH_FOR_UPDATE', 1 );
  33
  34 # Allowed values for $mOutputType
  35 define( 'OT_HTML', 1 );
  36 define( 'OT_WIKI', 2 );
  37 define( 'OT_MSG' , 3 );
  38
  39 # string parameter for extractTags which will cause it
  40 # to strip HTML comments in addition to regular
  41 # <XML>-style tags. This should not be anything we
  42 # may want to use in wikisyntax
  43 define( 'STRIP_COMMENTS', 'HTMLCommentStrip' );
  44
  45 # prefix for escaping, used in two functions at least
  46 define( 'UNIQ_PREFIX', 'NaodW29');
  47
  48 # Constants needed for external link processing
  49 define( 'URL_PROTOCOLS', 'http|https|ftp|irc|gopher|news|mailto' );
  50 define( 'HTTP_PROTOCOLS', 'http|https' );
  51 # Everything except bracket, space, or control characters
  52 define( 'EXT_LINK_URL_CLASS', '[^]<>"\\x00-\\x20\\x7F]' );
  53 # Including space
  54 define( 'EXT_LINK_TEXT_CLASS', '[^\]\\x00-\\x1F\\x7F]' );
  55 define( 'EXT_IMAGE_FNAME_CLASS', '[A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]' );
  56 define( 'EXT_IMAGE_EXTENSIONS', 'gif|png|jpg|jpeg' );
  57 define( 'EXT_LINK_BRACKETED',  '/\[(\b('.URL_PROTOCOLS.'):'.EXT_LINK_URL_CLASS.'+) *('.EXT_LINK_TEXT_CLASS.'*?)\]/S' );
  58 define( 'EXT_IMAGE_REGEX',
  59         '/^('.HTTP_PROTOCOLS.':)'.  # Protocol
  60         '('.EXT_LINK_URL_CLASS.'+)\\/'.  # Hostname and path
  61         '('.EXT_IMAGE_FNAME_CLASS.'+)\\.((?i)'.EXT_IMAGE_EXTENSIONS.')$/S' # Filename
  62 );
  63
  64 /**
  65  * PHP Parser
  66  *
  67  * Processes wiki markup
  68  *
  69  * <pre>
  70  * There are three main entry points into the Parser class:
  71  * parse()
  72  *   produces HTML output
  73  * preSaveTransform().
  74  *   produces altered wiki markup.
  75  * transformMsg()
  76  *   performs brace substitution on MediaWiki messages
  77  *
  78  * Globals used:
  79  *    objects:   $wgLang, $wgLinkCache
  80  *
  81  * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
  82  *
  83  * settings:
  84  *  $wgUseTex*, $wgUseDynamicDates*, $wgInterwikiMagic*,
  85  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  86  *  $wgLocaltimezone, $wgAllowSpecialInclusion*
  87  *
  88  *  * only within ParserOptions
  89  * </pre>
  90  *
  91  * @package MediaWiki
  92  */
  93 class Parser
  94 {
  95         /**#@+
  96          * @access private
  97          */
  98         # Persistent:
  99         var $mTagHooks;
 100
 101         # Cleared with clearState():
 102         var $mOutput, $mAutonumber, $mDTopen, $mStripState = array();
 103         var $mVariables, $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 104         var $mInterwikiLinkHolders, $mLinkHolders;
 105
 106         # Temporary:
 107         var $mOptions, $mTitle, $mOutputType,
 108             $mTemplates,        // cache of already loaded templates, avoids
 109                                 // multiple SQL queries for the same string
 110             $mTemplatePath;     // stores an unsorted hash of all the templates already loaded
 111                                 // in this path. Used for loop detection.
 112
 113         /**#@-*/
 114
 115         /**
 116          * Constructor
 117          *
 118          * @access public
 119          */
 120         function Parser() {
 121                 global $wgContLang;
 122                 $this->mTemplates = array();
 123                 $this->mTemplatePath = array();
 124                 $this->mTagHooks = array();
 125                 $this->clearState();
 126         }
 127
 128         /**
 129          * Clear Parser state
 130          *
 131          * @access private
 132          */
 133         function clearState() {
 134                 $this->mOutput = new ParserOutput;
 135                 $this->mAutonumber = 0;
 136                 $this->mLastSection = '';
 137                 $this->mDTopen = false;
 138                 $this->mVariables = false;
 139                 $this->mIncludeCount = array();
 140                 $this->mStripState = array();
 141                 $this->mArgStack = array();
 142                 $this->mInPre = false;
 143                 $this->mInterwikiLinkHolders = array(
 144                         'texts' => array(),
 145                         'titles' => array()
 146                 );
 147                 $this->mLinkHolders = array(
 148                         'namespaces' => array(),
 149                         'dbkeys' => array(),
 150                         'queries' => array(),
 151                         'texts' => array(),
 152                         'titles' => array()
 153                 );
 154         }
 155
 156         /**
 157          * First pass--just handle <nowiki> sections, pass the rest off
 158          * to internalParse() which does all the real work.
 159          *
 160          * @access private
 161          * @param string $text Text we want to parse
 162          * @param Title &$title A title object
 163          * @param array $options
 164          * @param boolean $linestart
 165          * @param boolean $clearState
 166          * @return ParserOutput a ParserOutput
 167          */
 168         function parse( $text, &$title, $options, $linestart = true, $clearState = true ) {
 169                 global $wgUseTidy, $wgContLang;
 170                 $fname = 'Parser::parse';
 171                 wfProfileIn( $fname );
 172
 173                 if ( $clearState ) {
 174                         $this->clearState();
 175                 }
 176
 177                 $this->mOptions = $options;
 178                 $this->mTitle =& $title;
 179                 $this->mOutputType = OT_HTML;
 180
 181                 $this->mStripState = NULL;
 182
 183                 //$text = $this->strip( $text, $this->mStripState );
 184                 // VOODOO MAGIC FIX! Sometimes the above segfaults in PHP5.
 185                 $x =& $this->mStripState;
 186                 $text = $this->strip( $text, $x );
 187
 188                 $text = $this->internalParse( $text );
 189
 190
 191                 $text = $this->unstrip( $text, $this->mStripState );
 192
 193                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 194                 $fixtags = array(
 195                         # french spaces, last one Guillemet-left
 196                         # only if there is something before the space
 197                         '/(.) (?=\\?|:|;|!|\\302\\273)/' => '\\1&nbsp;\\2',
 198                         # french spaces, Guillemet-right
 199                         '/(\\302\\253) /' => '\\1&nbsp;',
 200                         '/<center *>/i' => '<div class="center">',
 201                         '/<\\/center *>/i' => '</div>',
 202                 );
 203                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 204
 205                 # only once and last
 206                 $text = $this->doBlockLevels( $text, $linestart );
 207
 208                 $this->replaceLinkHolders( $text );
 209
 210                 # the position of the convert() call should not be changed. it
 211                 # assumes that the links are all replaces and the only thing left
 212                 # is the <nowiki> mark.
 213                 $text = $wgContLang->convert($text);
 214                 $this->mOutput->setTitleText($wgContLang->getParsedTitle());
 215
 216                 $text = $this->unstripNoWiki( $text, $this->mStripState );
 217
 218                 $text = Sanitizer::normalizeCharReferences( $text );
 219                 global $wgUseTidy;
 220                 if ($wgUseTidy) {
 221                         $text = Parser::tidy($text);
 222                 }
 223
 224                 $this->mOutput->setText( $text );
 225                 wfProfileOut( $fname );
 226                 return $this->mOutput;
 227         }
 228
 229         /**
 230          * Get a random string
 231          *
 232          * @access private
 233          * @static
 234          */
 235         function getRandomString() {
 236                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 237         }
 238
 239         /**
 240          * Replaces all occurrences of <$tag>content</$tag> in the text
 241          * with a random marker and returns the new text. the output parameter
 242          * $content will be an associative array filled with data on the form
 243          * $unique_marker => content.
 244          *
 245          * If $content is already set, the additional entries will be appended
 246          * If $tag is set to STRIP_COMMENTS, the function will extract
 247          * <!-- HTML comments -->
 248          *
 249          * @access private
 250          * @static
 251          */
 252         function extractTagsAndParams($tag, $text, &$content, &$tags, &$params, $uniq_prefix = ''){
 253                 $rnd = $uniq_prefix . '-' . $tag . Parser::getRandomString();
 254                 if ( !$content ) {
 255                         $content = array( );
 256                 }
 257                 $n = 1;
 258                 $stripped = '';
 259
 260                 if ( !$tags ) {
 261                         $tags = array( );
 262                 }
 263
 264                 if ( !$params ) {
 265                         $params = array( );
 266                 }
 267
 268                 if( $tag == STRIP_COMMENTS ) {
 269                         $start = '/<!--()/';
 270                         $end   = '/-->/';
 271                 } else {
 272                         $start = "/<$tag(\\s+[^>]*|\\s*)>/i";
 273                         $end   = "/<\\/$tag\\s*>/i";
 274                 }
 275
 276                 while ( '' != $text ) {
 277                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 278                         $stripped .= $p[0];
 279                         if( count( $p ) < 3 ) {
 280                                 break;
 281                         }
 282                         $attributes = $p[1];
 283                         $inside     = $p[2];
 284
 285                         $marker = $rnd . sprintf('%08X', $n++);
 286                         $stripped .= $marker;
 287
 288                         $tags[$marker] = "<$tag$attributes>";
 289                         $params[$marker] = Sanitizer::decodeTagAttributes( $attributes );
 290
 291                         $q = preg_split( $end, $inside, 2 );
 292                         $content[$marker] = $q[0];
 293                         if( count( $q ) < 1 ) {
 294                                 # No end tag -- let it run out to the end of the text.
 295                                 break;
 296                         } else {
 297                                 $text = $q[1];
 298                         }
 299                 }
 300                 return $stripped;
 301         }
 302
 303         /**
 304          * Wrapper function for extractTagsAndParams
 305          * for cases where $tags and $params isn't needed
 306          * i.e. where tags will never have params, like <nowiki>
 307          *
 308          * @access private
 309          * @static
 310          */
 311         function extractTags( $tag, $text, &$content, $uniq_prefix = '' ) {
 312                 $dummy_tags = array();
 313                 $dummy_params = array();
 314
 315                 return Parser::extractTagsAndParams( $tag, $text, $content,
 316                         $dummy_tags, $dummy_params, $uniq_prefix );
 317         }
 318
 319         /**
 320          * Strips and renders nowiki, pre, math, hiero
 321          * If $render is set, performs necessary rendering operations on plugins
 322          * Returns the text, and fills an array with data needed in unstrip()
 323          * If the $state is already a valid strip state, it adds to the state
 324          *
 325          * @param bool $stripcomments when set, HTML comments <!-- like this -->
 326          *  will be stripped in addition to other tags. This is important
 327          *  for section editing, where these comments cause confusion when
 328          *  counting the sections in the wikisource
 329          *
 330          * @access private
 331          */
 332         function strip( $text, &$state, $stripcomments = false ) {
 333                 $render = ($this->mOutputType == OT_HTML);
 334                 $html_content = array();
 335                 $nowiki_content = array();
 336                 $math_content = array();
 337                 $pre_content = array();
 338                 $comment_content = array();
 339                 $ext_content = array();
 340                 $ext_tags = array();
 341                 $ext_params = array();
 342                 $gallery_content = array();
 343
 344                 # Replace any instances of the placeholders
 345                 $uniq_prefix = UNIQ_PREFIX;
 346                 #$text = str_replace( $uniq_prefix, wfHtmlEscapeFirst( $uniq_prefix ), $text );
 347
 348                 # html
 349                 global $wgRawHtml;
 350                 if( $wgRawHtml ) {
 351                         $text = Parser::extractTags('html', $text, $html_content, $uniq_prefix);
 352                         foreach( $html_content as $marker => $content ) {
 353                                 if ($render ) {
 354                                         # Raw and unchecked for validity.
 355                                         $html_content[$marker] = $content;
 356                                 } else {
 357                                         $html_content[$marker] = '<html>'.$content.'</html>';
 358                                 }
 359                         }
 360                 }
 361
 362                 # nowiki
 363                 $text = Parser::extractTags('nowiki', $text, $nowiki_content, $uniq_prefix);
 364                 foreach( $nowiki_content as $marker => $content ) {
 365                         if( $render ){
 366                                 $nowiki_content[$marker] = wfEscapeHTMLTagsOnly( $content );
 367                         } else {
 368                                 $nowiki_content[$marker] = '<nowiki>'.$content.'</nowiki>';
 369                         }
 370                 }
 371
 372                 # math
 373                 $text = Parser::extractTags('math', $text, $math_content, $uniq_prefix);
 374                 foreach( $math_content as $marker => $content ){
 375                         if( $render ) {
 376                                 if( $this->mOptions->getUseTeX() ) {
 377                                         $math_content[$marker] = renderMath( $content );
 378                                 } else {
 379                                         $math_content[$marker] = '&lt;math&gt;'.$content.'&lt;math&gt;';
 380                                 }
 381                         } else {
 382                                 $math_content[$marker] = '<math>'.$content.'</math>';
 383                         }
 384                 }
 385
 386                 # pre
 387                 $text = Parser::extractTags('pre', $text, $pre_content, $uniq_prefix);
 388                 foreach( $pre_content as $marker => $content ){
 389                         if( $render ){
 390                                 $pre_content[$marker] = '<pre>' . wfEscapeHTMLTagsOnly( $content ) . '</pre>';
 391                         } else {
 392                                 $pre_content[$marker] = '<pre>'.$content.'</pre>';
 393                         }
 394                 }
 395
 396                 # gallery
 397                 $text = Parser::extractTags('gallery', $text, $gallery_content, $uniq_prefix);
 398                 foreach( $gallery_content as $marker => $content ) {
 399                         require_once( 'ImageGallery.php' ); // FIXME require in a foreach ?
 400                         if ( $render ) {
 401                                 $gallery_content[$marker] = Parser::renderImageGallery( $content );
 402                         } else {
 403                                 $gallery_content[$marker] = '<gallery>'.$content.'</gallery>';
 404                         }
 405                 }
 406
 407                 # Comments
 408                 if($stripcomments) {
 409                         $text = Parser::extractTags(STRIP_COMMENTS, $text, $comment_content, $uniq_prefix);
 410                         foreach( $comment_content as $marker => $content ){
 411                                 $comment_content[$marker] = '<!--'.$content.'-->';
 412                         }
 413                 }
 414
 415                 # Extensions
 416                 foreach ( $this->mTagHooks as $tag => $callback ) {
 417                         $ext_content[$tag] = array();
 418                         $text = Parser::extractTagsAndParams( $tag, $text, $ext_content[$tag],
 419                                 $ext_tags[$tag], $ext_params[$tag], $uniq_prefix );
 420                         foreach( $ext_content[$tag] as $marker => $content ) {
 421                                 $full_tag = $ext_tags[$tag][$marker];
 422                                 $params = $ext_params[$tag][$marker];
 423                                 if ( $render ) {
 424                                         $ext_content[$tag][$marker] = $callback( $content, $params );
 425                                 } else {
 426                                         $ext_content[$tag][$marker] = "$full_tag$content</$tag>";
 427                                 }
 428                         }
 429                 }
 430
 431                 # Merge state with the pre-existing state, if there is one
 432                 if ( $state ) {
 433                         $state['html'] = $state['html'] + $html_content;
 434                         $state['nowiki'] = $state['nowiki'] + $nowiki_content;
 435                         $state['math'] = $state['math'] + $math_content;
 436                         $state['pre'] = $state['pre'] + $pre_content;
 437                         $state['comment'] = $state['comment'] + $comment_content;
 438                         $state['gallery'] = $state['gallery'] + $gallery_content;
 439
 440                         foreach( $ext_content as $tag => $array ) {
 441                                 if ( array_key_exists( $tag, $state ) ) {
 442                                         $state[$tag] = $state[$tag] + $array;
 443                                 }
 444                         }
 445                 } else {
 446                         $state = array(
 447                           'html' => $html_content,
 448                           'nowiki' => $nowiki_content,
 449                           'math' => $math_content,
 450                           'pre' => $pre_content,
 451                           'comment' => $comment_content,
 452                           'gallery' => $gallery_content,
 453                         ) + $ext_content;
 454                 }
 455                 return $text;
 456         }
 457
 458         /**
 459          * restores pre, math, and hiero removed by strip()
 460          *
 461          * always call unstripNoWiki() after this one
 462          * @access private
 463          */
 464         function unstrip( $text, &$state ) {
 465                 # Must expand in reverse order, otherwise nested tags will be corrupted
 466                 $contentDict = end( $state );
 467                 for ( $contentDict = end( $state ); $contentDict !== false; $contentDict = prev( $state ) ) {
 468                         if( key($state) != 'nowiki' && key($state) != 'html') {
 469                                 for ( $content = end( $contentDict ); $content !== false; $content = prev( $contentDict ) ) {
 470                                         $text = str_replace( key( $contentDict ), $content, $text );
 471                                 }
 472                         }
 473                 }
 474
 475                 return $text;
 476         }
 477
 478         /**
 479          * always call this after unstrip() to preserve the order
 480          *
 481          * @access private
 482          */
 483         function unstripNoWiki( $text, &$state ) {
 484                 # Must expand in reverse order, otherwise nested tags will be corrupted
 485                 for ( $content = end($state['nowiki']); $content !== false; $content = prev( $state['nowiki'] ) ) {
 486                         $text = str_replace( key( $state['nowiki'] ), $content, $text );
 487                 }
 488
 489                 global $wgRawHtml;
 490                 if ($wgRawHtml) {
 491                         for ( $content = end($state['html']); $content !== false; $content = prev( $state['html'] ) ) {
 492                                 $text = str_replace( key( $state['html'] ), $content, $text );
 493                         }
 494                 }
 495
 496                 return $text;
 497         }
 498
 499         /**
 500          * Add an item to the strip state
 501          * Returns the unique tag which must be inserted into the stripped text
 502          * The tag will be replaced with the original text in unstrip()
 503          *
 504          * @access private
 505          */
 506         function insertStripItem( $text, &$state ) {
 507                 $rnd = UNIQ_PREFIX . '-item' . Parser::getRandomString();
 508                 if ( !$state ) {
 509                         $state = array(
 510                           'html' => array(),
 511                           'nowiki' => array(),
 512                           'math' => array(),
 513                           'pre' => array()
 514                         );
 515                 }
 516                 $state['item'][$rnd] = $text;
 517                 return $rnd;
 518         }
 519
 520         /**
 521          * Interface with html tidy, used if $wgUseTidy = true.
