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 =& $this->mStripState;
 186
 187                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$x ) );
 188                 $text = $this->strip( $text, $x );
 189                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$x ) );
 190
 191                 $text = $this->internalParse( $text );
 192
 193                 $text = $this->unstrip( $text, $this->mStripState );
 194
 195                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 196                 $fixtags = array(
 197                         # french spaces, last one Guillemet-left
 198                         # only if there is something before the space
 199                         '/(.) (?=\\?|:|;|!|\\302\\273)/' => '\\1&nbsp;\\2',
 200                         # french spaces, Guillemet-right
 201                         '/(\\302\\253) /' => '\\1&nbsp;',
 202                         '/<center *>/i' => '<div class="center">',
 203                         '/<\\/center *>/i' => '</div>',
 204                 );
 205                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 206
 207                 # only once and last
 208                 $text = $this->doBlockLevels( $text, $linestart );
 209
 210                 $this->replaceLinkHolders( $text );
 211
 212                 # the position of the convert() call should not be changed. it
 213                 # assumes that the links are all replaces and the only thing left
 214                 # is the <nowiki> mark.
 215                 $text = $wgContLang->convert($text);
 216                 $this->mOutput->setTitleText($wgContLang->getParsedTitle());
 217
 218                 $text = $this->unstripNoWiki( $text, $this->mStripState );
 219
 220                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 221
 222                 $text = Sanitizer::normalizeCharReferences( $text );
 223                 global $wgUseTidy;
 224                 if ($wgUseTidy) {
 225                         $text = Parser::tidy($text);
 226                 }
 227
 228                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 229
 230                 $this->mOutput->setText( $text );
 231                 wfProfileOut( $fname );
 232                 return $this->mOutput;
 233         }
 234
 235         /**
 236          * Get a random string
 237          *
 238          * @access private
 239          * @static
 240          */
 241         function getRandomString() {
 242                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 243         }
 244
 245         /**
 246          * Replaces all occurrences of <$tag>content</$tag> in the text
 247          * with a random marker and returns the new text. the output parameter
 248          * $content will be an associative array filled with data on the form
 249          * $unique_marker => content.
 250          *
 251          * If $content is already set, the additional entries will be appended
 252          * If $tag is set to STRIP_COMMENTS, the function will extract
 253          * <!-- HTML comments -->
 254          *
 255          * @access private
 256          * @static
 257          */
 258         function extractTagsAndParams($tag, $text, &$content, &$tags, &$params, $uniq_prefix = ''){
 259                 $rnd = $uniq_prefix . '-' . $tag . Parser::getRandomString();
 260                 if ( !$content ) {
 261                         $content = array( );
 262                 }
 263                 $n = 1;
 264                 $stripped = '';
 265
 266                 if ( !$tags ) {
 267                         $tags = array( );
 268                 }
 269
 270                 if ( !$params ) {
 271                         $params = array( );
 272                 }
 273
 274                 if( $tag == STRIP_COMMENTS ) {
 275                         $start = '/<!--()/';
 276                         $end   = '/-->/';
 277                 } else {
 278                         $start = "/<$tag(\\s+[^>]*|\\s*)>/i";
 279                         $end   = "/<\\/$tag\\s*>/i";
 280                 }
 281
 282                 while ( '' != $text ) {
 283                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 284                         $stripped .= $p[0];
 285                         if( count( $p ) < 3 ) {
 286                                 break;
 287                         }
 288                         $attributes = $p[1];
 289                         $inside     = $p[2];
 290
 291                         $marker = $rnd . sprintf('%08X', $n++);
 292                         $stripped .= $marker;
 293
 294                         $tags[$marker] = "<$tag$attributes>";
 295                         $params[$marker] = Sanitizer::decodeTagAttributes( $attributes );
 296
 297                         $q = preg_split( $end, $inside, 2 );
 298                         $content[$marker] = $q[0];
 299                         if( count( $q ) < 1 ) {
 300                                 # No end tag -- let it run out to the end of the text.
 301                                 break;
 302                         } else {
 303                                 $text = $q[1];
 304                         }
 305                 }
 306                 return $stripped;
 307         }
 308
 309         /**
 310          * Wrapper function for extractTagsAndParams
 311          * for cases where $tags and $params isn't needed
 312          * i.e. where tags will never have params, like <nowiki>
 313          *
 314          * @access private
 315          * @static
 316          */
 317         function extractTags( $tag, $text, &$content, $uniq_prefix = '' ) {
 318                 $dummy_tags = array();
 319                 $dummy_params = array();
 320
 321                 return Parser::extractTagsAndParams( $tag, $text, $content,
 322                         $dummy_tags, $dummy_params, $uniq_prefix );
 323         }
 324
 325         /**
 326          * Strips and renders nowiki, pre, math, hiero
 327          * If $render is set, performs necessary rendering operations on plugins
 328          * Returns the text, and fills an array with data needed in unstrip()
 329          * If the $state is already a valid strip state, it adds to the state
 330          *
 331          * @param bool $stripcomments when set, HTML comments <!-- like this -->
 332          *  will be stripped in addition to other tags. This is important
 333          *  for section editing, where these comments cause confusion when
 334          *  counting the sections in the wikisource
 335          *
 336          * @access private
 337          */
 338         function strip( $text, &$state, $stripcomments = false ) {
 339                 $render = ($this->mOutputType == OT_HTML);
 340                 $html_content = array();
 341                 $nowiki_content = array();
 342                 $math_content = array();
 343                 $pre_content = array();
 344                 $comment_content = array();
 345                 $ext_content = array();
 346                 $ext_tags = array();
 347                 $ext_params = array();
 348                 $gallery_content = array();
 349
 350                 # Replace any instances of the placeholders
 351                 $uniq_prefix = UNIQ_PREFIX;
 352                 #$text = str_replace( $uniq_prefix, wfHtmlEscapeFirst( $uniq_prefix ), $text );
 353
 354                 # html
 355                 global $wgRawHtml;
 356                 if( $wgRawHtml ) {
 357                         $text = Parser::extractTags('html', $text, $html_content, $uniq_prefix);
 358                         foreach( $html_content as $marker => $content ) {
 359                                 if ($render ) {
 360                                         # Raw and unchecked for validity.
 361                                         $html_content[$marker] = $content;
 362                                 } else {
 363                                         $html_content[$marker] = '<html>'.$content.'</html>';
 364                                 }
 365                         }
 366                 }
 367
 368                 # nowiki
 369                 $text = Parser::extractTags('nowiki', $text, $nowiki_content, $uniq_prefix);
 370                 foreach( $nowiki_content as $marker => $content ) {
 371                         if( $render ){
 372                                 $nowiki_content[$marker] = wfEscapeHTMLTagsOnly( $content );
 373                         } else {
 374                                 $nowiki_content[$marker] = '<nowiki>'.$content.'</nowiki>';
 375                         }
 376                 }
 377
 378                 # math
 379                 $text = Parser::extractTags('math', $text, $math_content, $uniq_prefix);
 380                 foreach( $math_content as $marker => $content ){
 381                         if( $render ) {
 382                                 if( $this->mOptions->getUseTeX() ) {
 383                                         $math_content[$marker] = renderMath( $content );
 384                                 } else {
 385                                         $math_content[$marker] = '&lt;math&gt;'.$content.'&lt;math&gt;';
 386                                 }
 387                         } else {
 388                                 $math_content[$marker] = '<math>'.$content.'</math>';
 389                         }
 390                 }
 391
 392                 # pre
 393                 $text = Parser::extractTags('pre', $text, $pre_content, $uniq_prefix);
 394                 foreach( $pre_content as $marker => $content ){
 395                         if( $render ){
 396                                 $pre_content[$marker] = '<pre>' . wfEscapeHTMLTagsOnly( $content ) . '</pre>';
 397                         } else {
 398                                 $pre_content[$marker] = '<pre>'.$content.'</pre>';
 399                         }
 400                 }
 401
 402                 # gallery
 403                 $text = Parser::extractTags('gallery', $text, $gallery_content, $uniq_prefix);
 404                 foreach( $gallery_content as $marker => $content ) {
 405                         require_once( 'ImageGallery.php' );
 406                         if ( $render ) {
 407                                 $gallery_content[$marker] = Parser::renderImageGallery( $content );
 408                         } else {
 409                                 $gallery_content[$marker] = '<gallery>'.$content.'</gallery>';
 410                         }
 411                 }
 412
 413                 # Comments
 414                 if($stripcomments) {
 415                         $text = Parser::extractTags(STRIP_COMMENTS, $text, $comment_content, $uniq_prefix);
 416                         foreach( $comment_content as $marker => $content ){
 417                                 $comment_content[$marker] = '<!--'.$content.'-->';
 418                         }
 419                 }
 420
 421                 # Extensions
 422                 foreach ( $this->mTagHooks as $tag => $callback ) {
 423                         $ext_content[$tag] = array();
 424                         $text = Parser::extractTagsAndParams( $tag, $text, $ext_content[$tag],
 425                                 $ext_tags[$tag], $ext_params[$tag], $uniq_prefix );
 426                         foreach( $ext_content[$tag] as $marker => $content ) {
 427                                 $full_tag = $ext_tags[$tag][$marker];
 428                                 $params = $ext_params[$tag][$marker];
 429                                 if ( $render ) {
 430                                         $ext_content[$tag][$marker] = $callback( $content, $params );
 431                                 } else {
 432                                         $ext_content[$tag][$marker] = "$full_tag$content</$tag>";
 433                                 }
 434                         }
 435                 }
 436
 437                 # Merge state with the pre-existing state, if there is one
 438                 if ( $state ) {
 439                         $state['html'] = $state['html'] + $html_content;
 440                         $state['nowiki'] = $state['nowiki'] + $nowiki_content;
 441                         $state['math'] = $state['math'] + $math_content;
 442                         $state['pre'] = $state['pre'] + $pre_content;
 443                         $state['comment'] = $state['comment'] + $comment_content;
 444                         $state['gallery'] = $state['gallery'] + $gallery_content;
 445
 446                         foreach( $ext_content as $tag => $array ) {
 447                                 if ( array_key_exists( $tag, $state ) ) {
 448                                         $state[$tag] = $state[$tag] + $array;
 449                                 }
 450                         }
 451                 } else {
 452                         $state = array(
 453                           'html' => $html_content,
 454                           'nowiki' => $nowiki_content,
 455                           'math' => $math_content,
 456                           'pre' => $pre_content,
 457                           'comment' => $comment_content,
 458                           'gallery' => $gallery_content,
 459                         ) + $ext_content;
 460                 }
 461                 return $text;
 462         }
 463
 464         /**
 465          * restores pre, math, and hiero removed by strip()
 466          *
 467          * always call unstripNoWiki() after this one
 468          * @access private
 469          */
 470         function unstrip( $text, &$state ) {
 471                 # Must expand in reverse order, otherwise nested tags will be corrupted
 472                 $contentDict = end( $state );
 473                 for ( $contentDict = end( $state ); $contentDict !== false; $contentDict = prev( $state ) ) {
 474                         if( key($state) != 'nowiki' && key($state) != 'html') {
 475                                 for ( $content = end( $contentDict ); $content !== false; $content = prev( $contentDict ) ) {
 476                                         $text = str_replace( key( $contentDict ), $content, $text );
 477                                 }
 478                         }
 479                 }
 480
 481                 return $text;
 482         }
 483
 484         /**
 485          * always call this after unstrip() to preserve the order
 486          *
 487          * @access private
 488          */
 489         function unstripNoWiki( $text, &$state ) {
 490                 # Must expand in reverse order, otherwise nested tags will be corrupted
 491                 for ( $content = end($state['nowiki']); $content !== false; $content = prev( $state['nowiki'] ) ) {
 492                         $text = str_replace( key( $state['nowiki'] ), $content, $text );
 493                 }
 494
 495                 global $wgRawHtml;
 496                 if ($wgRawHtml) {
 497                         for ( $content = end($state['html']); $content !== false; $content = prev( $state['html'] ) ) {
 498                                 $text = str_replace( key( $state['html'] ), $content, $text );
 499                         }
 500                 }
 501
 502                 return $text;
 503         }
 504
 505         /**
 506          * Add an item to the strip state
 507          * Returns the unique tag which must be inserted into the stripped text
 508          * The tag will be replaced with the original text in unstrip()
 509          *
 510          * @access private
 511          */
 512         function insertStripItem( $text, &$state ) {
 513                 $rnd = UNIQ_PREFIX . '-item' . Parser::getRandomString();
 514                 if ( !$state ) {
 515                         $state = array(
 516                           'html' => array(),
 517                           'nowiki' => array(),
 518                           'math' => array(),
 519                           'pre' => array(),
 520                           'comment' => array(),
 521                           'gallery' => array(),
 522                         );
 523                 }
 524                 $state['item'][$rnd] = $text;
 525                 return $rnd;
 526         }
 527
 528         /**
 529          * Interface with html tidy, used if $wgUseTidy = true.