 522          * If tidy isn't able to correct the markup, the original will be
 523          * returned in all its glory with a warning comment appended.
 524          *
 525          * Either the external tidy program or the in-process tidy extension
 526          * will be used depending on availability. Override the default
 527          * $wgTidyInternal setting to disable the internal if it's not working.
 528          *
 529          * @param string $text Hideous HTML input
 530          * @return string Corrected HTML output
 531          * @access public
 532          * @static
 533          */
 534         function tidy( $text ) {
 535                 global $wgTidyInternal;
 536                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 537 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 538 '<head><title>test</title></head><body>'.$text.'</body></html>';
 539                 if( $wgTidyInternal ) {
 540                         $correctedtext = Parser::internalTidy( $wrappedtext );
 541                 } else {
 542                         $correctedtext = Parser::externalTidy( $wrappedtext );
 543                 }
 544                 if( is_null( $correctedtext ) ) {
 545                         wfDebug( "Tidy error detected!\n" );
 546                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 547                 }
 548                 return $correctedtext;
 549         }
 550
 551         /**
 552          * Spawn an external HTML tidy process and get corrected markup back from it.
 553          *
 554          * @access private
 555          * @static
 556          */
 557         function externalTidy( $text ) {
 558                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 559                 $fname = 'Parser::externalTidy';
 560                 wfProfileIn( $fname );
 561
 562                 $cleansource = '';
 563                 $opts = ' -utf8';
 564
 565                 $descriptorspec = array(
 566                         0 => array('pipe', 'r'),
 567                         1 => array('pipe', 'w'),
 568                         2 => array('file', '/dev/null', 'a')
 569                 );
 570                 $pipes = array();
 571                 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 572                 if (is_resource($process)) {
 573                         fwrite($pipes[0], $text);
 574                         fclose($pipes[0]);
 575                         while (!feof($pipes[1])) {
 576                                 $cleansource .= fgets($pipes[1], 1024);
 577                         }
 578                         fclose($pipes[1]);
 579                         proc_close($process);
 580                 }
 581
 582                 wfProfileOut( $fname );
 583
 584                 if( $cleansource == '' && $text != '') {
 585                         // Some kind of error happened, so we couldn't get the corrected text.
 586                         // Just give up; we'll use the source text and append a warning.
 587                         return null;
 588                 } else {
 589                         return $cleansource;
 590                 }
 591         }
 592
 593         /**
 594          * Use the HTML tidy PECL extension to use the tidy library in-process,
 595          * saving the overhead of spawning a new process. Currently written to
 596          * the PHP 4.3.x version of the extension, may not work on PHP 5.
 597          *
 598          * 'pear install tidy' should be able to compile the extension module.
 599          *
 600          * @access private
 601          * @static
 602          */
 603         function internalTidy( $text ) {
 604                 global $wgTidyConf;
 605                 $fname = 'Parser::internalTidy';
 606                 wfProfileIn( $fname );
 607
 608                 tidy_load_config( $wgTidyConf );
 609                 tidy_set_encoding( 'utf8' );
 610                 tidy_parse_string( $text );
 611                 tidy_clean_repair();
 612                 if( tidy_get_status() == 2 ) {
 613                         // 2 is magic number for fatal error
 614                         // http://www.php.net/manual/en/function.tidy-get-status.php
 615                         $cleansource = null;
 616                 } else {
 617                         $cleansource = tidy_get_output();
 618                 }
 619                 wfProfileOut( $fname );
 620                 return $cleansource;
 621         }
 622
 623         /**
 624          * parse the wiki syntax used to render tables
 625          *
 626          * @access private
 627          */
 628         function doTableStuff ( $t ) {
 629                 $fname = 'Parser::doTableStuff';
 630                 wfProfileIn( $fname );
 631
 632                 $t = explode ( "\n" , $t ) ;
 633                 $td = array () ; # Is currently a td tag open?
 634                 $ltd = array () ; # Was it TD or TH?
 635                 $tr = array () ; # Is currently a tr tag open?
 636                 $ltr = array () ; # tr attributes
 637                 $indent_level = 0; # indent level of the table
 638                 foreach ( $t AS $k => $x )
 639                 {
 640                         $x = trim ( $x ) ;
 641                         $fc = substr ( $x , 0 , 1 ) ;
 642                         if ( preg_match( '/^(:*)\{\|(.*)$/', $x, $matches ) ) {
 643                                 $indent_level = strlen( $matches[1] );
 644                                 $t[$k] = str_repeat( '<dl><dd>', $indent_level ) .
 645                                         '<table' . Sanitizer::fixTagAttributes ( $matches[2], 'table' ) . '>' ;
 646                                 array_push ( $td , false ) ;
 647                                 array_push ( $ltd , '' ) ;
 648                                 array_push ( $tr , false ) ;
 649                                 array_push ( $ltr , '' ) ;
 650                         }
 651                         else if ( count ( $td ) == 0 ) { } # Don't do any of the following
 652                         else if ( '|}' == substr ( $x , 0 , 2 ) ) {
 653                                 $z = "</table>" . substr ( $x , 2);
 654                                 $l = array_pop ( $ltd ) ;
 655                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 656                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 657                                 array_pop ( $ltr ) ;
 658                                 $t[$k] = $z . str_repeat( '</dd></dl>', $indent_level );
 659                         }
 660                         else if ( '|-' == substr ( $x , 0 , 2 ) ) { # Allows for |---------------
 661                                 $x = substr ( $x , 1 ) ;
 662                                 while ( $x != '' && substr ( $x , 0 , 1 ) == '-' ) $x = substr ( $x , 1 ) ;
 663                                 $z = '' ;
 664                                 $l = array_pop ( $ltd ) ;
 665                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 666                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 667                                 array_pop ( $ltr ) ;
 668                                 $t[$k] = $z ;
 669                                 array_push ( $tr , false ) ;
 670                                 array_push ( $td , false ) ;
 671                                 array_push ( $ltd , '' ) ;
 672                                 array_push ( $ltr , Sanitizer::fixTagAttributes ( $x, 'tr' ) ) ;
 673                         }
 674                         else if ( '|' == $fc || '!' == $fc || '|+' == substr ( $x , 0 , 2 ) ) { # Caption
 675                                 # $x is a table row
 676                                 if ( '|+' == substr ( $x , 0 , 2 ) ) {
 677                                         $fc = '+' ;
 678                                         $x = substr ( $x , 1 ) ;
 679                                 }
 680                                 $after = substr ( $x , 1 ) ;
 681                                 if ( $fc == '!' ) $after = str_replace ( '!!' , '||' , $after ) ;
 682                                 $after = explode ( '||' , $after ) ;
 683                                 $t[$k] = '' ;
 684
 685                                 # Loop through each table cell
 686                                 foreach ( $after AS $theline )
 687                                 {
 688                                         $z = '' ;
 689                                         if ( $fc != '+' )
 690                                         {
 691                                                 $tra = array_pop ( $ltr ) ;
 692                                                 if ( !array_pop ( $tr ) ) $z = '<tr'.$tra.">\n" ;
 693                                                 array_push ( $tr , true ) ;
 694                                                 array_push ( $ltr , '' ) ;
 695                                         }
 696
 697                                         $l = array_pop ( $ltd ) ;
 698                                         if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 699                                         if ( $fc == '|' ) $l = 'td' ;
 700                                         else if ( $fc == '!' ) $l = 'th' ;
 701                                         else if ( $fc == '+' ) $l = 'caption' ;
 702                                         else $l = '' ;
 703                                         array_push ( $ltd , $l ) ;
 704
 705                                         # Cell parameters
 706                                         $y = explode ( '|' , $theline , 2 ) ;
 707                                         # Note that a '|' inside an invalid link should not
 708                                         # be mistaken as delimiting cell parameters
 709                                         if ( strpos( $y[0], '[[' ) !== false ) {
 710                                                 $y = array ($theline);
 711                                         }
 712                                         if ( count ( $y ) == 1 )
 713                                                 $y = "{$z}<{$l}>{$y[0]}" ;
 714                                         else $y = $y = "{$z}<{$l}".Sanitizer::fixTagAttributes($y[0], $l).">{$y[1]}" ;
 715                                         $t[$k] .= $y ;
 716                                         array_push ( $td , true ) ;
 717                                 }
 718                         }
 719                 }
 720
 721                 # Closing open td, tr && table
 722                 while ( count ( $td ) > 0 )
 723                 {
 724                         if ( array_pop ( $td ) ) $t[] = '</td>' ;
 725                         if ( array_pop ( $tr ) ) $t[] = '</tr>' ;
 726                         $t[] = '</table>' ;
 727                 }
 728
 729                 $t = implode ( "\n" , $t ) ;
 730                 wfProfileOut( $fname );
 731                 return $t ;
 732         }
 733
 734         /**
 735          * Helper function for parse() that transforms wiki markup into
 736          * HTML. Only called for $mOutputType == OT_HTML.
 737          *
 738          * @access private
 739          */
 740         function internalParse( $text ) {
 741                 global $wgContLang;
 742                 $args = array();
 743                 $isMain = true;
 744                 $fname = 'Parser::internalParse';
 745                 wfProfileIn( $fname );
 746
 747                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'replaceVariables' ) );
 748                 $text = $this->replaceVariables( $text, $args );
 749
 750                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 751
 752                 $text = $this->doHeadings( $text );
 753                 if($this->mOptions->getUseDynamicDates()) {
 754                         $df =& DateFormatter::getInstance();
 755                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 756                 }
 757                 $text = $this->doAllQuotes( $text );
 758                 $text = $this->replaceInternalLinks( $text );
 759                 $text = $this->replaceExternalLinks( $text );
 760
 761                 # replaceInternalLinks may sometimes leave behind
 762                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 763                 $text = str_replace("http-noparse://","http://",$text);
 764
 765                 $text = $this->doMagicLinks( $text );
 766                 $text = $this->doTableStuff( $text );
 767                 $text = $this->formatHeadings( $text, $isMain );
 768
 769                 wfProfileOut( $fname );
 770                 return $text;
 771         }
 772
 773         /**
 774          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 775          * magic external links.
 776          *
 777          * @access private
 778          */
 779         function &doMagicLinks( &$text ) {
 780                 $text = $this->magicISBN( $text );
 781                 $text = $this->magicRFC( $text, 'RFC ', 'rfcurl' );
 782                 $text = $this->magicRFC( $text, 'PMID ', 'pubmedurl' );
 783                 return $text;
 784         }
 785
 786         /**
 787          * Parse ^^ tokens and return html
 788          *
 789          * @access private
 790          */
 791         function doExponent( $text ) {
 792                 $fname = 'Parser::doExponent';
 793                 wfProfileIn( $fname );
 794                 $text = preg_replace('/\^\^(.*)\^\^/','<small><sup>\\1</sup></small>', $text);
 795                 wfProfileOut( $fname );
 796                 return $text;
 797         }
 798
 799         /**
 800          * Parse headers and return html
 801          *
 802          * @access private
 803          */
 804         function doHeadings( $text ) {
 805                 $fname = 'Parser::doHeadings';
 806                 wfProfileIn( $fname );
 807                 for ( $i = 6; $i >= 1; --$i ) {
 808                         $h = substr( '======', 0, $i );
 809                         $text = preg_replace( "/^{$h}(.+){$h}(\\s|$)/m",
 810                           "<h{$i}>\\1</h{$i}>\\2", $text );
 811                 }
 812                 wfProfileOut( $fname );
 813                 return $text;
 814         }
 815
 816         /**
 817          * Replace single quotes with HTML markup
 818          * @access private
 819          * @return string the altered text
 820          */
 821         function doAllQuotes( $text ) {
 822                 $fname = 'Parser::doAllQuotes';
 823                 wfProfileIn( $fname );
 824                 $outtext = '';
 825                 $lines = explode( "\n", $text );
 826                 foreach ( $lines as $line ) {
 827                         $outtext .= $this->doQuotes ( $line ) . "\n";
 828                 }
 829                 $outtext = substr($outtext, 0,-1);
 830                 wfProfileOut( $fname );
 831                 return $outtext;
 832         }
 833
 834         /**
 835          * Helper function for doAllQuotes()
 836          * @access private
 837          */
 838         function doQuotes( $text ) {
 839                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
 840                 if ( count( $arr ) == 1 )
 841                         return $text;
 842                 else
 843                 {
 844                         # First, do some preliminary work. This may shift some apostrophes from
 845                         # being mark-up to being text. It also counts the number of occurrences
 846                         # of bold and italics mark-ups.
 847                         $i = 0;
 848                         $numbold = 0;
 849                         $numitalics = 0;
 850                         foreach ( $arr as $r )
 851                         {
 852                                 if ( ( $i % 2 ) == 1 )
 853                                 {
 854                                         # If there are ever four apostrophes, assume the first is supposed to
 855                                         # be text, and the remaining three constitute mark-up for bold text.
 856                                         if ( strlen( $arr[$i] ) == 4 )
 857                                         {
 858                                                 $arr[$i-1] .= "'";
 859                                                 $arr[$i] = "'''";
 860                                         }
 861                                         # If there are more than 5 apostrophes in a row, assume they're all
 862                                         # text except for the last 5.
 863                                         else if ( strlen( $arr[$i] ) > 5 )
 864                                         {
 865                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
 866                                                 $arr[$i] = "'''''";
 867                                         }
 868                                         # Count the number of occurrences of bold and italics mark-ups.
 869                                         # We are not counting sequences of five apostrophes.
 870                                         if ( strlen( $arr[$i] ) == 2 ) $numitalics++;  else
 871                                         if ( strlen( $arr[$i] ) == 3 ) $numbold++;     else
 872                                         if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
 873                                 }
 874                                 $i++;
 875                         }
 876
 877                         # If there is an odd number of both bold and italics, it is likely
 878                         # that one of the bold ones was meant to be an apostrophe followed
 879                         # by italics. Which one we cannot know for certain, but it is more
 880                         # likely to be one that has a single-letter word before it.
 881                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
 882                         {
 883                                 $i = 0;
 884                                 $firstsingleletterword = -1;
 885                                 $firstmultiletterword = -1;
 886                                 $firstspace = -1;
 887                                 foreach ( $arr as $r )
 888                                 {
 889                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
 890                                         {
 891                                                 $x1 = substr ($arr[$i-1], -1);
 892                                                 $x2 = substr ($arr[$i-1], -2, 1);
 893                                                 if ($x1 == ' ') {
 894                                                         if ($firstspace == -1) $firstspace = $i;
 895                                                 } else if ($x2 == ' ') {
 896                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
 897                                                 } else {
 898                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
 899                                                 }
 900                                         }
 901                                         $i++;
 902                                 }
 903
 904                                 # If there is a single-letter word, use it!