 530          * If tidy isn't able to correct the markup, the original will be
 531          * returned in all its glory with a warning comment appended.
 532          *
 533          * Either the external tidy program or the in-process tidy extension
 534          * will be used depending on availability. Override the default
 535          * $wgTidyInternal setting to disable the internal if it's not working.
 536          *
 537          * @param string $text Hideous HTML input
 538          * @return string Corrected HTML output
 539          * @access public
 540          * @static
 541          */
 542         function tidy( $text ) {
 543                 global $wgTidyInternal;
 544                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 545 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 546 '<head><title>test</title></head><body>'.$text.'</body></html>';
 547                 if( $wgTidyInternal ) {
 548                         $correctedtext = Parser::internalTidy( $wrappedtext );
 549                 } else {
 550                         $correctedtext = Parser::externalTidy( $wrappedtext );
 551                 }
 552                 if( is_null( $correctedtext ) ) {
 553                         wfDebug( "Tidy error detected!\n" );
 554                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 555                 }
 556                 return $correctedtext;
 557         }
 558
 559         /**
 560          * Spawn an external HTML tidy process and get corrected markup back from it.
 561          *
 562          * @access private
 563          * @static
 564          */
 565         function externalTidy( $text ) {
 566                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 567                 $fname = 'Parser::externalTidy';
 568                 wfProfileIn( $fname );
 569
 570                 $cleansource = '';
 571                 $opts = ' -utf8';
 572
 573                 $descriptorspec = array(
 574                         0 => array('pipe', 'r'),
 575                         1 => array('pipe', 'w'),
 576                         2 => array('file', '/dev/null', 'a')
 577                 );
 578                 $pipes = array();
 579                 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 580                 if (is_resource($process)) {
 581                         fwrite($pipes[0], $text);
 582                         fclose($pipes[0]);
 583                         while (!feof($pipes[1])) {
 584                                 $cleansource .= fgets($pipes[1], 1024);
 585                         }
 586                         fclose($pipes[1]);
 587                         proc_close($process);
 588                 }
 589
 590                 wfProfileOut( $fname );
 591
 592                 if( $cleansource == '' && $text != '') {
 593                         // Some kind of error happened, so we couldn't get the corrected text.
 594                         // Just give up; we'll use the source text and append a warning.
 595                         return null;
 596                 } else {
 597                         return $cleansource;
 598                 }
 599         }
 600
 601         /**
 602          * Use the HTML tidy PECL extension to use the tidy library in-process,
 603          * saving the overhead of spawning a new process. Currently written to
 604          * the PHP 4.3.x version of the extension, may not work on PHP 5.
 605          *
 606          * 'pear install tidy' should be able to compile the extension module.
 607          *
 608          * @access private
 609          * @static
 610          */
 611         function internalTidy( $text ) {
 612                 global $wgTidyConf;
 613                 $fname = 'Parser::internalTidy';
 614                 wfProfileIn( $fname );
 615
 616                 tidy_load_config( $wgTidyConf );
 617                 tidy_set_encoding( 'utf8' );
 618                 tidy_parse_string( $text );
 619                 tidy_clean_repair();
 620                 if( tidy_get_status() == 2 ) {
 621                         // 2 is magic number for fatal error
 622                         // http://www.php.net/manual/en/function.tidy-get-status.php
 623                         $cleansource = null;
 624                 } else {
 625                         $cleansource = tidy_get_output();
 626                 }
 627                 wfProfileOut( $fname );
 628                 return $cleansource;
 629         }
 630
 631         /**
 632          * parse the wiki syntax used to render tables
 633          *
 634          * @access private
 635          */
 636         function doTableStuff ( $t ) {
 637                 $fname = 'Parser::doTableStuff';
 638                 wfProfileIn( $fname );
 639
 640                 $t = explode ( "\n" , $t ) ;
 641                 $td = array () ; # Is currently a td tag open?
 642                 $ltd = array () ; # Was it TD or TH?
 643                 $tr = array () ; # Is currently a tr tag open?
 644                 $ltr = array () ; # tr attributes
 645                 $indent_level = 0; # indent level of the table
 646                 foreach ( $t AS $k => $x )
 647                 {
 648                         $x = trim ( $x ) ;
 649                         $fc = substr ( $x , 0 , 1 ) ;
 650                         if ( preg_match( '/^(:*)\{\|(.*)$/', $x, $matches ) ) {
 651                                 $indent_level = strlen( $matches[1] );
 652                                 $t[$k] = str_repeat( '<dl><dd>', $indent_level ) .
 653                                         '<table' . Sanitizer::fixTagAttributes ( $matches[2], 'table' ) . '>' ;
 654                                 array_push ( $td , false ) ;
 655                                 array_push ( $ltd , '' ) ;
 656                                 array_push ( $tr , false ) ;
 657                                 array_push ( $ltr , '' ) ;
 658                         }
 659                         else if ( count ( $td ) == 0 ) { } # Don't do any of the following
 660                         else if ( '|}' == substr ( $x , 0 , 2 ) ) {
 661                                 $z = "</table>" . substr ( $x , 2);
 662                                 $l = array_pop ( $ltd ) ;
 663                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 664                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 665                                 array_pop ( $ltr ) ;
 666                                 $t[$k] = $z . str_repeat( '</dd></dl>', $indent_level );
 667                         }
 668                         else if ( '|-' == substr ( $x , 0 , 2 ) ) { # Allows for |---------------
 669                                 $x = substr ( $x , 1 ) ;
 670                                 while ( $x != '' && substr ( $x , 0 , 1 ) == '-' ) $x = substr ( $x , 1 ) ;
 671                                 $z = '' ;
 672                                 $l = array_pop ( $ltd ) ;
 673                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 674                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 675                                 array_pop ( $ltr ) ;
 676                                 $t[$k] = $z ;
 677                                 array_push ( $tr , false ) ;
 678                                 array_push ( $td , false ) ;
 679                                 array_push ( $ltd , '' ) ;
 680                                 array_push ( $ltr , Sanitizer::fixTagAttributes ( $x, 'tr' ) ) ;
 681                         }
 682                         else if ( '|' == $fc || '!' == $fc || '|+' == substr ( $x , 0 , 2 ) ) { # Caption
 683                                 # $x is a table row
 684                                 if ( '|+' == substr ( $x , 0 , 2 ) ) {
 685                                         $fc = '+' ;
 686                                         $x = substr ( $x , 1 ) ;
 687                                 }
 688                                 $after = substr ( $x , 1 ) ;
 689                                 if ( $fc == '!' ) $after = str_replace ( '!!' , '||' , $after ) ;
 690                                 $after = explode ( '||' , $after ) ;
 691                                 $t[$k] = '' ;
 692
 693                                 # Loop through each table cell
 694                                 foreach ( $after AS $theline )
 695                                 {
 696                                         $z = '' ;
 697                                         if ( $fc != '+' )
 698                                         {
 699                                                 $tra = array_pop ( $ltr ) ;
 700                                                 if ( !array_pop ( $tr ) ) $z = '<tr'.$tra.">\n" ;
 701                                                 array_push ( $tr , true ) ;
 702                                                 array_push ( $ltr , '' ) ;
 703                                         }
 704
 705                                         $l = array_pop ( $ltd ) ;
 706                                         if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 707                                         if ( $fc == '|' ) $l = 'td' ;
 708                                         else if ( $fc == '!' ) $l = 'th' ;
 709                                         else if ( $fc == '+' ) $l = 'caption' ;
 710                                         else $l = '' ;
 711                                         array_push ( $ltd , $l ) ;
 712
 713                                         # Cell parameters
 714                                         $y = explode ( '|' , $theline , 2 ) ;
 715                                         # Note that a '|' inside an invalid link should not
 716                                         # be mistaken as delimiting cell parameters
 717                                         if ( strpos( $y[0], '[[' ) !== false ) {
 718                                                 $y = array ($theline);
 719                                         }
 720                                         if ( count ( $y ) == 1 )
 721                                                 $y = "{$z}<{$l}>{$y[0]}" ;
 722                                         else $y = $y = "{$z}<{$l}".Sanitizer::fixTagAttributes($y[0], $l).">{$y[1]}" ;
 723                                         $t[$k] .= $y ;
 724                                         array_push ( $td , true ) ;
 725                                 }
 726                         }
 727                 }
 728
 729                 # Closing open td, tr && table
 730                 while ( count ( $td ) > 0 )
 731                 {
 732                         if ( array_pop ( $td ) ) $t[] = '</td>' ;
 733                         if ( array_pop ( $tr ) ) $t[] = '</tr>' ;
 734                         $t[] = '</table>' ;
 735                 }
 736
 737                 $t = implode ( "\n" , $t ) ;
 738                 wfProfileOut( $fname );
 739                 return $t ;
 740         }
 741
 742         /**
 743          * Helper function for parse() that transforms wiki markup into
 744          * HTML. Only called for $mOutputType == OT_HTML.
 745          *
 746          * @access private
 747          */
 748         function internalParse( $text ) {
 749                 global $wgContLang;
 750                 $args = array();
 751                 $isMain = true;
 752                 $fname = 'Parser::internalParse';
 753                 wfProfileIn( $fname );
 754
 755                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'replaceVariables' ) );
 756                 $text = $this->replaceVariables( $text, $args );
 757
 758                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 759
 760                 $text = $this->doHeadings( $text );
 761                 if($this->mOptions->getUseDynamicDates()) {
 762                         $df =& DateFormatter::getInstance();
 763                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 764                 }
 765                 $text = $this->doAllQuotes( $text );
 766                 $text = $this->replaceInternalLinks( $text );
 767                 $text = $this->replaceExternalLinks( $text );
 768
 769                 # replaceInternalLinks may sometimes leave behind
 770                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 771                 $text = str_replace("http-noparse://","http://",$text);
 772
 773                 $text = $this->doMagicLinks( $text );
 774                 $text = $this->doTableStuff( $text );
 775                 $text = $this->formatHeadings( $text, $isMain );
 776
 777                 wfProfileOut( $fname );
 778                 return $text;
 779         }
 780
 781         /**
 782          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 783          * magic external links.