 905                                 if ($firstsingleletterword > -1)
 906                                 {
 907                                         $arr [ $firstsingleletterword ] = "''";
 908                                         $arr [ $firstsingleletterword-1 ] .= "'";
 909                                 }
 910                                 # If not, but there's a multi-letter word, use that one.
 911                                 else if ($firstmultiletterword > -1)
 912                                 {
 913                                         $arr [ $firstmultiletterword ] = "''";
 914                                         $arr [ $firstmultiletterword-1 ] .= "'";
 915                                 }
 916                                 # ... otherwise use the first one that has neither.
 917                                 # (notice that it is possible for all three to be -1 if, for example,
 918                                 # there is only one pentuple-apostrophe in the line)
 919                                 else if ($firstspace > -1)
 920                                 {
 921                                         $arr [ $firstspace ] = "''";
 922                                         $arr [ $firstspace-1 ] .= "'";
 923                                 }
 924                         }
 925
 926                         # Now let's actually convert our apostrophic mush to HTML!
 927                         $output = '';
 928                         $buffer = '';
 929                         $state = '';
 930                         $i = 0;
 931                         foreach ($arr as $r)
 932                         {
 933                                 if (($i % 2) == 0)
 934                                 {
 935                                         if ($state == 'both')
 936                                                 $buffer .= $r;
 937                                         else
 938                                                 $output .= $r;
 939                                 }
 940                                 else
 941                                 {
 942                                         if (strlen ($r) == 2)
 943                                         {
 944                                                 if ($state == 'i')
 945                                                 { $output .= '</i>'; $state = ''; }
 946                                                 else if ($state == 'bi')
 947                                                 { $output .= '</i>'; $state = 'b'; }
 948                                                 else if ($state == 'ib')
 949                                                 { $output .= '</b></i><b>'; $state = 'b'; }
 950                                                 else if ($state == 'both')
 951                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
 952                                                 else # $state can be 'b' or ''
 953                                                 { $output .= '<i>'; $state .= 'i'; }
 954                                         }
 955                                         else if (strlen ($r) == 3)
 956                                         {
 957                                                 if ($state == 'b')
 958                                                 { $output .= '</b>'; $state = ''; }
 959                                                 else if ($state == 'bi')
 960                                                 { $output .= '</i></b><i>'; $state = 'i'; }
 961                                                 else if ($state == 'ib')
 962                                                 { $output .= '</b>'; $state = 'i'; }
 963                                                 else if ($state == 'both')
 964                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
 965                                                 else # $state can be 'i' or ''
 966                                                 { $output .= '<b>'; $state .= 'b'; }
 967                                         }
 968                                         else if (strlen ($r) == 5)
 969                                         {
 970                                                 if ($state == 'b')
 971                                                 { $output .= '</b><i>'; $state = 'i'; }
 972                                                 else if ($state == 'i')
 973                                                 { $output .= '</i><b>'; $state = 'b'; }
 974                                                 else if ($state == 'bi')
 975                                                 { $output .= '</i></b>'; $state = ''; }
 976                                                 else if ($state == 'ib')
 977                                                 { $output .= '</b></i>'; $state = ''; }
 978                                                 else if ($state == 'both')
 979                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
 980                                                 else # ($state == '')
 981                                                 { $buffer = ''; $state = 'both'; }
 982                                         }
 983                                 }
 984                                 $i++;
 985                         }
 986                         # Now close all remaining tags.  Notice that the order is important.
 987                         if ($state == 'b' || $state == 'ib')
 988                                 $output .= '</b>';
 989                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
 990                                 $output .= '</i>';
 991                         if ($state == 'bi')
 992                                 $output .= '</b>';
 993                         if ($state == 'both')
 994                                 $output .= '<b><i>'.$buffer.'</i></b>';
 995                         return $output;
 996                 }
 997         }
 998
 999         /**
1000          * Replace external links
1001          *
1002          * Note: this is all very hackish and the order of execution matters a lot.
1003          * Make sure to run maintenance/parserTests.php if you change this code.
1004          *
1005          * @access private
1006          */
1007         function replaceExternalLinks( $text ) {
1008                 global $wgContLang;
1009                 $fname = 'Parser::replaceExternalLinks';
1010                 wfProfileIn( $fname );
1011
1012                 $sk =& $this->mOptions->getSkin();
1013
1014                 $bits = preg_split( EXT_LINK_BRACKETED, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1015
1016                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
1017
1018                 $i = 0;
1019                 while ( $i<count( $bits ) ) {
1020                         $url = $bits[$i++];
1021                         $protocol = $bits[$i++];
1022                         $text = $bits[$i++];
1023                         $trail = $bits[$i++];
1024
1025                         # The characters '<' and '>' (which were escaped by
1026                         # removeHTMLtags()) should not be included in
1027                         # URLs, per RFC 2396.
1028                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1029                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1030                                 $url = substr($url, 0, $m2[0][1]);
1031                         }
1032
1033                         # If the link text is an image URL, replace it with an <img> tag
1034                         # This happened by accident in the original parser, but some people used it extensively
1035                         $img = $this->maybeMakeExternalImage( $text );
1036                         if ( $img !== false ) {
1037                                 $text = $img;
1038                         }
1039
1040                         $dtrail = '';
1041
1042                         # Set linktype for CSS - if URL==text, link is essentially free
1043                         $linktype = ($text == $url) ? 'free' : 'text';
1044
1045                         # No link text, e.g. [http://domain.tld/some.link]
1046                         if ( $text == '' ) {
1047                                 # Autonumber if allowed
1048                                 if ( strpos( HTTP_PROTOCOLS, $protocol ) !== false ) {
1049                                         $text = '[' . ++$this->mAutonumber . ']';
1050                                         $linktype = 'autonumber';
1051                                 } else {
1052                                         # Otherwise just use the URL
1053                                         $text = htmlspecialchars( $url );
1054                                         $linktype = 'free';
1055                                 }
1056                         } else {
1057                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1058                                 # Check for trail
1059                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1060                         }
1061
1062                         $text = $wgContLang->markNoConversion($text);
1063
1064                         # Replace &amp; from obsolete syntax with &.
1065                         # All HTML entities will be escaped by makeExternalLink()
1066                         # or maybeMakeExternalImage()
1067                         $url = str_replace( '&amp;', '&', $url );
1068
1069                         # Process the trail (i.e. everything after this link up until start of the next link),
1070                         # replacing any non-bracketed links
1071                         $trail = $this->replaceFreeExternalLinks( $trail );
1072
1073
1074                         # Use the encoded URL
1075                         # This means that users can paste URLs directly into the text
1076                         # Funny characters like &ouml; aren't valid in URLs anyway
1077                         # This was changed in August 2004
1078                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype ) . $dtrail . $trail;
1079                 }
1080
1081                 wfProfileOut( $fname );
1082                 return $s;
1083         }
1084
1085         /**
1086          * Replace anything that looks like a URL with a link
1087          * @access private
1088          */
1089         function replaceFreeExternalLinks( $text ) {
1090                 global $wgContLang;
1091                 $fname = 'Parser::replaceFreeExternalLinks';
1092                 wfProfileIn( $fname );
1093
1094                 $bits = preg_split( '/(\b(?:'.URL_PROTOCOLS.'):)/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1095                 $s = array_shift( $bits );
1096                 $i = 0;
1097
1098                 $sk =& $this->mOptions->getSkin();
1099
1100                 while ( $i < count( $bits ) ){
1101                         $protocol = $bits[$i++];
1102                         $remainder = $bits[$i++];
1103
1104                         if ( preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1105                                 # Found some characters after the protocol that look promising
1106                                 $url = $protocol . $m[1];
1107                                 $trail = $m[2];
1108
1109                                 # The characters '<' and '>' (which were escaped by
1110                                 # removeHTMLtags()) should not be included in
1111                                 # URLs, per RFC 2396.
1112                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1113                                         $trail = substr($url, $m2[0][1]) . $trail;
1114                                         $url = substr($url, 0, $m2[0][1]);
1115                                 }
1116
1117                                 # Move trailing punctuation to $trail
1118                                 $sep = ',;\.:!?';
1119                                 # If there is no left bracket, then consider right brackets fair game too
1120                                 if ( strpos( $url, '(' ) === false ) {
1121                                         $sep .= ')';
1122                                 }
1123
1124                                 $numSepChars = strspn( strrev( $url ), $sep );
1125                                 if ( $numSepChars ) {
1126                                         $trail = substr( $url, -$numSepChars ) . $trail;
1127                                         $url = substr( $url, 0, -$numSepChars );
1128                                 }
1129
1130                                 # Replace &amp; from obsolete syntax with &.
1131                                 # All HTML entities will be escaped by makeExternalLink()
1132                                 # or maybeMakeExternalImage()
1133                                 $url = str_replace( '&amp;', '&', $url );
1134
1135                                 # Is this an external image?
1136                                 $text = $this->maybeMakeExternalImage( $url );
1137                                 if ( $text === false ) {
1138                                         # Not an image, make a link
1139                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free' );
1140                                 }
1141                                 $s .= $text . $trail;
1142                         } else {
1143                                 $s .= $protocol . $remainder;
1144                         }
1145                 }
1146                 wfProfileOut();
1147                 return $s;
1148         }
1149
1150         /**
1151          * make an image if it's allowed
1152          * @access private
1153          */
1154         function maybeMakeExternalImage( $url ) {
1155                 $sk =& $this->mOptions->getSkin();
1156                 $text = false;
1157                 if ( $this->mOptions->getAllowExternalImages() ) {
1158                         if ( preg_match( EXT_IMAGE_REGEX, $url ) ) {
1159                                 # Image found
1160                                 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
1161                         }
1162                 }
1163                 return $text;
1164         }
1165
1166         /**
1167          * Process [[ ]] wikilinks
1168          *
1169          * @access private
1170          */
1171         function replaceInternalLinks( $s ) {
1172                 global $wgContLang, $wgLinkCache;
1173                 static $fname = 'Parser::replaceInternalLinks' ;
1174
1175                 wfProfileIn( $fname );
1176
1177                 wfProfileIn( $fname.'-setup' );
1178                 static $tc = FALSE;
1179                 # the % is needed to support urlencoded titles as well
1180                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1181
1182                 $sk =& $this->mOptions->getSkin();
1183
1184                 #split the entire text string on occurences of [[
1185                 $a = explode( '[[', ' ' . $s );
1186                 #get the first element (all text up to first [[), and remove the space we added
1187                 $s = array_shift( $a );
1188                 $s = substr( $s, 1 );
1189
1190                 # Match a link having the form [[namespace:link|alternate]]trail
1191                 static $e1 = FALSE;
1192                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1193                 # Match cases where there is no "]]", which might still be images
1194                 static $e1_img = FALSE;
1195                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1196                 # Match the end of a line for a word that's not followed by whitespace,
1197                 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1198                 static $e2 = '/^(.*?)([a-zA-Z\x80-\xff]+)$/sD';
1199
1200                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1201
1202                 if( is_null( $this->mTitle ) ) {
1203                         wfDebugDieBacktrace( 'nooo' );
1204                 }
1205                 $nottalk = !$this->mTitle->isTalkPage();
1206
1207                 if ( $useLinkPrefixExtension ) {
1208                         if ( preg_match( $e2, $s, $m ) ) {
1209                                 $first_prefix = $m[2];
1210                                 $s = $m[1];
1211                         } else {
1212                                 $first_prefix = false;
1213                         }
1214                 } else {
1215                         $prefix = '';
1216                 }
1217
1218                 $selflink = $this->mTitle->getPrefixedText();
1219                 wfProfileOut( $fname.'-setup' );
1220
1221                 $checkVariantLink = sizeof($wgContLang->getVariants())>1;
1222                 $useSubpages = $this->areSubpagesAllowed();
1223
1224                 # Loop for each link
1225                 for ($k = 0; isset( $a[$k] ); $k++) {
1226                         $line = $a[$k];
1227                         if ( $useLinkPrefixExtension ) {
1228                                 wfProfileIn( $fname.'-prefixhandling' );
1229                                 if ( preg_match( $e2, $s, $m ) ) {
1230                                         $prefix = $m[2];
1231                                         $s = $m[1];
1232                                 } else {
1233                                         $prefix='';
1234                                 }
1235                                 # first link
1236                                 if($first_prefix) {
1237                                         $prefix = $first_prefix;
1238                                         $first_prefix = false;
1239                                 }
1240                                 wfProfileOut( $fname.'-prefixhandling' );
1241                         }
1242
1243                         $might_be_img = false;
1244
1245                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1246                                 $text = $m[2];
1247                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1248                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1249                                 # the real problem is with the $e1 regex
1250                                 # See bug 1300.
1251                                 #
1252                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1253                                 # and no image is in sight. See bug 2095.
1254                                 #
1255                                 if( $text !== '' && preg_match( "/^\](.*)/s", $m[3], $n ) ) {
1256                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1257                                         $m[3] = $n[1];
1258                                 }
1259                                 # fix up urlencoded title texts
1260                                 if(preg_match('/%/', $m[1] )) $m[1] = urldecode($m[1]);
1261                                 $trail = $m[3];
1262                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1263                                 $might_be_img = true;
1264                                 $text = $m[2];
1265                                 if(preg_match('/%/', $m[1] )) $m[1] = urldecode($m[1]);
1266                                 $trail = "";
1267                         } else { # Invalid form; output directly
1268                                 $s .= $prefix . '[[' . $line ;
1269                                 continue;
1270                         }
1271
1272                         # Don't allow internal links to pages containing
1273                         # PROTO: where PROTO is a valid URL protocol; these
1274                         # should be external links.
1275                         if (preg_match('/^(\b(?:'.URL_PROTOCOLS.'):)/', $m[1])) {
1276                                 $s .= $prefix . '[[' . $line ;
1277                                 continue;
1278                         }
1279
1280                         # Make subpage if necessary
1281                         if( $useSubpages ) {
1282                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1283                         } else {
1284                                 $link = $m[1];
1285                         }
1286
1287                         $noforce = (substr($m[1], 0, 1) != ':');
1288                         if (!$noforce) {
1289                                 # Strip off leading ':'
1290                                 $link = substr($link, 1);
1291                         }
1292
1293                         $nt =& Title::newFromText( $this->unstripNoWiki($link, $this->mStripState) );
1294                         if( !$nt ) {
1295                                 $s .= $prefix . '[[' . $line;
1296                                 continue;
1297                         }
1298
1299                         #check other language variants of the link
1300                         #if the article does not exist
1301                         if( $checkVariantLink
1302                             && $nt->getArticleID() == 0 ) {
1303                                 $wgContLang->findVariantLink($link, $nt);
1304                         }
1305
1306                         $ns = $nt->getNamespace();
1307                         $iw = $nt->getInterWiki();
1308
1309                         if ($might_be_img) { # if this is actually an invalid link
1310                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1311                                         $found = false;
1312                                         while (isset ($a[$k+1]) ) {
1313                                                 #look at the next 'line' to see if we can close it there
1314                                                 $next_line =  array_shift(array_splice( $a, $k + 1, 1) );
1315                                                 if( preg_match("/^(.*?]].*?)]](.*)$/sD", $next_line, $m) ) {
1316                                                 # the first ]] closes the inner link, the second the image
1317                                                         $found = true;
1318                                                         $text .= '[[' . $m[1];
1319                                                         $trail = $m[2];
1320                                                         break;
1321                                                 } elseif( preg_match("/^.*?]].*$/sD", $next_line, $m) ) {
1322                                                         #if there's exactly one ]] that's fine, we'll keep looking
1323                                                         $text .= '[[' . $m[0];
1324                                                 } else {
1325                                                         #if $next_line is invalid too, we need look no further
1326                                                         $text .= '[[' . $next_line;
1327                                                         break;
1328                                                 }
1329                                         }
1330                                         if ( !$found ) {
1331                                                 # we couldn't find the end of this imageLink, so output it raw
1332                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1333                                                 $text = $this->replaceInternalLinks($text);
1334                                                 $s .= $prefix . '[[' . $link . '|' . $text;
1335                                                 # note: no $trail, because without an end, there *is* no trail
1336                                                 continue;
1337                                         }
1338                                 } else { #it's not an image, so output it raw
1339                                         $s .= $prefix . '[[' . $link . '|' . $text;
1340                                         # note: no $trail, because without an end, there *is* no trail
1341                                         continue;
1342                                 }
1343                         }
1344
1345                         $wasblank = ( '' == $text );
1346                         if( $wasblank ) $text = $link;
1347
1348
1349                         # Link not escaped by : , create the various objects
1350                         if( $noforce ) {
1351
1352                                 # Interwikis
1353                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1354                                         array_push( $this->mOutput->mLanguageLinks, $nt->getFullText() );
1355                                         $s = rtrim($s . "\n");
1356                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1357                                         continue;
1358                                 }
1359
1360                                 if ( $ns == NS_IMAGE ) {
1361                                         wfProfileIn( "$fname-image" );
1362                                         if ( !wfIsBadImage( $nt->getDBkey() ) ) {
1363                                                 # recursively parse links inside the image caption
1364                                                 # actually, this will parse them in any other parameters, too,
1365                                                 # but it might be hard to fix that, and it doesn't matter ATM
1366                                                 $text = $this->replaceExternalLinks($text);
1367                                                 $text = $this->replaceInternalLinks($text);
1368
1369                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1370                                                 $s .= $prefix . str_replace('http://', 'http-noparse://', $this->makeImage( $nt, $text ) ) . $trail;
1371                                                 $wgLinkCache->addImageLinkObj( $nt );
1372
1373                                                 wfProfileOut( "$fname-image" );
1374                                                 continue;
1375                                         }
1376                                         wfProfileOut( "$fname-image" );
1377
1378                                 }
1379
1380                                 if ( $ns == NS_CATEGORY ) {
1381                                         wfProfileIn( "$fname-category" );
1382                                         $t = $wgContLang->convert($nt->getText());
1383                                         $s = rtrim($s . "\n"); # bug 87
1384
1385                                         $wgLinkCache->suspend(); # Don't save in links/brokenlinks
1386                                         $t = $sk->makeLinkObj( $nt, $t, '', '' , $prefix );
1387                                         $wgLinkCache->resume();
1388
1389                                         if ( $wasblank ) {
1390                                                 if ( $this->mTitle->getNamespace() == NS_CATEGORY ) {
1391                                                         $sortkey = $this->mTitle->getText();
1392                                                 } else {
1393                                                         $sortkey = $this->mTitle->getPrefixedText();
1394                                                 }
1395                                         } else {
1396                                                 $sortkey = $text;
1397                                         }
1398                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1399                                         $wgLinkCache->addCategoryLinkObj( $nt, $sortkey );
1400                                         $this->mOutput->addCategoryLink( $t );
1401
1402                                         /**
1403                                          * Strip the whitespace Category links produce, see bug 87
1404                                          * @todo We might want to use trim($tmp, "\n") here.