 784          *
 785          * @access private
 786          */
 787         function &doMagicLinks( &$text ) {
 788                 $text = $this->magicISBN( $text );
 789                 $text = $this->magicRFC( $text, 'RFC ', 'rfcurl' );
 790                 $text = $this->magicRFC( $text, 'PMID ', 'pubmedurl' );
 791                 return $text;
 792         }
 793
 794         /**
 795          * Parse ^^ tokens and return html
 796          *
 797          * @access private
 798          */
 799         function doExponent( $text ) {
 800                 $fname = 'Parser::doExponent';
 801                 wfProfileIn( $fname );
 802                 $text = preg_replace('/\^\^(.*)\^\^/','<small><sup>\\1</sup></small>', $text);
 803                 wfProfileOut( $fname );
 804                 return $text;
 805         }
 806
 807         /**
 808          * Parse headers and return html
 809          *
 810          * @access private
 811          */
 812         function doHeadings( $text ) {
 813                 $fname = 'Parser::doHeadings';
 814                 wfProfileIn( $fname );
 815                 for ( $i = 6; $i >= 1; --$i ) {
 816                         $h = substr( '======', 0, $i );
 817                         $text = preg_replace( "/^{$h}(.+){$h}(\\s|$)/m",
 818                           "<h{$i}>\\1</h{$i}>\\2", $text );
 819                 }
 820                 wfProfileOut( $fname );
 821                 return $text;
 822         }
 823
 824         /**
 825          * Replace single quotes with HTML markup
 826          * @access private
 827          * @return string the altered text
 828          */
 829         function doAllQuotes( $text ) {
 830                 $fname = 'Parser::doAllQuotes';
 831                 wfProfileIn( $fname );
 832                 $outtext = '';
 833                 $lines = explode( "\n", $text );
 834                 foreach ( $lines as $line ) {
 835                         $outtext .= $this->doQuotes ( $line ) . "\n";
 836                 }
 837                 $outtext = substr($outtext, 0,-1);
 838                 wfProfileOut( $fname );
 839                 return $outtext;
 840         }
 841
 842         /**
 843          * Helper function for doAllQuotes()
 844          * @access private
 845          */
 846         function doQuotes( $text ) {
 847                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
 848                 if ( count( $arr ) == 1 )
 849                         return $text;
 850                 else
 851                 {
 852                         # First, do some preliminary work. This may shift some apostrophes from
 853                         # being mark-up to being text. It also counts the number of occurrences
 854                         # of bold and italics mark-ups.
 855                         $i = 0;
 856                         $numbold = 0;
 857                         $numitalics = 0;
 858                         foreach ( $arr as $r )
 859                         {
 860                                 if ( ( $i % 2 ) == 1 )
 861                                 {
 862                                         # If there are ever four apostrophes, assume the first is supposed to
 863                                         # be text, and the remaining three constitute mark-up for bold text.
 864                                         if ( strlen( $arr[$i] ) == 4 )
 865                                         {
 866                                                 $arr[$i-1] .= "'";
 867                                                 $arr[$i] = "'''";
 868                                         }
 869                                         # If there are more than 5 apostrophes in a row, assume they're all
 870                                         # text except for the last 5.
 871                                         else if ( strlen( $arr[$i] ) > 5 )
 872                                         {
 873                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
 874                                                 $arr[$i] = "'''''";
 875                                         }
 876                                         # Count the number of occurrences of bold and italics mark-ups.
 877                                         # We are not counting sequences of five apostrophes.
 878                                         if ( strlen( $arr[$i] ) == 2 ) $numitalics++;  else
 879                                         if ( strlen( $arr[$i] ) == 3 ) $numbold++;     else
 880                                         if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
 881                                 }
 882                                 $i++;
 883                         }
 884
 885                         # If there is an odd number of both bold and italics, it is likely
 886                         # that one of the bold ones was meant to be an apostrophe followed
 887                         # by italics. Which one we cannot know for certain, but it is more
 888                         # likely to be one that has a single-letter word before it.
 889                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
 890                         {
 891                                 $i = 0;
 892                                 $firstsingleletterword = -1;
 893                                 $firstmultiletterword = -1;
 894                                 $firstspace = -1;
 895                                 foreach ( $arr as $r )
 896                                 {
 897                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
 898                                         {
 899                                                 $x1 = substr ($arr[$i-1], -1);
 900                                                 $x2 = substr ($arr[$i-1], -2, 1);
 901                                                 if ($x1 == ' ') {
 902                                                         if ($firstspace == -1) $firstspace = $i;
 903                                                 } else if ($x2 == ' ') {
 904                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
 905                                                 } else {
 906                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
 907                                                 }
 908                                         }
 909                                         $i++;
 910                                 }
 911
 912                                 # If there is a single-letter word, use it!
 913                                 if ($firstsingleletterword > -1)
 914                                 {
 915                                         $arr [ $firstsingleletterword ] = "''";
 916                                         $arr [ $firstsingleletterword-1 ] .= "'";
 917                                 }
 918                                 # If not, but there's a multi-letter word, use that one.
 919                                 else if ($firstmultiletterword > -1)
 920                                 {
 921                                         $arr [ $firstmultiletterword ] = "''";
 922                                         $arr [ $firstmultiletterword-1 ] .= "'";
 923                                 }
 924                                 # ... otherwise use the first one that has neither.
 925                                 # (notice that it is possible for all three to be -1 if, for example,
 926                                 # there is only one pentuple-apostrophe in the line)
 927                                 else if ($firstspace > -1)
 928                                 {
 929                                         $arr [ $firstspace ] = "''";
 930                                         $arr [ $firstspace-1 ] .= "'";
 931                                 }
 932                         }
 933
 934                         # Now let's actually convert our apostrophic mush to HTML!
 935                         $output = '';
 936                         $buffer = '';
 937                         $state = '';
 938                         $i = 0;
 939                         foreach ($arr as $r)
 940                         {
 941                                 if (($i % 2) == 0)
 942                                 {
 943                                         if ($state == 'both')
 944                                                 $buffer .= $r;
 945                                         else
 946                                                 $output .= $r;
 947                                 }
 948                                 else
 949                                 {
 950                                         if (strlen ($r) == 2)
 951                                         {
 952                                                 if ($state == 'i')
 953                                                 { $output .= '</i>'; $state = ''; }
 954                                                 else if ($state == 'bi')
 955                                                 { $output .= '</i>'; $state = 'b'; }
 956                                                 else if ($state == 'ib')
 957                                                 { $output .= '</b></i><b>'; $state = 'b'; }
 958                                                 else if ($state == 'both')
 959                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
 960                                                 else # $state can be 'b' or ''
 961                                                 { $output .= '<i>'; $state .= 'i'; }
 962                                         }
 963                                         else if (strlen ($r) == 3)
 964                                         {
 965                                                 if ($state == 'b')
 966                                                 { $output .= '</b>'; $state = ''; }
 967                                                 else if ($state == 'bi')
 968                                                 { $output .= '</i></b><i>'; $state = 'i'; }
 969                                                 else if ($state == 'ib')
 970                                                 { $output .= '</b>'; $state = 'i'; }
 971                                                 else if ($state == 'both')
 972                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
 973                                                 else # $state can be 'i' or ''
 974                                                 { $output .= '<b>'; $state .= 'b'; }
 975                                         }
 976                                         else if (strlen ($r) == 5)
 977                                         {
 978                                                 if ($state == 'b')
 979                                                 { $output .= '</b><i>'; $state = 'i'; }
 980                                                 else if ($state == 'i')
 981                                                 { $output .= '</i><b>'; $state = 'b'; }
 982                                                 else if ($state == 'bi')
 983                                                 { $output .= '</i></b>'; $state = ''; }
 984                                                 else if ($state == 'ib')
 985                                                 { $output .= '</b></i>'; $state = ''; }
 986                                                 else if ($state == 'both')
 987                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
 988                                                 else # ($state == '')
 989                                                 { $buffer = ''; $state = 'both'; }
 990                                         }
 991                                 }
 992                                 $i++;
 993                         }
 994                         # Now close all remaining tags.  Notice that the order is important.
 995                         if ($state == 'b' || $state == 'ib')
 996                                 $output .= '</b>';
 997                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
 998                                 $output .= '</i>';
 999                         if ($state == 'bi')
1000                                 $output .= '</b>';
1001                         if ($state == 'both')
1002                                 $output .= '<b><i>'.$buffer.'</i></b>';
1003                         return $output;
1004                 }
1005         }
1006
1007         /**
1008          * Replace external links
1009          *
1010          * Note: this is all very hackish and the order of execution matters a lot.
1011          * Make sure to run maintenance/parserTests.php if you change this code.
1012          *
1013          * @access private
1014          */
1015         function replaceExternalLinks( $text ) {
1016                 global $wgContLang;
1017                 $fname = 'Parser::replaceExternalLinks';
1018                 wfProfileIn( $fname );
1019
1020                 $sk =& $this->mOptions->getSkin();
1021
1022                 $bits = preg_split( EXT_LINK_BRACKETED, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1023
1024                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
1025
1026                 $i = 0;
1027                 while ( $i<count( $bits ) ) {
1028                         $url = $bits[$i++];
1029                         $protocol = $bits[$i++];
1030                         $text = $bits[$i++];
1031                         $trail = $bits[$i++];
1032
1033                         # The characters '<' and '>' (which were escaped by
1034                         # removeHTMLtags()) should not be included in
1035                         # URLs, per RFC 2396.
1036                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1037                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1038                                 $url = substr($url, 0, $m2[0][1]);
1039                         }
1040
1041                         # If the link text is an image URL, replace it with an <img> tag
1042                         # This happened by accident in the original parser, but some people used it extensively
1043                         $img = $this->maybeMakeExternalImage( $text );
1044                         if ( $img !== false ) {
1045                                 $text = $img;
1046                         }
1047
1048                         $dtrail = '';
1049
1050                         # Set linktype for CSS - if URL==text, link is essentially free
1051                         $linktype = ($text == $url) ? 'free' : 'text';
1052
1053                         # No link text, e.g. [http://domain.tld/some.link]
1054                         if ( $text == '' ) {
1055                                 # Autonumber if allowed
1056                                 if ( strpos( HTTP_PROTOCOLS, $protocol ) !== false ) {
1057                                         $text = '[' . ++$this->mAutonumber . ']';
1058                                         $linktype = 'autonumber';
1059                                 } else {
1060                                         # Otherwise just use the URL
1061                                         $text = htmlspecialchars( $url );
1062                                         $linktype = 'free';
1063                                 }
1064                         } else {
1065                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1066                                 # Check for trail
1067                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1068                         }
1069
1070                         $text = $wgContLang->markNoConversion($text);
1071
1072                         # Replace &amp; from obsolete syntax with &.
1073                         # All HTML entities will be escaped by makeExternalLink()
1074                         # or maybeMakeExternalImage()
1075                         $url = str_replace( '&amp;', '&', $url );
1076
1077                         # Process the trail (i.e. everything after this link up until start of the next link),
1078                         # replacing any non-bracketed links
1079                         $trail = $this->replaceFreeExternalLinks( $trail );
1080
1081
1082                         # Use the encoded URL
1083                         # This means that users can paste URLs directly into the text
1084                         # Funny characters like &ouml; aren't valid in URLs anyway
1085                         # This was changed in August 2004
1086                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype ) . $dtrail . $trail;
1087                 }
1088
1089                 wfProfileOut( $fname );
1090                 return $s;
1091         }
1092
1093         /**
1094          * Replace anything that looks like a URL with a link
1095          * @access private
1096          */
1097         function replaceFreeExternalLinks( $text ) {
1098                 global $wgContLang;
1099                 $fname = 'Parser::replaceFreeExternalLinks';
1100                 wfProfileIn( $fname );
1101
1102                 $bits = preg_split( '/(\b(?:'.URL_PROTOCOLS.'):)/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1103                 $s = array_shift( $bits );
1104                 $i = 0;
1105
1106                 $sk =& $this->mOptions->getSkin();
1107
1108                 while ( $i < count( $bits ) ){
1109                         $protocol = $bits[$i++];
1110                         $remainder = $bits[$i++];
1111
1112                         if ( preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1113                                 # Found some characters after the protocol that look promising
1114                                 $url = $protocol . $m[1];
1115                                 $trail = $m[2];
1116
1117                                 # The characters '<' and '>' (which were escaped by
1118                                 # removeHTMLtags()) should not be included in
1119                                 # URLs, per RFC 2396.