1405                                          */
1406                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1407
1408                                         wfProfileOut( "$fname-category" );
1409                                         continue;
1410                                 }
1411                         }
1412
1413                         if( ( $nt->getPrefixedText() === $selflink ) &&
1414                             ( $nt->getFragment() === '' ) ) {
1415                                 # Self-links are handled specially; generally de-link and change to bold.
1416                                 $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1417                                 continue;
1418                         }
1419
1420                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1421                         if( $ns == NS_MEDIA ) {
1422                                 $s .= $prefix . $sk->makeMediaLinkObj( $nt, $text, true ) . $trail;
1423                                 $wgLinkCache->addImageLinkObj( $nt );
1424                                 continue;
1425                         } elseif( $ns == NS_SPECIAL ) {
1426                                 $s .= $prefix . $sk->makeKnownLinkObj( $nt, $text, '', $trail );
1427                                 continue;
1428                         }
1429                         if( !$nt->isExternal() && $nt->isAlwaysKnown() ) {
1430                                 /**
1431                                  * Skip lookups for special pages and self-links.
1432                                  * External interwiki links are not included here because
1433                                  * the HTTP urls would break output in the next parse step;
1434                                  * they will have placeholders kept.
1435                                  */
1436                                 $s .= $sk->makeKnownLinkObj( $nt, $text, '', $trail, $prefix );
1437                         } else {
1438                                 /**
1439                                  * Add a link placeholder
1440                                  * Later, this will be replaced by a real link, after the existence or
1441                                  * non-existence of all the links is known
1442                                  */
1443                                 $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1444                         }
1445                 }
1446                 wfProfileOut( $fname );
1447                 return $s;
1448         }
1449
1450         /**
1451          * Make a link placeholder. The text returned can be later resolved to a real link with
1452          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1453          * parsing of interwiki links, and secondly to allow all extistence checks and
1454          * article length checks (for stub links) to be bundled into a single query.
1455          *
1456          */
1457         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1458                 if ( ! is_object($nt) ) {
1459                         # Fail gracefully
1460                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1461                 } else {
1462                         # Separate the link trail from the rest of the link
1463                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1464
1465                         if ( $nt->isExternal() ) {
1466                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1467                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1468                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1469                         } else {
1470                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1471                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1472                                 $this->mLinkHolders['queries'][] = $query;
1473                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1474                                 $this->mLinkHolders['titles'][] = $nt;
1475
1476                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1477                         }
1478                 }
1479                 return $retVal;
1480         }
1481
1482         /**
1483          * Return true if subpage links should be expanded on this page.
1484          * @return bool
1485          */
1486         function areSubpagesAllowed() {
1487                 # Some namespaces don't allow subpages
1488                 global $wgNamespacesWithSubpages;
1489                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1490         }
1491
1492         /**
1493          * Handle link to subpage if necessary
1494          * @param string $target the source of the link
1495          * @param string &$text the link text, modified as necessary
1496          * @return string the full name of the link
1497          * @access private
1498          */
1499         function maybeDoSubpageLink($target, &$text) {
1500                 # Valid link forms:
1501                 # Foobar -- normal
1502                 # :Foobar -- override special treatment of prefix (images, language links)
1503                 # /Foobar -- convert to CurrentPage/Foobar
1504                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1505                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1506                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1507
1508                 $fname = 'Parser::maybeDoSubpageLink';
1509                 wfProfileIn( $fname );
1510                 $ret = $target; # default return value is no change
1511
1512                 # Some namespaces don't allow subpages,
1513                 # so only perform processing if subpages are allowed
1514                 if( $this->areSubpagesAllowed() ) {
1515                         # Look at the first character
1516                         if( $target != '' && $target{0} == '/' ) {
1517                                 # / at end means we don't want the slash to be shown
1518                                 if( substr( $target, -1, 1 ) == '/' ) {
1519                                         $target = substr( $target, 1, -1 );
1520                                         $noslash = $target;
1521                                 } else {
1522                                         $noslash = substr( $target, 1 );
1523                                 }
1524
1525                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash);
1526                                 if( '' === $text ) {
1527                                         $text = $target;
1528                                 } # this might be changed for ugliness reasons
1529                         } else {
1530                                 # check for .. subpage backlinks
1531                                 $dotdotcount = 0;
1532                                 $nodotdot = $target;
1533                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1534                                         ++$dotdotcount;
1535                                         $nodotdot = substr( $nodotdot, 3 );
1536                                 }
1537                                 if($dotdotcount > 0) {
1538                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1539                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1540                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1541                                                 # / at the end means don't show full path
1542                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1543                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1544                                                         if( '' === $text ) {
1545                                                                 $text = $nodotdot;
1546                                                         }
1547                                                 }
1548                                                 $nodotdot = trim( $nodotdot );
1549                                                 if( $nodotdot != '' ) {
1550                                                         $ret .= '/' . $nodotdot;
1551                                                 }
1552                                         }
1553                                 }
1554                         }
1555                 }
1556
1557                 wfProfileOut( $fname );
1558                 return $ret;
1559         }
1560
1561         /**#@+
1562          * Used by doBlockLevels()
1563          * @access private
1564          */
1565         /* private */ function closeParagraph() {
1566                 $result = '';
1567                 if ( '' != $this->mLastSection ) {
1568                         $result = '</' . $this->mLastSection  . ">\n";
1569                 }
1570                 $this->mInPre = false;
1571                 $this->mLastSection = '';
1572                 return $result;
1573         }
1574         # getCommon() returns the length of the longest common substring
1575         # of both arguments, starting at the beginning of both.
1576         #
1577         /* private */ function getCommon( $st1, $st2 ) {
1578                 $fl = strlen( $st1 );
1579                 $shorter = strlen( $st2 );
1580                 if ( $fl < $shorter ) { $shorter = $fl; }
1581
1582                 for ( $i = 0; $i < $shorter; ++$i ) {
1583                         if ( $st1{$i} != $st2{$i} ) { break; }
1584                 }
1585                 return $i;
1586         }
1587         # These next three functions open, continue, and close the list
1588         # element appropriate to the prefix character passed into them.
1589         #
1590         /* private */ function openList( $char ) {
1591                 $result = $this->closeParagraph();
1592
1593                 if ( '*' == $char ) { $result .= '<ul><li>'; }
1594                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
1595                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
1596                 else if ( ';' == $char ) {
1597                         $result .= '<dl><dt>';
1598                         $this->mDTopen = true;
1599                 }
1600                 else { $result = '<!-- ERR 1 -->'; }
1601
1602                 return $result;
1603         }
1604
1605         /* private */ function nextItem( $char ) {
1606                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
1607                 else if ( ':' == $char || ';' == $char ) {
1608                         $close = '</dd>';
1609                         if ( $this->mDTopen ) { $close = '</dt>'; }
1610                         if ( ';' == $char ) {
1611                                 $this->mDTopen = true;
1612                                 return $close . '<dt>';
1613                         } else {
1614                                 $this->mDTopen = false;
1615                                 return $close . '<dd>';
1616                         }
1617                 }
1618                 return '<!-- ERR 2 -->';
1619         }
1620
1621         /* private */ function closeList( $char ) {
1622                 if ( '*' == $char ) { $text = '</li></ul>'; }
1623                 else if ( '#' == $char ) { $text = '</li></ol>'; }
1624                 else if ( ':' == $char ) {
1625                         if ( $this->mDTopen ) {
1626                                 $this->mDTopen = false;
1627                                 $text = '</dt></dl>';
1628                         } else {
1629                                 $text = '</dd></dl>';
1630                         }
1631                 }
1632                 else {  return '<!-- ERR 3 -->'; }
1633                 return $text."\n";
1634         }
1635         /**#@-*/
1636
1637         /**
1638          * Make lists from lines starting with ':', '*', '#', etc.
1639          *
1640          * @access private
1641          * @return string the lists rendered as HTML
1642          */
1643         function doBlockLevels( $text, $linestart ) {
1644                 $fname = 'Parser::doBlockLevels';
1645                 wfProfileIn( $fname );
1646
1647                 # Parsing through the text line by line.  The main thing
1648                 # happening here is handling of block-level elements p, pre,
1649                 # and making lists from lines starting with * # : etc.
1650                 #
1651                 $textLines = explode( "\n", $text );
1652
1653                 $lastPrefix = $output = '';
1654                 $this->mDTopen = $inBlockElem = false;
1655                 $prefixLength = 0;
1656                 $paragraphStack = false;
1657
1658                 if ( !$linestart ) {
1659                         $output .= array_shift( $textLines );
1660                 }
1661                 foreach ( $textLines as $oLine ) {
1662                         $lastPrefixLength = strlen( $lastPrefix );
1663                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
1664                         $preOpenMatch = preg_match('/<pre/i', $oLine );
1665                         if ( !$this->mInPre ) {
1666                                 # Multiple prefixes may abut each other for nested lists.
1667                                 $prefixLength = strspn( $oLine, '*#:;' );
1668                                 $pref = substr( $oLine, 0, $prefixLength );
1669
1670                                 # eh?
1671                                 $pref2 = str_replace( ';', ':', $pref );
1672                                 $t = substr( $oLine, $prefixLength );
1673                                 $this->mInPre = !empty($preOpenMatch);
1674                         } else {
1675                                 # Don't interpret any other prefixes in preformatted text
1676                                 $prefixLength = 0;
1677                                 $pref = $pref2 = '';
1678                                 $t = $oLine;
1679                         }
1680
1681                         # List generation
1682                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
1683                                 # Same as the last item, so no need to deal with nesting or opening stuff
1684                                 $output .= $this->nextItem( substr( $pref, -1 ) );
1685                                 $paragraphStack = false;
1686
1687                                 if ( substr( $pref, -1 ) == ';') {
1688                                         # The one nasty exception: definition lists work like this:
1689                                         # ; title : definition text
1690                                         # So we check for : in the remainder text to split up the
1691                                         # title and definition, without b0rking links.
1692                                         $term = $t2 = '';
1693                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
1694                                                 $t = $t2;
1695                                                 $output .= $term . $this->nextItem( ':' );
1696                                         }
1697                                 }
1698                         } elseif( $prefixLength || $lastPrefixLength ) {
1699                                 # Either open or close a level...
1700                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
1701                                 $paragraphStack = false;
1702
1703                                 while( $commonPrefixLength < $lastPrefixLength ) {
1704                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
1705                                         --$lastPrefixLength;
1706                                 }
1707                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
1708                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
1709                                 }
1710                                 while ( $prefixLength > $commonPrefixLength ) {
1711                                         $char = substr( $pref, $commonPrefixLength, 1 );
1712                                         $output .= $this->openList( $char );
1713
1714                                         if ( ';' == $char ) {
1715                                                 # FIXME: This is dupe of code above
1716                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
1717                                                         $t = $t2;
1718                                                         $output .= $term . $this->nextItem( ':' );
1719                                                 }
1720                                         }
1721                                         ++$commonPrefixLength;
1722                                 }
1723                                 $lastPrefix = $pref2;
1724                         }
1725                         if( 0 == $prefixLength ) {
1726                                 wfProfileIn( "$fname-paragraph" );
1727                                 # No prefix (not in list)--go to paragraph mode
1728                                 $uniq_prefix = UNIQ_PREFIX;
1729                                 // XXX: use a stack for nestable elements like span, table and div
1730                                 $openmatch = preg_match('/(<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
1731                                 $closematch = preg_match(
1732                                         '/(<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
1733                                         '<td|<th|<div|<\\/div|<hr|<\\/pre|<\\/p|'.$uniq_prefix.'-pre|<\\/li|<\\/ul)/iS', $t );
1734                                 if ( $openmatch or $closematch ) {
1735                                         $paragraphStack = false;
1736                                         $output .= $this->closeParagraph();
1737                                         if($preOpenMatch and !$preCloseMatch) {
1738                                                 $this->mInPre = true;
1739                                         }
1740                                         if ( $closematch ) {
1741                                                 $inBlockElem = false;
1742                                         } else {
1743                                                 $inBlockElem = true;
1744                                         }
1745                                 } else if ( !$inBlockElem && !$this->mInPre ) {
1746                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
1747                                                 // pre
1748                                                 if ($this->mLastSection != 'pre') {
1749                                                         $paragraphStack = false;
1750                                                         $output .= $this->closeParagraph().'<pre>';
1751                                                         $this->mLastSection = 'pre';
1752                                                 }
1753                                                 $t = substr( $t, 1 );
1754                                         } else {
1755                                                 // paragraph
1756                                                 if ( '' == trim($t) ) {
1757                                                         if ( $paragraphStack ) {
1758                                                                 $output .= $paragraphStack.'<br />';
1759                                                                 $paragraphStack = false;
1760                                                                 $this->mLastSection = 'p';
1761                                                         } else {
1762                                                                 if ($this->mLastSection != 'p' ) {
1763                                                                         $output .= $this->closeParagraph();
1764                                                                         $this->mLastSection = '';
1765                                                                         $paragraphStack = '<p>';
1766                                                                 } else {
1767                                                                         $paragraphStack = '</p><p>';
1768                                                                 }
1769                                                         }
1770                                                 } else {
1771                                                         if ( $paragraphStack ) {
1772                                                                 $output .= $paragraphStack;
1773                                                                 $paragraphStack = false;
1774                                                                 $this->mLastSection = 'p';
1775                                                         } else if ($this->mLastSection != 'p') {
1776                                                                 $output .= $this->closeParagraph().'<p>';
1777                                                                 $this->mLastSection = 'p';
1778                                                         }
1779                                                 }
1780                                         }
1781                                 }
1782                                 wfProfileOut( "$fname-paragraph" );
1783                         }
1784                         if ($paragraphStack === false) {
1785                                 $output .= $t."\n";
1786                         }
1787                 }
1788                 while ( $prefixLength ) {
1789                         $output .= $this->closeList( $pref2{$prefixLength-1} );
1790                         --$prefixLength;
1791                 }
1792                 if ( '' != $this->mLastSection ) {
1793                         $output .= '</' . $this->mLastSection . '>';
1794                         $this->mLastSection = '';
1795                 }
1796
1797                 wfProfileOut( $fname );
1798                 return $output;
1799         }
1800
1801         /**
1802          * Split up a string on ':', ignoring any occurences inside
1803          * <a>..</a> or <span>...</span>
1804          * @param string $str the string to split
1805          * @param string &$before set to everything before the ':'
1806          * @param string &$after set to everything after the ':'
1807          * return string the position of the ':', or false if none found
1808          */
1809         function findColonNoLinks($str, &$before, &$after) {
1810                 # I wonder if we should make this count all tags, not just <a>
1811                 # and <span>. That would prevent us from matching a ':' that
1812                 # comes in the middle of italics other such formatting....