1120                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1121                                         $trail = substr($url, $m2[0][1]) . $trail;
1122                                         $url = substr($url, 0, $m2[0][1]);
1123                                 }
1124
1125                                 # Move trailing punctuation to $trail
1126                                 $sep = ',;\.:!?';
1127                                 # If there is no left bracket, then consider right brackets fair game too
1128                                 if ( strpos( $url, '(' ) === false ) {
1129                                         $sep .= ')';
1130                                 }
1131
1132                                 $numSepChars = strspn( strrev( $url ), $sep );
1133                                 if ( $numSepChars ) {
1134                                         $trail = substr( $url, -$numSepChars ) . $trail;
1135                                         $url = substr( $url, 0, -$numSepChars );
1136                                 }
1137
1138                                 # Replace &amp; from obsolete syntax with &.
1139                                 # All HTML entities will be escaped by makeExternalLink()
1140                                 # or maybeMakeExternalImage()
1141                                 $url = str_replace( '&amp;', '&', $url );
1142
1143                                 # Is this an external image?
1144                                 $text = $this->maybeMakeExternalImage( $url );
1145                                 if ( $text === false ) {
1146                                         # Not an image, make a link
1147                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free' );
1148                                 }
1149                                 $s .= $text . $trail;
1150                         } else {
1151                                 $s .= $protocol . $remainder;
1152                         }
1153                 }
1154                 wfProfileOut();
1155                 return $s;
1156         }
1157
1158         /**
1159          * make an image if it's allowed
1160          * @access private
1161          */
1162         function maybeMakeExternalImage( $url ) {
1163                 $sk =& $this->mOptions->getSkin();
1164                 $text = false;
1165                 if ( $this->mOptions->getAllowExternalImages() ) {
1166                         if ( preg_match( EXT_IMAGE_REGEX, $url ) ) {
1167                                 # Image found
1168                                 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
1169                         }
1170                 }
1171                 return $text;
1172         }
1173
1174         /**
1175          * Process [[ ]] wikilinks
1176          *
1177          * @access private
1178          */
1179         function replaceInternalLinks( $s ) {
1180                 global $wgContLang, $wgLinkCache;
1181                 static $fname = 'Parser::replaceInternalLinks' ;
1182
1183                 wfProfileIn( $fname );
1184
1185                 wfProfileIn( $fname.'-setup' );
1186                 static $tc = FALSE;
1187                 # the % is needed to support urlencoded titles as well
1188                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1189
1190                 $sk =& $this->mOptions->getSkin();
1191
1192                 #split the entire text string on occurences of [[
1193                 $a = explode( '[[', ' ' . $s );
1194                 #get the first element (all text up to first [[), and remove the space we added
1195                 $s = array_shift( $a );
1196                 $s = substr( $s, 1 );
1197
1198                 # Match a link having the form [[namespace:link|alternate]]trail
1199                 static $e1 = FALSE;
1200                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1201                 # Match cases where there is no "]]", which might still be images
1202                 static $e1_img = FALSE;
1203                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1204                 # Match the end of a line for a word that's not followed by whitespace,
1205                 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1206                 static $e2 = '/^(.*?)([a-zA-Z\x80-\xff]+)$/sD';
1207
1208                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1209
1210                 if( is_null( $this->mTitle ) ) {
1211                         wfDebugDieBacktrace( 'nooo' );
1212                 }
1213                 $nottalk = !$this->mTitle->isTalkPage();
1214
1215                 if ( $useLinkPrefixExtension ) {
1216                         if ( preg_match( $e2, $s, $m ) ) {
1217                                 $first_prefix = $m[2];
1218                                 $s = $m[1];
1219                         } else {
1220                                 $first_prefix = false;
1221                         }
1222                 } else {
1223                         $prefix = '';
1224                 }
1225
1226                 $selflink = $this->mTitle->getPrefixedText();
1227                 wfProfileOut( $fname.'-setup' );
1228
1229                 $checkVariantLink = sizeof($wgContLang->getVariants())>1;
1230                 $useSubpages = $this->areSubpagesAllowed();
1231
1232                 # Loop for each link
1233                 for ($k = 0; isset( $a[$k] ); $k++) {
1234                         $line = $a[$k];
1235                         if ( $useLinkPrefixExtension ) {
1236                                 wfProfileIn( $fname.'-prefixhandling' );
1237                                 if ( preg_match( $e2, $s, $m ) ) {
1238                                         $prefix = $m[2];
1239                                         $s = $m[1];
1240                                 } else {
1241                                         $prefix='';
1242                                 }
1243                                 # first link
1244                                 if($first_prefix) {
1245                                         $prefix = $first_prefix;
1246                                         $first_prefix = false;
1247                                 }
1248                                 wfProfileOut( $fname.'-prefixhandling' );
1249                         }
1250
1251                         $might_be_img = false;
1252
1253                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1254                                 $text = $m[2];
1255                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1256                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1257                                 # the real problem is with the $e1 regex
1258                                 # See bug 1300.
1259                                 #
1260                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1261                                 # and no image is in sight. See bug 2095.