1813                 # -- Wil
1814                 $fname = 'Parser::findColonNoLinks';
1815                 wfProfileIn( $fname );
1816                 $pos = 0;
1817                 do {
1818                         $colon = strpos($str, ':', $pos);
1819
1820                         if ($colon !== false) {
1821                                 $before = substr($str, 0, $colon);
1822                                 $after = substr($str, $colon + 1);
1823
1824                                 # Skip any ':' within <a> or <span> pairs
1825                                 $a = substr_count($before, '<a');
1826                                 $s = substr_count($before, '<span');
1827                                 $ca = substr_count($before, '</a>');
1828                                 $cs = substr_count($before, '</span>');
1829
1830                                 if ($a <= $ca and $s <= $cs) {
1831                                         # Tags are balanced before ':'; ok
1832                                         break;
1833                                 }
1834                                 $pos = $colon + 1;
1835                         }
1836                 } while ($colon !== false);
1837                 wfProfileOut( $fname );
1838                 return $colon;
1839         }
1840
1841         /**
1842          * Return value of a magic variable (like PAGENAME)
1843          *
1844          * @access private
1845          */
1846         function getVariableValue( $index ) {
1847                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgArticle, $wgScriptPath;
1848
1849                 /**
1850                  * Some of these require message or data lookups and can be
1851                  * expensive to check many times.
1852                  */
1853                 static $varCache = array();
1854                 if( isset( $varCache[$index] ) ) return $varCache[$index];
1855
1856                 switch ( $index ) {
1857                         case MAG_CURRENTMONTH:
1858                                 return $varCache[$index] = $wgContLang->formatNum( date( 'm' ) );
1859                         case MAG_CURRENTMONTHNAME:
1860                                 return $varCache[$index] = $wgContLang->getMonthName( date('n') );
1861                         case MAG_CURRENTMONTHNAMEGEN:
1862                                 return $varCache[$index] = $wgContLang->getMonthNameGen( date('n') );
1863                         case MAG_CURRENTMONTHABBREV:
1864                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( date('n') );
1865                         case MAG_CURRENTDAY:
1866                                 return $varCache[$index] = $wgContLang->formatNum( date('j') );
1867                         case MAG_PAGENAME:
1868                                 return $this->mTitle->getText();
1869                         case MAG_PAGENAMEE:
1870                                 return $this->mTitle->getPartialURL();
1871                         case MAG_REVISIONID:
1872                                 return $wgArticle->getRevIdFetched();
1873                         case MAG_NAMESPACE:
1874                                 # return Namespace::getCanonicalName($this->mTitle->getNamespace());
1875                                 return $wgContLang->getNsText($this->mTitle->getNamespace()); # Patch by Dori
1876                         case MAG_CURRENTDAYNAME:
1877                                 return $varCache[$index] = $wgContLang->getWeekdayName( date('w')+1 );
1878                         case MAG_CURRENTYEAR:
1879                                 return $varCache[$index] = $wgContLang->formatNum( date( 'Y' ), true );
1880                         case MAG_CURRENTTIME:
1881                                 return $varCache[$index] = $wgContLang->time( wfTimestampNow(), false );
1882                         case MAG_CURRENTWEEK:
1883                                 return $varCache[$index] = $wgContLang->formatNum( date('W') );
1884                         case MAG_CURRENTDOW:
1885                                 return $varCache[$index] = $wgContLang->formatNum( date('w') );
1886                         case MAG_NUMBEROFARTICLES:
1887                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfArticles() );
1888                         case MAG_NUMBEROFFILES:
1889                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfFiles() );
1890                         case MAG_SITENAME:
1891                                 return $wgSitename;
1892                         case MAG_SERVER:
1893                                 return $wgServer;
1894                         case MAG_SERVERNAME:
1895                                 return $wgServerName;
1896                         case MAG_SCRIPTPATH:
1897                                 return $wgScriptPath;
1898                         default:
1899                                 return NULL;
1900                 }
1901         }
1902
1903         /**
1904          * initialise the magic variables (like CURRENTMONTHNAME)
1905          *
1906          * @access private
1907          */
1908         function initialiseVariables() {
1909                 $fname = 'Parser::initialiseVariables';
1910                 wfProfileIn( $fname );
1911                 global $wgVariableIDs;
1912                 $this->mVariables = array();
1913                 foreach ( $wgVariableIDs as $id ) {
1914                         $mw =& MagicWord::get( $id );
1915                         $mw->addToArray( $this->mVariables, $id );
1916                 }
1917                 wfProfileOut( $fname );
1918         }
1919
1920         /**
1921          * Replace magic variables, templates, and template arguments
1922          * with the appropriate text. Templates are substituted recursively,
1923          * taking care to avoid infinite loops.
1924          *
1925          * Note that the substitution depends on value of $mOutputType:
1926          *  OT_WIKI: only {{subst:}} templates
1927          *  OT_MSG: only magic variables
1928          *  OT_HTML: all templates and magic variables
1929          *
1930          * @param string $tex The text to transform
1931          * @param array $args Key-value pairs representing template parameters to substitute
1932          * @access private
1933          */
1934         function replaceVariables( $text, $args = array() ) {
1935
1936                 # Prevent too big inclusions
1937                 if( strlen( $text ) > MAX_INCLUDE_SIZE ) {
1938                         return $text;
1939                 }
1940
1941                 $fname = 'Parser::replaceVariables';
1942                 wfProfileIn( $fname );
1943
1944                 $titleChars = Title::legalChars();
1945
1946                 # This function is called recursively. To keep track of arguments we need a stack:
1947                 array_push( $this->mArgStack, $args );
1948
1949                 # Variable substitution
1950                 $text = preg_replace_callback( "/{{([$titleChars]*?)}}/", array( &$this, 'variableSubstitution' ), $text );
1951
1952                 if ( $this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI ) {
1953                         # Argument substitution
1954                         $text = preg_replace_callback( "/{{{([$titleChars]*?)}}}/", array( &$this, 'argSubstitution' ), $text );
1955                 }
1956                 # Template substitution
1957                 $regex = '/(\\n|{)?{{(['.$titleChars.']*)(\\|.*?|)}}/s';
1958                 $text = preg_replace_callback( $regex, array( &$this, 'braceSubstitution' ), $text );
1959
1960                 array_pop( $this->mArgStack );
1961
1962                 wfProfileOut( $fname );
1963                 return $text;
1964         }
1965
1966         /**
1967          * Replace magic variables
1968          * @access private
1969          */
1970         function variableSubstitution( $matches ) {
1971                 $fname = 'parser::variableSubstitution';
1972                 $varname = $matches[1];
1973                 wfProfileIn( $fname );
1974                 if ( !$this->mVariables ) {
1975                         $this->initialiseVariables();
1976                 }
1977                 $skip = false;
1978                 if ( $this->mOutputType == OT_WIKI ) {
1979                         # Do only magic variables prefixed by SUBST
1980                         $mwSubst =& MagicWord::get( MAG_SUBST );
1981                         if (!$mwSubst->matchStartAndRemove( $varname ))
1982                                 $skip = true;
1983                         # Note that if we don't substitute the variable below,
1984                         # we don't remove the {{subst:}} magic word, in case
1985                         # it is a template rather than a magic variable.
1986                 }
1987                 if ( !$skip && array_key_exists( $varname, $this->mVariables ) ) {
1988                         $id = $this->mVariables[$varname];
1989                         $text = $this->getVariableValue( $id );
1990                         $this->mOutput->mContainsOldMagic = true;
1991                 } else {
1992                         $text = $matches[0];
1993                 }
1994                 wfProfileOut( $fname );
1995                 return $text;
1996         }
1997
1998         # Split template arguments
1999         function getTemplateArgs( $argsString ) {
2000                 if ( $argsString === '' ) {
2001                         return array();
2002                 }
2003
2004                 $args = explode( '|', substr( $argsString, 1 ) );
2005
2006                 # If any of the arguments contains a '[[' but no ']]', it needs to be
2007                 # merged with the next arg because the '|' character between belongs
2008                 # to the link syntax and not the template parameter syntax.
2009                 $argc = count($args);
2010
2011                 for ( $i = 0; $i < $argc-1; $i++ ) {
2012                         if ( substr_count ( $args[$i], '[[' ) != substr_count ( $args[$i], ']]' ) ) {
2013                                 $args[$i] .= '|'.$args[$i+1];
2014                                 array_splice($args, $i+1, 1);
2015                                 $i--;
2016                                 $argc--;
2017                         }
2018                 }
2019
2020                 return $args;
2021         }
2022
2023         /**
2024          * Return the text of a template, after recursively
2025          * replacing any variables or templates within the template.
2026          *
2027          * @param array $matches The parts of the template
2028          *  $matches[1]: the title, i.e. the part before the |
2029          *  $matches[2]: the parameters (including a leading |), if  any
2030          * @return string the text of the template
2031          * @access private
2032          */
2033         function braceSubstitution( $matches ) {
2034                 global $wgLinkCache, $wgContLang;
2035                 $fname = 'Parser::braceSubstitution';
2036                 wfProfileIn( $fname );
2037
2038                 $found = false;
2039                 $nowiki = false;
2040                 $noparse = false;
2041
2042                 $title = NULL;
2043
2044                 # Need to know if the template comes at the start of a line,
2045                 # to treat the beginning of the template like the beginning
2046                 # of a line for tables and block-level elements.
2047                 $linestart = $matches[1];
2048
2049                 # $part1 is the bit before the first |, and must contain only title characters
2050                 # $args is a list of arguments, starting from index 0, not including $part1
2051
2052                 $part1 = $matches[2];
2053                 # If the third subpattern matched anything, it will start with |
2054
2055                 $args = $this->getTemplateArgs($matches[3]);
2056                 $argc = count( $args );
2057
2058                 # Don't parse {{{}}} because that's only for template arguments
2059                 if ( $linestart === '{' ) {
2060                         $text = $matches[0];
2061                         $found = true;
2062                         $noparse = true;
2063                 }
2064
2065                 # SUBST
2066                 if ( !$found ) {
2067                         $mwSubst =& MagicWord::get( MAG_SUBST );
2068                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor ($this->mOutputType == OT_WIKI) ) {
2069                                 # One of two possibilities is true:
2070                                 # 1) Found SUBST but not in the PST phase
2071                                 # 2) Didn't find SUBST and in the PST phase
2072                                 # In either case, return without further processing
2073                                 $text = $matches[0];
2074                                 $found = true;
2075                                 $noparse = true;
2076                         }
2077                 }
2078
2079                 # MSG, MSGNW and INT
2080                 if ( !$found ) {
2081                         # Check for MSGNW:
2082                         $mwMsgnw =& MagicWord::get( MAG_MSGNW );
2083                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2084                                 $nowiki = true;
2085                         } else {
2086                                 # Remove obsolete MSG:
2087                                 $mwMsg =& MagicWord::get( MAG_MSG );
2088                                 $mwMsg->matchStartAndRemove( $part1 );
2089                         }
2090
2091                         # Check if it is an internal message
2092                         $mwInt =& MagicWord::get( MAG_INT );
2093                         if ( $mwInt->matchStartAndRemove( $part1 ) ) {
2094                                 if ( $this->incrementIncludeCount( 'int:'.$part1 ) ) {
2095                                         $text = $linestart . wfMsgReal( $part1, $args, true );
2096                                         $found = true;
2097                                 }
2098                         }
2099                 }
2100
2101                 # NS
2102                 if ( !$found ) {
2103                         # Check for NS: (namespace expansion)
2104                         $mwNs = MagicWord::get( MAG_NS );
2105                         if ( $mwNs->matchStartAndRemove( $part1 ) ) {
2106                                 if ( intval( $part1 ) ) {
2107                                         $text = $linestart . $wgContLang->getNsText( intval( $part1 ) );
2108                                         $found = true;
2109                                 } else {
2110                                         $index = Namespace::getCanonicalIndex( strtolower( $part1 ) );
2111                                         if ( !is_null( $index ) ) {
2112                                                 $text = $linestart . $wgContLang->getNsText( $index );
2113                                                 $found = true;
2114                                         }
2115                                 }
2116                         }
2117                 }
2118
2119                 # LOCALURL and LOCALURLE
2120                 if ( !$found ) {
2121                         $mwLocal = MagicWord::get( MAG_LOCALURL );
2122                         $mwLocalE = MagicWord::get( MAG_LOCALURLE );
2123
2124                         if ( $mwLocal->matchStartAndRemove( $part1 ) ) {
2125                                 $func = 'getLocalURL';
2126                         } elseif ( $mwLocalE->matchStartAndRemove( $part1 ) ) {
2127                                 $func = 'escapeLocalURL';
2128                         } else {
2129                                 $func = '';
2130                         }
2131
2132                         if ( $func !== '' ) {
2133                                 $title = Title::newFromText( $part1 );
2134                                 if ( !is_null( $title ) ) {
2135                                         if ( $argc > 0 ) {
2136                                                 $text = $linestart . $title->$func( $args[0] );
2137                                         } else {
2138                                                 $text = $linestart . $title->$func();
2139                                         }
2140                                         $found = true;
2141                                 }
2142                         }
2143                 }
2144
2145                 # GRAMMAR
2146                 if ( !$found && $argc == 1 ) {
2147                         $mwGrammar =& MagicWord::get( MAG_GRAMMAR );
2148                         if ( $mwGrammar->matchStartAndRemove( $part1 ) ) {
2149                                 $text = $linestart . $wgContLang->convertGrammar( $args[0], $part1 );
2150                                 $found = true;
2151                         }
2152                 }
2153
2154                 # Template table test
2155
2156                 # Did we encounter this template already? If yes, it is in the cache
2157                 # and we need to check for loops.
2158                 if ( !$found && isset( $this->mTemplates[$part1] ) ) {
2159                         $found = true;
2160
2161                         # Infinite loop test
2162                         if ( isset( $this->mTemplatePath[$part1] ) ) {
2163                                 $noparse = true;
2164                                 $found = true;
2165                                 $text = $linestart .
2166                                         "\{\{$part1}}" .
2167                                         '<!-- WARNING: template loop detected -->';
2168                                 wfDebug( "$fname: template loop broken at '$part1'\n" );
2169                         } else {
2170                                 # set $text to cached message.
2171                                 $text = $linestart . $this->mTemplates[$part1];
2172                         }
2173                 }
2174
2175                 # Load from database
2176                 $replaceHeadings = false;
2177                 $isHTML = false;
2178                 $lastPathLevel = $this->mTemplatePath;
2179                 if ( !$found ) {
2180                         $ns = NS_TEMPLATE;
2181                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage='' );
2182                         if ($subpage !== '') {
2183                                 $ns = $this->mTitle->getNamespace();
2184                         }
2185                         $title = Title::newFromText( $part1, $ns );
2186                         if ( !is_null( $title ) && !$title->isExternal() ) {
2187                                 # Check for excessive inclusion
2188                                 $dbk = $title->getPrefixedDBkey();
2189                                 if ( $this->incrementIncludeCount( $dbk ) ) {
2190                                         if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() ) {
2191                                                 # Capture special page output
2192                                                 $text = SpecialPage::capturePath( $title );
2193                                                 if ( $text && !is_object( $text ) ) {
2194                                                         $found = true;
2195                                                         $noparse = true;
2196                                                         $isHTML = true;
2197                                                         $this->mOutput->setCacheTime( -1 );
2198                                                 }
2199                                         } else {
2200                                                 $article = new Article( $title );
2201                                                 $articleContent = $article->getContentWithoutUsingSoManyDamnGlobals();
2202                                                 if ( $articleContent !== false ) {
2203                                                         $found = true;
2204                                                         $text = $articleContent;
2205                                                         $replaceHeadings = true;
2206                                                 }
2207                                         }
2208                                 }
2209
2210                                 # If the title is valid but undisplayable, make a link to it
2211                                 if ( $this->mOutputType == OT_HTML && !$found ) {
2212                                         $text = '[['.$title->getPrefixedText().']]';
2213                                         $found = true;
2214                                 }
2215
2216                                 # Template cache array insertion
2217                                 if( $found ) {
2218                                         $this->mTemplates[$part1] = $text;
2219                                         $text = $linestart . $text;
2220                                 }
2221                         }
2222                 }
2223
2224                 # Recursive parsing, escaping and link table handling
2225                 # Only for HTML output
2226                 if ( $nowiki && $found && $this->mOutputType == OT_HTML ) {
2227                         $text = wfEscapeWikiText( $text );
2228                 } elseif ( ($this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI) && $found && !$noparse) {
2229                         # Clean up argument array
2230                         $assocArgs = array();
2231                         $index = 1;
2232                         foreach( $args as $arg ) {
2233                                 $eqpos = strpos( $arg, '=' );
2234                                 if ( $eqpos === false ) {
2235                                         $assocArgs[$index++] = $arg;
2236                                 } else {
2237                                         $name = trim( substr( $arg, 0, $eqpos ) );
2238                                         $value = trim( substr( $arg, $eqpos+1 ) );
2239                                         if ( $value === false ) {
2240                                                 $value = '';
2241                                         }
2242                                         if ( $name !== false ) {
2243                                                 $assocArgs[$name] = $value;
2244                                         }
2245                                 }
2246                         }
2247
2248                         # Add a new element to the templace recursion path
2249                         $this->mTemplatePath[$part1] = 1;
2250
2251                         if( $this->mOutputType == OT_HTML ) {
2252                                 $text = $this->strip( $text, $this->mStripState );
2253                                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'replaceVariables' ), $assocArgs );
2254                         }
2255                         $text = $this->replaceVariables( $text, $assocArgs );
2256
2257                         # Resume the link cache and register the inclusion as a link
2258                         if ( $this->mOutputType == OT_HTML && !is_null( $title ) ) {
2259                                 $wgLinkCache->addLinkObj( $title );
2260                         }
2261
2262                         # If the template begins with a table or block-level
2263                         # element, it should be treated as beginning a new line.
2264                         if ($linestart !== '\n' && preg_match('/^({\\||:|;|#|\*)/', $text)) {
2265                                 $text = "\n" . $text;
2266                         }
2267                 }
2268                 # Prune lower levels off the recursion check path
2269                 $this->mTemplatePath = $lastPathLevel;
2270
2271                 if ( !$found ) {
2272                         wfProfileOut( $fname );
2273                         return $matches[0];
2274                 } else {
2275                         if ( $isHTML ) {
2276                                 # Replace raw HTML by a placeholder
2277                                 # Add a blank line preceding, to prevent it from mucking up
2278                                 # immediately preceding headings
2279                                 $text = "\n\n" . $this->insertStripItem( $text, $this->mStripState );
2280                         } else {
2281                                 # replace ==section headers==
2282                                 # XXX this needs to go away once we have a better parser.
2283                                 if ( $this->mOutputType != OT_WIKI && $replaceHeadings ) {
2284                                         if( !is_null( $title ) )
2285                                                 $encodedname = base64_encode($title->getPrefixedDBkey());
2286                                         else
2287                                                 $encodedname = base64_encode("");
2288                                         $m = preg_split('/(^={1,6}.*?={1,6}\s*?$)/m', $text, -1,
2289                                                 PREG_SPLIT_DELIM_CAPTURE);
2290                                         $text = '';
2291                                         $nsec = 0;
2292                                         for( $i = 0; $i < count($m); $i += 2 ) {
2293                                                 $text .= $m[$i];
2294                                                 if (!isset($m[$i + 1]) || $m[$i + 1] == "") continue;
2295                                                 $hl = $m[$i + 1];
2296                                                 if( strstr($hl, "<!--MWTEMPLATESECTION") ) {
2297                                                         $text .= $hl;
2298                                                         continue;
2299                                                 }
2300                                                 preg_match('/^(={1,6})(.*?)(={1,6})\s*?$/m', $hl, $m2);
2301                                                 $text .= $m2[1] . $m2[2] . "<!--MWTEMPLATESECTION="
2302                                                         . $encodedname . "&" . base64_encode("$nsec") . "-->" . $m2[3];
2303
2304                                                 $nsec++;
2305                                         }
2306                                 }
2307                         }
2308                 }
2309
2310                 # Prune lower levels off the recursion check path
2311                 $this->mTemplatePath = $lastPathLevel;
2312
2313                 if ( !$found ) {
2314                         wfProfileOut( $fname );
2315                         return $matches[0];
2316                 } else {
2317                         wfProfileOut( $fname );
2318                         return $text;
2319                 }
2320         }
2321
2322         /**
2323          * Triple brace replacement -- used for template arguments
2324          * @access private
2325          */
2326         function argSubstitution( $matches ) {
2327                 $arg = trim( $matches[1] );
2328                 $text = $matches[0];
2329                 $inputArgs = end( $this->mArgStack );
2330
2331                 if ( array_key_exists( $arg, $inputArgs ) ) {
2332                         $text = $inputArgs[$arg];
2333                 }
2334
2335                 return $text;
2336         }
2337
2338         /**
2339          * Returns true if the function is allowed to include this entity
2340          * @access private
2341          */
2342         function incrementIncludeCount( $dbk ) {
2343                 if ( !array_key_exists( $dbk, $this->mIncludeCount ) ) {
2344                         $this->mIncludeCount[$dbk] = 0;
2345                 }
2346                 if ( ++$this->mIncludeCount[$dbk] <= MAX_INCLUDE_REPEAT ) {
2347                         return true;
2348                 } else {
2349                         return false;
2350                 }
2351         }
2352
2353         /**
2354          * This function accomplishes several tasks:
2355          * 1) Auto-number headings if that option is enabled
2356          * 2) Add an [edit] link to sections for logged in users who have enabled the option
2357          * 3) Add a Table of contents on the top for users who have enabled the option
2358          * 4) Auto-anchor headings
2359          *
2360          * It loops through all headlines, collects the necessary data, then splits up the
2361          * string and re-inserts the newly formatted headlines.
2362          *
2363          * @param string $text
2364          * @param boolean $isMain
2365          * @access private
2366          */
2367         function formatHeadings( $text, $isMain=true ) {
2368                 global $wgMaxTocLevel, $wgContLang, $wgLinkHolders, $wgInterwikiLinkHolders;
2369
2370                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
2371                 $doShowToc = true;
2372                 $forceTocHere = false;
2373                 if( !$this->mTitle->userCanEdit() ) {
2374                         $showEditLink = 0;
2375                 } else {
2376                         $showEditLink = $this->mOptions->getEditSection();
2377                 }
2378
2379                 # Inhibit editsection links if requested in the page
2380                 $esw =& MagicWord::get( MAG_NOEDITSECTION );
2381                 if( $esw->matchAndRemove( $text ) ) {
2382                         $showEditLink = 0;
2383                 }
2384                 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
2385                 # do not add TOC
2386                 $mw =& MagicWord::get( MAG_NOTOC );
2387                 if( $mw->matchAndRemove( $text ) ) {
2388                         $doShowToc = false;
2389                 }
2390
2391                 # Get all headlines for numbering them and adding funky stuff like [edit]
2392                 # links - this is for later, but we need the number of headlines right now
2393                 $numMatches = preg_match_all( '/<H([1-6])(.*?'.'>)(.*?)<\/H[1-6] *>/i', $text, $matches );
2394
2395                 # if there are fewer than 4 headlines in the article, do not show TOC
2396                 if( $numMatches < 4 ) {
2397                         $doShowToc = false;
2398                 }
2399
2400                 # if the string __TOC__ (not case-sensitive) occurs in the HTML,
2401                 # override above conditions and always show TOC at that place
2402
2403                 $mw =& MagicWord::get( MAG_TOC );
2404                 if($mw->match( $text ) ) {
2405                         $doShowToc = true;
2406                         $forceTocHere = true;
2407                 } else {
2408                         # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
2409                         # override above conditions and always show TOC above first header
2410                         $mw =& MagicWord::get( MAG_FORCETOC );
2411                         if ($mw->matchAndRemove( $text ) ) {
2412                                 $doShowToc = true;
2413                         }
2414                 }
2415
2416                 # Never ever show TOC if no headers
2417                 if( $numMatches < 1 ) {
2418                         $doShowToc = false;
2419                 }
2420
2421                 # We need this to perform operations on the HTML
2422                 $sk =& $this->mOptions->getSkin();
2423
2424                 # headline counter
2425                 $headlineCount = 0;
2426                 $sectionCount = 0; # headlineCount excluding template sections
2427
2428                 # Ugh .. the TOC should have neat indentation levels which can be
2429                 # passed to the skin functions. These are determined here
2430                 $toc = '';
2431                 $full = '';
2432                 $head = array();
2433                 $sublevelCount = array();
2434                 $levelCount = array();
2435                 $toclevel = 0;
2436                 $level = 0;
2437                 $prevlevel = 0;
2438                 $toclevel = 0;
2439                 $prevtoclevel = 0;
2440
2441                 foreach( $matches[3] as $headline ) {
2442                         $istemplate = 0;
2443                         $templatetitle = '';
2444                         $templatesection = 0;
2445                         $numbering = '';
2446
2447                         if (preg_match("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", $headline, $mat)) {
2448                                 $istemplate = 1;
2449                                 $templatetitle = base64_decode($mat[1]);
2450                                 $templatesection = 1 + (int)base64_decode($mat[2]);
2451                                 $headline = preg_replace("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", "", $headline);
2452                         }
2453
2454                         if( $toclevel ) {
2455                                 $prevlevel = $level;
2456                                 $prevtoclevel = $toclevel;
2457                         }
2458                         $level = $matches[1][$headlineCount];
2459
2460                         if( $doNumberHeadings || $doShowToc ) {
2461
2462                                 if ( $level > $prevlevel ) {
2463                                         # Increase TOC level
2464                                         $toclevel++;
2465                                         $sublevelCount[$toclevel] = 0;
2466                                         $toc .= $sk->tocIndent();
2467                                 }
2468                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
2469                                         # Decrease TOC level, find level to jump to
2470
2471                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
2472                                                 # Can only go down to level 1
2473                                                 $toclevel = 1;
2474                                         } else {
2475                                                 for ($i = $toclevel; $i > 0; $i--) {
2476                                                         if ( $levelCount[$i] == $level ) {
2477                                                                 # Found last matching level
2478                                                                 $toclevel = $i;
2479                                                                 break;
2480                                                         }
2481                                                         elseif ( $levelCount[$i] < $level ) {
2482                                                                 # Found first matching level below current level
2483                                                                 $toclevel = $i + 1;
2484                                                                 break;
2485                                                         }
2486                                                 }
2487                                         }
2488
2489                                         $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
2490                                 }
2491                                 else {
2492                                         # No change in level, end TOC line
2493                                         $toc .= $sk->tocLineEnd();
2494                                 }
2495
2496                                 $levelCount[$toclevel] = $level;
2497
2498                                 # count number of headlines for each level
2499                                 @$sublevelCount[$toclevel]++;
2500                                 $dot = 0;
2501                                 for( $i = 1; $i <= $toclevel; $i++ ) {
2502                                         if( !empty( $sublevelCount[$i] ) ) {
2503                                                 if( $dot ) {
2504                                                         $numbering .= '.';
2505                                                 }
2506                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
2507                                                 $dot = 1;
2508                                         }
2509                                 }
2510                         }
2511
2512                         # The canonized header is a version of the header text safe to use for links
2513                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
2514                         $canonized_headline = $this->unstrip( $headline, $this->mStripState );
2515                         $canonized_headline = $this->unstripNoWiki( $canonized_headline, $this->mStripState );
2516
2517                         # Remove link placeholders by the link text.
2518                         #     <!--LINK number-->
2519                         # turns into
2520                         #     link text with suffix
2521                         $canonized_headline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
2522                                                             "\$this->mLinkHolders['texts'][\$1]",
2523                                                             $canonized_headline );
2524                         $canonized_headline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
2525                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
2526                                                             $canonized_headline );
2527
2528                         # strip out HTML
2529                         $canonized_headline = preg_replace( '/<.*?' . '>/','',$canonized_headline );
2530                         $tocline = trim( $canonized_headline );
2531                         $canonized_headline = urlencode( Sanitizer::decodeCharReferences( str_replace(' ', '_', $tocline) ) );
2532                         $replacearray = array(
2533                                 '%3A' => ':',
2534                                 '%' => '.'
2535                         );
2536                         $canonized_headline = str_replace(array_keys($replacearray),array_values($replacearray),$canonized_headline);
2537                         $refers[$headlineCount] = $canonized_headline;
2538
2539                         # count how many in assoc. array so we can track dupes in anchors
2540                         @$refers[$canonized_headline]++;
2541                         $refcount[$headlineCount]=$refers[$canonized_headline];
2542
2543                         # Don't number the heading if it is the only one (looks silly)
2544                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
2545                                 # the two are different if the line contains a link
2546                                 $headline=$numbering . ' ' . $headline;
2547                         }
2548
2549                         # Create the anchor for linking from the TOC to the section
2550                         $anchor = $canonized_headline;
2551                         if($refcount[$headlineCount] > 1 ) {
2552                                 $anchor .= '_' . $refcount[$headlineCount];
2553                         }
2554                         if( $doShowToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
2555                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
2556                         }
2557                         if( $showEditLink && ( !$istemplate || $templatetitle !== "" ) ) {
2558                                 if ( empty( $head[$headlineCount] ) ) {
2559                                         $head[$headlineCount] = '';
2560                                 }
2561                                 if( $istemplate )
2562                                         $head[$headlineCount] .= $sk->editSectionLinkForOther($templatetitle, $templatesection);
2563                                 else
2564                                         $head[$headlineCount] .= $sk->editSectionLink($this->mTitle, $sectionCount+1);
2565                         }
2566
2567                         # give headline the correct <h#> tag
2568                         @$head[$headlineCount] .= "<a name=\"$anchor\"></a><h".$level.$matches[2][$headlineCount] .$headline.'</h'.$level.'>';
2569
2570                         $headlineCount++;
2571                         if( !$istemplate )
2572                                 $sectionCount++;
2573                 }
2574
2575                 if( $doShowToc ) {
2576                         $toc .= $sk->tocUnindent( $toclevel - 1 );
2577                         $toc = $sk->tocList( $toc );
2578                 }
2579
2580                 # split up and insert constructed headlines
2581
2582                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
2583                 $i = 0;
2584
2585                 foreach( $blocks as $block ) {
2586                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
2587                                 # This is the [edit] link that appears for the top block of text when
2588                                 # section editing is enabled
2589
2590                                 # Disabled because it broke block formatting
2591                                 # For example, a bullet point in the top line
2592                                 # $full .= $sk->editSectionLink(0);
2593                         }
2594                         $full .= $block;
2595                         if( $doShowToc && !$i && $isMain && !$forceTocHere) {
2596                         # Top anchor now in skin
2597                                 $full = $full.$toc;
2598                         }
2599
2600                         if( !empty( $head[$i] ) ) {
2601                                 $full .= $head[$i];
2602                         }
2603                         $i++;
2604                 }
2605                 if($forceTocHere) {
2606                         $mw =& MagicWord::get( MAG_TOC );
2607                         return $mw->replace( $toc, $full );
2608                 } else {
2609                         return $full;
2610                 }
2611         }
2612
2613         /**
2614          * Return an HTML link for the "ISBN 123456" text
2615          * @access private
2616          */
2617         function magicISBN( $text ) {
2618                 $fname = 'Parser::magicISBN';
2619                 wfProfileIn( $fname );
2620
2621                 $a = split( 'ISBN ', ' '.$text );
2622                 if ( count ( $a ) < 2 ) {
2623                         wfProfileOut( $fname );
2624                         return $text;
2625                 }
2626                 $text = substr( array_shift( $a ), 1);
2627                 $valid = '0123456789-Xx';
2628
2629                 foreach ( $a as $x ) {
2630                         $isbn = $blank = '' ;
2631                         while ( ' ' == $x{0} ) {
2632                                 $blank .= ' ';
2633                                 $x = substr( $x, 1 );
2634                         }
2635                         if ( $x == '' ) { # blank isbn
2636                                 $text .= "ISBN $blank";
2637                                 continue;
2638                         }
2639                         while ( strstr( $valid, $x{0} ) != false ) {
2640                                 $isbn .= $x{0};
2641                                 $x = substr( $x, 1 );
2642                         }
2643                         $num = str_replace( '-', '', $isbn );
2644                         $num = str_replace( ' ', '', $num );
2645                         $num = str_replace( 'x', 'X', $num );
2646
2647                         if ( '' == $num ) {
2648                                 $text .= "ISBN $blank$x";
2649                         } else {
2650                                 $titleObj = Title::makeTitle( NS_SPECIAL, 'Booksources' );
2651                                 $text .= '<a href="' .
2652                                 $titleObj->escapeLocalUrl( 'isbn='.$num ) .
2653                                         "\" class=\"internal\">ISBN $isbn</a>";
2654                                 $text .= $x;
2655                         }
2656                 }
2657                 wfProfileOut( $fname );
2658                 return $text;
2659         }
2660
2661         /**
2662          * Return an HTML link for the "RFC 1234" text
2663          *
2664          * @access private
2665          * @param string $text     Text to be processed
2666          * @param string $keyword  Magic keyword to use (default RFC)
2667          * @param string $urlmsg   Interface message to use (default rfcurl)
2668          * @return string
2669          */
2670         function magicRFC( $text, $keyword='RFC ', $urlmsg='rfcurl'  ) {
2671
2672                 $valid = '0123456789';
2673                 $internal = false;
2674
2675                 $a = split( $keyword, ' '.$text );
2676                 if ( count ( $a ) < 2 ) {
2677                         return $text;
2678                 }
2679                 $text = substr( array_shift( $a ), 1);
2680
2681                 /* Check if keyword is preceed by [[.
2682                  * This test is made here cause of the array_shift above
2683                  * that prevent the test to be done in the foreach.
2684                  */
2685                 if ( substr( $text, -2 ) == '[[' ) {
2686                         $internal = true;
2687                 }
2688
2689                 foreach ( $a as $x ) {
2690                         /* token might be empty if we have RFC RFC 1234 */
2691                         if ( $x=='' ) {
2692                                 $text.=$keyword;
2693                                 continue;
2694                                 }
2695
2696                         $id = $blank = '' ;
2697
2698                         /** remove and save whitespaces in $blank */
2699                         while ( $x{0} == ' ' ) {
2700                                 $blank .= ' ';
2701                                 $x = substr( $x, 1 );
2702                         }
2703
2704                         /** remove and save the rfc number in $id */
2705                         while ( strstr( $valid, $x{0} ) != false ) {
2706                                 $id .= $x{0};
2707                                 $x = substr( $x, 1 );
2708                         }
2709
2710                         if ( $id == '' ) {
2711                                 /* call back stripped spaces*/
2712                                 $text .= $keyword.$blank.$x;
2713                         } elseif( $internal ) {
2714                                 /* normal link */
2715                                 $text .= $keyword.$id.$x;
2716                         } else {
2717                                 /* build the external link*/
2718                                 $url = wfMsg( $urlmsg, $id);
2719                                 $sk =& $this->mOptions->getSkin();
2720                                 $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
2721                                 $text .= "<a href='{$url}'{$la}>{$keyword}{$id}</a>{$x}";
2722                         }
2723
2724                         /* Check if the next RFC keyword is preceed by [[ */
2725                         $internal = ( substr($x,-2) == '[[' );
2726                 }
2727                 return $text;
2728         }
2729
2730         /**
2731          * Transform wiki markup when saving a page by doing \r\n -> \n
2732          * conversion, substitting signatures, {{subst:}} templates, etc.
2733          *
2734          * @param string $text the text to transform
2735          * @param Title &$title the Title object for the current article
2736          * @param User &$user the User object describing the current user
2737          * @param ParserOptions $options parsing options
2738          * @param bool $clearState whether to clear the parser state first
2739          * @return string the altered wiki markup
2740          * @access public
2741          */
2742         function preSaveTransform( $text, &$title, &$user, $options, $clearState = true ) {
2743                 $this->mOptions = $options;
2744                 $this->mTitle =& $title;
2745                 $this->mOutputType = OT_WIKI;
2746
2747                 if ( $clearState ) {
2748                         $this->clearState();
2749                 }
2750
2751                 $stripState = false;
2752                 $pairs = array(
2753                         "\r\n" => "\n",
2754                 );
2755                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
2756                 $text = $this->strip( $text, $stripState, false );
2757                 $text = $this->pstPass2( $text, $user );
2758                 $text = $this->unstrip( $text, $stripState );
2759                 $text = $this->unstripNoWiki( $text, $stripState );
2760                 return $text;
2761         }
2762
2763         /**
2764          * Pre-save transform helper function
2765          * @access private
2766          */
2767         function pstPass2( $text, &$user ) {
2768                 global $wgContLang, $wgLocaltimezone;
2769
2770                 # Variable replacement
2771                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
2772                 $text = $this->replaceVariables( $text );
2773
2774                 # Signatures
2775                 #
2776                 $n = $user->getName();
2777                 $k = $user->getOption( 'nickname' );
2778                 if ( '' == $k ) { $k = $n; }
2779                 if ( isset( $wgLocaltimezone ) ) {
2780                         $oldtz = getenv( 'TZ' );
2781                         putenv( 'TZ='.$wgLocaltimezone );
2782                 }
2783
2784                 /* Note: This is the timestamp saved as hardcoded wikitext to
2785                  * the database, we use $wgContLang here in order to give
2786                  * everyone the same signiture and use the default one rather
2787                  * than the one selected in each users preferences.
2788                  */
2789                 $d = $wgContLang->timeanddate( wfTimestampNow(), false, false) .
2790                   ' (' . date( 'T' ) . ')';
2791                 if ( isset( $wgLocaltimezone ) ) {
2792                         putenv( 'TZ='.$oldtz );
2793                 }
2794
2795                 if( $user->getOption( 'fancysig' ) ) {
2796                         $sigText = $k;
2797                 } else {
2798                         $sigText = '[[' . $wgContLang->getNsText( NS_USER ) . ":$n|$k]]";
2799                 }
2800                 $text = preg_replace( '/~~~~~/', $d, $text );
2801                 $text = preg_replace( '/~~~~/', "$sigText $d", $text );
2802                 $text = preg_replace( '/~~~/', $sigText, $text );
2803
2804                 # Context links: [[|name]] and [[name (context)|]]
2805                 #
2806                 $tc = "[&;%\\-,.\\(\\)' _0-9A-Za-z\\/:\\x80-\\xff]";
2807                 $np = "[&;%\\-,.' _0-9A-Za-z\\/:\\x80-\\xff]"; # No parens
2808                 $namespacechar = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
2809                 $conpat = "/^({$np}+) \\(({$tc}+)\\)$/";
2810
2811                 $p1 = "/\[\[({$np}+) \\(({$np}+)\\)\\|]]/";             # [[page (context)|]]
2812                 $p2 = "/\[\[\\|({$tc}+)]]/";                                    # [[|page]]
2813                 $p3 = "/\[\[(:*$namespacechar+):({$np}+)\\|]]/";                # [[namespace:page|]] and [[:namespace:page|]]
2814                 $p4 = "/\[\[(:*$namespacechar+):({$np}+) \\(({$np}+)\\)\\|]]/"; # [[ns:page (cont)|]] and [[:ns:page (cont)|]]
2815                 $context = '';
2816                 $t = $this->mTitle->getText();
2817                 if ( preg_match( $conpat, $t, $m ) ) {
2818                         $context = $m[2];
2819                 }
2820                 $text = preg_replace( $p4, '[[\\1:\\2 (\\3)|\\2]]', $text );
2821                 $text = preg_replace( $p1, '[[\\1 (\\2)|\\1]]', $text );
2822                 $text = preg_replace( $p3, '[[\\1:\\2|\\2]]', $text );
2823
2824                 if ( '' == $context ) {
2825                         $text = preg_replace( $p2, '[[\\1]]', $text );
2826                 } else {
2827                         $text = preg_replace( $p2, "[[\\1 ({$context})|\\1]]", $text );
2828                 }
2829
2830                 # Trim trailing whitespace
2831                 # MAG_END (__END__) tag allows for trailing
2832                 # whitespace to be deliberately included
2833                 $text = rtrim( $text );
2834                 $mw =& MagicWord::get( MAG_END );
2835                 $mw->matchAndRemove( $text );
2836
2837                 return $text;
2838         }
2839
2840         /**
2841          * Set up some variables which are usually set up in parse()
2842          * so that an external function can call some class members with confidence
2843          * @access public
2844          */
2845         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
2846                 $this->mTitle =& $title;
2847                 $this->mOptions = $options;
2848                 $this->mOutputType = $outputType;
2849                 if ( $clearState ) {
2850                         $this->clearState();
2851                 }
2852         }
2853
2854         /**
2855          * Transform a MediaWiki message by replacing magic variables.
2856          *
2857          * @param string $text the text to transform
2858          * @param ParserOptions $options  options
2859          * @return string the text with variables substituted
2860          * @access public
2861          */
2862         function transformMsg( $text, $options ) {
2863                 global $wgTitle;
2864                 static $executing = false;
2865
2866                 # Guard against infinite recursion
2867                 if ( $executing ) {
2868                         return $text;
2869                 }
2870                 $executing = true;
2871
2872                 $this->mTitle = $wgTitle;
2873                 $this->mOptions = $options;
2874                 $this->mOutputType = OT_MSG;
2875                 $this->clearState();
2876                 $text = $this->replaceVariables( $text );
2877
2878                 $executing = false;
2879                 return $text;
2880         }
2881
2882         /**
2883          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
2884          * Callback will be called with the text within
2885          * Transform and return the text within
2886          * @access public
2887          */
2888         function setHook( $tag, $callback ) {
2889                 $oldVal = @$this->mTagHooks[$tag];
2890                 $this->mTagHooks[$tag] = $callback;
2891                 return $oldVal;
2892         }
2893
2894         /**
2895          * Replace <!--LINK--> link placeholders with actual links, in the buffer
2896          * Placeholders created in Skin::makeLinkObj()
2897          * Returns an array of links found, indexed by PDBK:
2898          *  0 - broken
2899          *  1 - normal link
2900          *  2 - stub
2901          * $options is a bit field, RLH_FOR_UPDATE to select for update
2902          */
2903         function replaceLinkHolders( &$text, $options = 0 ) {
2904                 global $wgUser, $wgLinkCache;
2905                 global $wgOutputReplace;
2906
2907                 $fname = 'Parser::replaceLinkHolders';
2908                 wfProfileIn( $fname );
2909
2910                 $pdbks = array();
2911                 $colours = array();
2912                 $sk = $this->mOptions->getSkin();
2913
2914                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
2915                         wfProfileIn( $fname.'-check' );
2916                         $dbr =& wfGetDB( DB_SLAVE );
2917                         $page = $dbr->tableName( 'page' );
2918                         $threshold = $wgUser->getOption('stubthreshold');
2919
2920                         # Sort by namespace
2921                         asort( $this->mLinkHolders['namespaces'] );
2922
2923                         # Generate query
2924                         $query = false;
2925                         foreach ( $this->mLinkHolders['namespaces'] as $key => $val ) {
2926                                 # Make title object
2927                                 $title = $this->mLinkHolders['titles'][$key];
2928
2929                                 # Skip invalid entries.
2930                                 # Result will be ugly, but prevents crash.
2931                                 if ( is_null( $title ) ) {
2932                                         continue;
2933                                 }
2934                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
2935
2936                                 # Check if it's in the link cache already
2937                                 if ( $wgLinkCache->getGoodLinkID( $pdbk ) ) {
2938                                         $colours[$pdbk] = 1;
2939                                 } elseif ( $wgLinkCache->isBadLink( $pdbk ) ) {
2940                                         $colours[$pdbk] = 0;
2941                                 } else {
2942                                         # Not in the link cache, add it to the query
2943                                         if ( !isset( $current ) ) {
2944                                                 $current = $val;
2945                                                 $query =  "SELECT page_id, page_namespace, page_title";
2946                                                 if ( $threshold > 0 ) {
2947                                                         $query .= ', page_len, page_is_redirect';
2948                                                 }
2949                                                 $query .= " FROM $page WHERE (page_namespace=$val AND page_title IN(";
2950                                         } elseif ( $current != $val ) {
2951                                                 $current = $val;
2952                                                 $query .= ")) OR (page_namespace=$val AND page_title IN(";
2953                                         } else {
2954                                                 $query .= ', ';
2955                                         }
2956
2957                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
2958                                 }
2959                         }
2960                         if ( $query ) {
2961                                 $query .= '))';
2962                                 if ( $options & RLH_FOR_UPDATE ) {
2963                                         $query .= ' FOR UPDATE';
2964                                 }
2965
2966                                 $res = $dbr->query( $query, $fname );
2967
2968                                 # Fetch data and form into an associative array
2969                                 # non-existent = broken
2970                                 # 1 = known
2971                                 # 2 = stub
2972                                 while ( $s = $dbr->fetchObject($res) ) {
2973                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
2974                                         $pdbk = $title->getPrefixedDBkey();
2975                                         $wgLinkCache->addGoodLinkObj( $s->page_id, $title );
2976
2977                                         if ( $threshold >  0 ) {
2978                                                 $size = $s->page_len;
2979                                                 if ( $s->page_is_redirect || $s->page_namespace != 0 || $size >= $threshold ) {
2980                                                         $colours[$pdbk] = 1;
2981                                                 } else {
2982                                                         $colours[$pdbk] = 2;
2983                                                 }
2984                                         } else {
2985                                                 $colours[$pdbk] = 1;
2986                                         }
2987                                 }
2988                         }
2989                         wfProfileOut( $fname.'-check' );
2990
2991                         # Construct search and replace arrays
2992                         wfProfileIn( $fname.'-construct' );
2993                         $wgOutputReplace = array();
2994                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
2995                                 $pdbk = $pdbks[$key];
2996                                 $searchkey = "<!--LINK $key-->";
2997                                 $title = $this->mLinkHolders['titles'][$key];
2998                                 if ( empty( $colours[$pdbk] ) ) {
2999                                         $wgLinkCache->addBadLinkObj( $title );
3000                                         $colours[$pdbk] = 0;
3001                                         $wgOutputReplace[$searchkey] = $sk->makeBrokenLinkObj( $title,
3002                                                                         $this->mLinkHolders['texts'][$key],
3003                                                                         $this->mLinkHolders['queries'][$key] );
3004                                 } elseif ( $colours[$pdbk] == 1 ) {
3005                                         $wgOutputReplace[$searchkey] = $sk->makeKnownLinkObj( $title,
3006                                                                         $this->mLinkHolders['texts'][$key],
3007                                                                         $this->mLinkHolders['queries'][$key] );
3008                                 } elseif ( $colours[$pdbk] == 2 ) {
3009                                         $wgOutputReplace[$searchkey] = $sk->makeStubLinkObj( $title,
3010                                                                         $this->mLinkHolders['texts'][$key],
3011                                                                         $this->mLinkHolders['queries'][$key] );
3012                                 }
3013                         }
3014                         wfProfileOut( $fname.'-construct' );
3015
3016                         # Do the thing
3017                         wfProfileIn( $fname.'-replace' );
3018
3019                         $text = preg_replace_callback(
3020                                 '/(<!--LINK .*?-->)/',
3021                                 "wfOutputReplaceMatches",
3022                                 $text);
3023
3024                         wfProfileOut( $fname.'-replace' );
3025                 }
3026
3027                 # Now process interwiki link holders
3028                 # This is quite a bit simpler than internal links
3029                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
3030                         wfProfileIn( $fname.'-interwiki' );
3031                         # Make interwiki link HTML
3032                         $wgOutputReplace = array();
3033                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
3034                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
3035                                 $wgOutputReplace[$key] = $sk->makeLinkObj( $title, $link );
3036                         }
3037
3038                         $text = preg_replace_callback(
3039                                 '/<!--IWLINK (.*?)-->/',
3040                                 "wfOutputReplaceMatches",
3041                                 $text );
3042                         wfProfileOut( $fname.'-interwiki' );
3043                 }
3044
3045                 wfProfileOut( $fname );
3046                 return $colours;
3047         }
3048
3049         /**
3050          * Replace <!--LINK--> link placeholders with plain text of links
3051          * (not HTML-formatted).
3052          * @param string $text
3053          * @return string
3054          */
3055         function replaceLinkHoldersText( $text ) {
3056                 global $wgUser, $wgLinkCache;
3057                 global $wgOutputReplace;
3058
3059                 $fname = 'Parser::replaceLinkHoldersText';
3060                 wfProfileIn( $fname );
3061
3062                 $text = preg_replace_callback(
3063                         '/<!--(LINK|IWLINK) (.*?)-->/',
3064                         array( &$this, 'replaceLinkHoldersTextCallback' ),
3065                         $text );
3066
3067                 wfProfileOut( $fname );
3068                 return $text;
3069         }
3070
3071         /**
3072          * @param array $matches
3073          * @return string
3074          * @access private
3075          */
3076         function replaceLinkHoldersTextCallback( $matches ) {
3077                 $type = $matches[1];
3078                 $key  = $matches[2];
3079                 if( $type == 'LINK' ) {
3080                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
3081                                 return $this->mLinkHolders['texts'][$key];
3082                         }
3083                 } elseif( $type == 'IWLINK' ) {
3084                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
3085                                 return $this->mInterwikiLinkHolders['texts'][$key];
3086                         }
3087                 }
3088                 return $matches[0];
3089         }
3090
3091         /**
3092          * Renders an image gallery from a text with one line per image.
3093          * text labels may be given by using |-style alternative text. E.g.
3094          *   Image:one.jpg|The number "1"
3095          *   Image:tree.jpg|A tree
3096          * given as text will return the HTML of a gallery with two images,
3097          * labeled 'The number "1"' and
3098          * 'A tree'.
3099          *
3100          * @static
3101          */
3102         function renderImageGallery( $text ) {
3103                 # Setup the parser
3104                 global $wgUser, $wgTitle;
3105                 $parserOptions = ParserOptions::newFromUser( $wgUser );
3106                 $localParser = new Parser();
3107
3108                 global $wgLinkCache;
3109                 $ig = new ImageGallery();
3110                 $ig->setShowBytes( false );
3111                 $ig->setShowFilename( false );
3112                 $lines = explode( "\n", $text );
3113
3114                 foreach ( $lines as $line ) {
3115                         # match lines like these:
3116                         # Image:someimage.jpg|This is some image
3117                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
3118                         # Skip empty lines
3119                         if ( count( $matches ) == 0 ) {
3120                                 continue;
3121                         }
3122                         $nt = Title::newFromURL( $matches[1] );
3123                         if( is_null( $nt ) ) {
3124                                 # Bogus title. Ignore these so we don't bomb out later.
3125                                 continue;
3126                         }
3127                         if ( isset( $matches[3] ) ) {
3128                                 $label = $matches[3];
3129                         } else {
3130                                 $label = '';
3131                         }
3132
3133                         $html = $localParser->parse( $label , $wgTitle, $parserOptions );
3134                         $html = $html->mText;
3135
3136                         $ig->add( new Image( $nt ), $html );
3137                         $wgLinkCache->addImageLinkObj( $nt );
3138                 }
3139                 return $ig->toHTML();
3140         }
3141
3142         /**
3143          * Parse image options text and use it to make an image
3144          */
3145         function makeImage( &$nt, $options ) {
3146                 global $wgContLang, $wgUseImageResize;
3147                 global $wgUser, $wgThumbLimits;
3148
3149                 $align = '';
3150
3151                 # Check if the options text is of the form "options|alt text"
3152                 # Options are:
3153                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
3154                 #  * left               no resizing, just left align. label is used for alt= only
3155                 #  * right              same, but right aligned
3156                 #  * none               same, but not aligned
3157                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
3158                 #  * center             center the image
3159                 #  * framed             Keep original image size, no magnify-button.
3160
3161                 $part = explode( '|', $options);
3162
3163                 $mwThumb  =& MagicWord::get( MAG_IMG_THUMBNAIL );
3164                 $mwLeft   =& MagicWord::get( MAG_IMG_LEFT );
3165                 $mwRight  =& MagicWord::get( MAG_IMG_RIGHT );
3166                 $mwNone   =& MagicWord::get( MAG_IMG_NONE );
3167                 $mwWidth  =& MagicWord::get( MAG_IMG_WIDTH );
3168                 $mwCenter =& MagicWord::get( MAG_IMG_CENTER );
3169                 $mwFramed =& MagicWord::get( MAG_IMG_FRAMED );
3170                 $caption = '';
3171
3172                 $width = $height = $framed = $thumb = false;
3173                 $manual_thumb = '' ;
3174
3175                 foreach( $part as $key => $val ) {
3176                         $val_parts = explode ( '=' , $val , 2 ) ;
3177                         $left_part = array_shift ( $val_parts ) ;
3178                         if ( $wgUseImageResize && ! is_null( $mwThumb->matchVariableStartToEnd($val) ) ) {
3179                                 $thumb=true;
3180                         } elseif ( $wgUseImageResize && count ( $val_parts ) == 1 && ! is_null( $mwThumb->matchVariableStartToEnd($left_part) ) ) {
3181                                 # use manually specified thumbnail
3182                                 $thumb=true;
3183                                 $manual_thumb = array_shift ( $val_parts ) ;
3184                         } elseif ( ! is_null( $mwRight->matchVariableStartToEnd($val) ) ) {
3185                                 # remember to set an alignment, don't render immediately
3186                                 $align = 'right';
3187                         } elseif ( ! is_null( $mwLeft->matchVariableStartToEnd($val) ) ) {
3188                                 # remember to set an alignment, don't render immediately
3189                                 $align = 'left';
3190                         } elseif ( ! is_null( $mwCenter->matchVariableStartToEnd($val) ) ) {
3191                                 # remember to set an alignment, don't render immediately
3192                                 $align = 'center';
3193                         } elseif ( ! is_null( $mwNone->matchVariableStartToEnd($val) ) ) {
3194                                 # remember to set an alignment, don't render immediately
3195                                 $align = 'none';
3196                         } elseif ( $wgUseImageResize && ! is_null( $match = $mwWidth->matchVariableStartToEnd($val) ) ) {
3197                                 wfDebug( "MAG_IMG_WIDTH match: $match\n" );
3198                                 # $match is the image width in pixels
3199                                 if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $match, $m ) ) {
3200                                         $width = intval( $m[1] );
3201                                         $height = intval( $m[2] );
3202                                 } else {
3203                                         $width = intval($match);
3204                                 }
3205                         } elseif ( ! is_null( $mwFramed->matchVariableStartToEnd($val) ) ) {
3206                                 $framed=true;
3207                         } else {
3208                                 $caption = $val;
3209                         }
3210                 }
3211                 # Strip bad stuff out of the alt text
3212                 $alt = $this->replaceLinkHoldersText( $caption );
3213                 $alt = Sanitizer::stripAllTags( $alt );
3214
3215                 # Linker does the rest
3216                 $sk =& $this->mOptions->getSkin();
3217                 return $sk->makeImageLinkObj( $nt, $caption, $alt, $align, $width, $height, $framed, $thumb, $manual_thumb );
3218         }
3219 }
3220
3221 /**
3222  * @todo document
3223  * @package MediaWiki
3224  */
3225 class ParserOutput
3226 {
3227         var $mText, $mLanguageLinks, $mCategoryLinks, $mContainsOldMagic;
3228         var $mCacheTime; # Timestamp on this article, or -1 for uncacheable. Used in ParserCache.
3229         var $mVersion;   # Compatibility check
3230         var $mTitleText; # title text of the chosen language variant
3231
3232         function ParserOutput( $text = '', $languageLinks = array(), $categoryLinks = array(),
3233                 $containsOldMagic = false, $titletext = '' )
3234         {
3235                 $this->mText = $text;
3236                 $this->mLanguageLinks = $languageLinks;
3237                 $this->mCategoryLinks = $categoryLinks;
3238                 $this->mContainsOldMagic = $containsOldMagic;
3239                 $this->mCacheTime = '';
3240                 $this->mVersion = MW_PARSER_VERSION;
3241                 $this->mTitleText = $titletext;
3242         }
3243
3244         function getText()                   { return $this->mText; }
3245         function getLanguageLinks()          { return $this->mLanguageLinks; }
3246         function getCategoryLinks()          { return array_keys( $this->mCategoryLinks ); }
3247         function getCacheTime()              { return $this->mCacheTime; }
3248         function getTitleText()              { return $this->mTitleText; }
3249         function containsOldMagic()          { return $this->mContainsOldMagic; }
3250         function setText( $text )            { return wfSetVar( $this->mText, $text ); }
3251         function setLanguageLinks( $ll )     { return wfSetVar( $this->mLanguageLinks, $ll ); }
3252         function setCategoryLinks( $cl )     { return wfSetVar( $this->mCategoryLinks, $cl ); }
3253         function setContainsOldMagic( $com ) { return wfSetVar( $this->mContainsOldMagic, $com ); }
3254         function setCacheTime( $t )          { return wfSetVar( $this->mCacheTime, $t ); }
3255         function setTitleText( $t )          { return wfSetVar ($this->mTitleText, $t); }
3256
3257         function addCategoryLink( $c )       { $this->mCategoryLinks[$c] = 1; }
3258
3259         function merge( $other ) {
3260                 $this->mLanguageLinks = array_merge( $this->mLanguageLinks, $other->mLanguageLinks );
3261                 $this->mCategoryLinks = array_merge( $this->mCategoryLinks, $this->mLanguageLinks );
3262                 $this->mContainsOldMagic = $this->mContainsOldMagic || $other->mContainsOldMagic;
3263         }
3264
3265         /**
3266          * Return true if this cached output object predates the global or
3267          * per-article cache invalidation timestamps, or if it comes from
3268          * an incompatible older version.
3269          *
3270          * @param string $touched the affected article's last touched timestamp
3271          * @return bool
3272          * @access public
3273          */
3274         function expired( $touched ) {
3275                 global $wgCacheEpoch;
3276                 return $this->getCacheTime() == -1 || // parser says it's uncacheable
3277                        $this->getCacheTime() <= $touched ||
3278                        $this->getCacheTime() <= $wgCacheEpoch ||
3279                        !isset( $this->mVersion ) ||
3280                        version_compare( $this->mVersion, MW_PARSER_VERSION, "lt" );
3281         }
3282 }
3283
3284 /**
3285  * Set options of the Parser
3286  * @todo document
3287  * @package MediaWiki
3288  */
3289 class ParserOptions
3290 {
3291         # All variables are private
3292         var $mUseTeX;                    # Use texvc to expand <math> tags
3293         var $mUseDynamicDates;           # Use DateFormatter to format dates
3294         var $mInterwikiMagic;            # Interlanguage links are removed and returned in an array
3295         var $mAllowExternalImages;       # Allow external images inline
3296         var $mSkin;                      # Reference to the preferred skin
3297         var $mDateFormat;                # Date format index
3298         var $mEditSection;               # Create "edit section" links
3299         var $mNumberHeadings;            # Automatically number headings
3300         var $mAllowSpecialInclusion;     # Allow inclusion of special pages
3301
3302         function getUseTeX()                        { return $this->mUseTeX; }
3303         function getUseDynamicDates()               { return $this->mUseDynamicDates; }
3304         function getInterwikiMagic()                { return $this->mInterwikiMagic; }
3305         function getAllowExternalImages()           { return $this->mAllowExternalImages; }
3306         function getSkin()                          { return $this->mSkin; }
3307         function getDateFormat()                    { return $this->mDateFormat; }
3308         function getEditSection()                   { return $this->mEditSection; }
3309         function getNumberHeadings()                { return $this->mNumberHeadings; }
3310         function getAllowSpecialInclusion()         { return $this->mAllowSpecialInclusion; }
3311
3312
3313         function setUseTeX( $x )                    { return wfSetVar( $this->mUseTeX, $x ); }
3314         function setUseDynamicDates( $x )           { return wfSetVar( $this->mUseDynamicDates, $x ); }
3315         function setInterwikiMagic( $x )            { return wfSetVar( $this->mInterwikiMagic, $x ); }
3316         function setAllowExternalImages( $x )       { return wfSetVar( $this->mAllowExternalImages, $x ); }
3317         function setDateFormat( $x )                { return wfSetVar( $this->mDateFormat, $x ); }
3318         function setEditSection( $x )               { return wfSetVar( $this->mEditSection, $x ); }
3319         function setNumberHeadings( $x )            { return wfSetVar( $this->mNumberHeadings, $x ); }
3320         function setAllowSpecialInclusion( $x )     { return wfSetVar( $this->mAllowSpecialInclusion, $x ); }
3321
3322         function setSkin( &$x ) { $this->mSkin =& $x; }
3323
3324         function ParserOptions() {
3325                 global $wgUser;
3326                 $this->initialiseFromUser( $wgUser );
3327         }
3328
3329         /**
3330          * Get parser options
3331          * @static
3332          */
3333         function newFromUser( &$user ) {
3334                 $popts = new ParserOptions;
3335                 $popts->initialiseFromUser( $user );
3336                 return $popts;
3337         }
3338
3339         /** Get user options */
3340         function initialiseFromUser( &$userInput ) {
3341                 global $wgUseTeX, $wgUseDynamicDates, $wgInterwikiMagic, $wgAllowExternalImages,
3342                        $wgAllowSpecialInclusion;
3343                 $fname = 'ParserOptions::initialiseFromUser';
3344                 wfProfileIn( $fname );
3345                 if ( !$userInput ) {
3346                         $user = new User;
3347                         $user->setLoaded( true );
3348                 } else {
3349                         $user =& $userInput;
3350                 }
3351
3352                 $this->mUseTeX = $wgUseTeX;
3353                 $this->mUseDynamicDates = $wgUseDynamicDates;
3354                 $this->mInterwikiMagic = $wgInterwikiMagic;
3355                 $this->mAllowExternalImages = $wgAllowExternalImages;
3356                 wfProfileIn( $fname.'-skin' );
3357                 $this->mSkin =& $user->getSkin();
3358                 wfProfileOut( $fname.'-skin' );
3359                 $this->mDateFormat = $user->getOption( 'date' );
3360                 $this->mEditSection = $user->getOption( 'editsection' );
3361                 $this->mNumberHeadings = $user->getOption( 'numberheadings' );
3362                 $this->mAllowSpecialInclusion = $wgAllowSpecialInclusion;
3363                 wfProfileOut( $fname );
3364         }
3365 }
3366
3367 /**
3368  * Callback function used by Parser::replaceLinkHolders()
3369  * to substitute link placeholders.
3370  */
3371 function &wfOutputReplaceMatches( $matches ) {
3372         global $wgOutputReplace;
3373         return $wgOutputReplace[$matches[1]];
3374 }
3375
3376 /**
3377  * Return the total number of articles
3378  */
3379 function wfNumberOfArticles() {
3380         global $wgNumberOfArticles;
3381
3382         wfLoadSiteStats();
3383         return $wgNumberOfArticles;
3384 }
3385
3386 /**
3387  * Return the number of files
3388  */
3389 function wfNumberOfFiles() {
3390         $fname = 'Parser::wfNumberOfFiles';
3391
3392         wfProfileIn( $fname );
3393         $dbr =& wfGetDB( DB_SLAVE );
3394         $res = $dbr->selectField('image', 'COUNT(*)', array(), $fname );
3395         wfProfileOut( $fname );
3396
3397         return $res;
3398 }
3399
3400 /**
3401  * Get various statistics from the database
3402  * @private
3403  */
3404 function wfLoadSiteStats() {
3405         global $wgNumberOfArticles, $wgTotalViews, $wgTotalEdits;
3406         $fname = 'wfLoadSiteStats';
3407
3408         if ( -1 != $wgNumberOfArticles ) return;
3409         $dbr =& wfGetDB( DB_SLAVE );
3410         $s = $dbr->selectRow( 'site_stats',
3411                 array( 'ss_total_views', 'ss_total_edits', 'ss_good_articles' ),
3412                 array( 'ss_row_id' => 1 ), $fname
3413         );
3414
3415         if ( $s === false ) {
3416                 return;
3417         } else {
3418                 $wgTotalViews = $s->ss_total_views;
3419                 $wgTotalEdits = $s->ss_total_edits;
3420                 $wgNumberOfArticles = $s->ss_good_articles;
3421         }
3422 }
3423
3424 /**
3425  * Escape html tags
3426  * Basicly replacing " > and < with HTML entities ( &quot;, &gt;, &lt;)
3427  *
3428  * @param string $in Text that might contain HTML tags
3429  * @return string Escaped string
3430  */
3431 function wfEscapeHTMLTagsOnly( $in ) {
3432         return str_replace(
3433                 array( '"', '>', '<' ),
3434                 array( '&quot;', '&gt;', '&lt;' ),
3435                 $in );
3436 }
3437
3438 ?>