1262                                 #
1263                                 if( $text !== '' && preg_match( "/^\](.*)/s", $m[3], $n ) ) {
1264                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1265                                         $m[3] = $n[1];
1266                                 }
1267                                 # fix up urlencoded title texts
1268                                 if(preg_match('/%/', $m[1] )) $m[1] = urldecode($m[1]);
1269                                 $trail = $m[3];
1270                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1271                                 $might_be_img = true;
1272                                 $text = $m[2];
1273                                 if(preg_match('/%/', $m[1] )) $m[1] = urldecode($m[1]);
1274                                 $trail = "";
1275                         } else { # Invalid form; output directly
1276                                 $s .= $prefix . '[[' . $line ;
1277                                 continue;
1278                         }
1279
1280                         # Don't allow internal links to pages containing
1281                         # PROTO: where PROTO is a valid URL protocol; these
1282                         # should be external links.
1283                         if (preg_match('/^(\b(?:'.URL_PROTOCOLS.'):)/', $m[1])) {
1284                                 $s .= $prefix . '[[' . $line ;
1285                                 continue;
1286                         }
1287
1288                         # Make subpage if necessary
1289                         if( $useSubpages ) {
1290                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1291                         } else {
1292                                 $link = $m[1];
1293                         }
1294
1295                         $noforce = (substr($m[1], 0, 1) != ':');
1296                         if (!$noforce) {
1297                                 # Strip off leading ':'
1298                                 $link = substr($link, 1);
1299                         }
1300
1301                         $nt =& Title::newFromText( $this->unstripNoWiki($link, $this->mStripState) );
1302                         if( !$nt ) {
1303                                 $s .= $prefix . '[[' . $line;
1304                                 continue;
1305                         }
1306
1307                         #check other language variants of the link
1308                         #if the article does not exist
1309                         if( $checkVariantLink
1310                             && $nt->getArticleID() == 0 ) {
1311                                 $wgContLang->findVariantLink($link, $nt);
1312                         }
1313
1314                         $ns = $nt->getNamespace();
1315                         $iw = $nt->getInterWiki();
1316
1317                         if ($might_be_img) { # if this is actually an invalid link
1318                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1319                                         $found = false;
1320                                         while (isset ($a[$k+1]) ) {
1321                                                 #look at the next 'line' to see if we can close it there
1322                                                 $next_line =  array_shift(array_splice( $a, $k + 1, 1) );
1323                                                 if( preg_match("/^(.*?]].*?)]](.*)$/sD", $next_line, $m) ) {
1324                                                 # the first ]] closes the inner link, the second the image
1325                                                         $found = true;
1326                                                         $text .= '[[' . $m[1];
1327                                                         $trail = $m[2];
1328                                                         break;
1329                                                 } elseif( preg_match("/^.*?]].*$/sD", $next_line, $m) ) {
1330                                                         #if there's exactly one ]] that's fine, we'll keep looking
1331                                                         $text .= '[[' . $m[0];
1332                                                 } else {
1333                                                         #if $next_line is invalid too, we need look no further
1334                                                         $text .= '[[' . $next_line;
1335                                                         break;
1336                                                 }
1337                                         }
1338                                         if ( !$found ) {
1339                                                 # we couldn't find the end of this imageLink, so output it raw
1340                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1341                                                 $text = $this->replaceInternalLinks($text);
1342                                                 $s .= $prefix . '[[' . $link . '|' . $text;
1343                                                 # note: no $trail, because without an end, there *is* no trail
1344                                                 continue;
1345                                         }
1346                                 } else { #it's not an image, so output it raw
1347                                         $s .= $prefix . '[[' . $link . '|' . $text;
1348                                         # note: no $trail, because without an end, there *is* no trail
1349                                         continue;
1350                                 }
1351                         }
1352
1353                         $wasblank = ( '' == $text );
1354                         if( $wasblank ) $text = $link;
1355
1356
1357                         # Link not escaped by : , create the various objects
1358                         if( $noforce ) {
1359
1360                                 # Interwikis
1361                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1362                                         array_push( $this->mOutput->mLanguageLinks, $nt->getFullText() );
1363                                         $s = rtrim($s . "\n");
1364                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1365                                         continue;
1366                                 }
1367
1368                                 if ( $ns == NS_IMAGE ) {
1369                                         wfProfileIn( "$fname-image" );
1370                                         if ( !wfIsBadImage( $nt->getDBkey() ) ) {
1371                                                 # recursively parse links inside the image caption
1372                                                 # actually, this will parse them in any other parameters, too,
1373                                                 # but it might be hard to fix that, and it doesn't matter ATM
1374                                                 $text = $this->replaceExternalLinks($text);
1375                                                 $text = $this->replaceInternalLinks($text);
1376
1377                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1378                                                 $s .= $prefix . str_replace('http://', 'http-noparse://', $this->makeImage( $nt, $text ) ) . $trail;
1379                                                 $wgLinkCache->addImageLinkObj( $nt );
1380
1381                                                 wfProfileOut( "$fname-image" );
1382                                                 continue;
1383                                         }
1384                                         wfProfileOut( "$fname-image" );
1385
1386                                 }
1387
1388                                 if ( $ns == NS_CATEGORY ) {
1389                                         wfProfileIn( "$fname-category" );
1390                                         $t = $wgContLang->convertHtml( $nt->getText() );
1391                                         $s = rtrim($s . "\n"); # bug 87
1392
1393                                         $wgLinkCache->suspend(); # Don't save in links/brokenlinks
1394                                         $t = $sk->makeLinkObj( $nt, $t, '', '' , $prefix );
1395                                         $wgLinkCache->resume();
1396
1397                                         if ( $wasblank ) {
1398                                                 if ( $this->mTitle->getNamespace() == NS_CATEGORY ) {
1399                                                         $sortkey = $this->mTitle->getText();
1400                                                 } else {
1401                                                         $sortkey = $this->mTitle->getPrefixedText();
1402                                                 }
1403                                         } else {
1404                                                 $sortkey = $text;
1405                                         }
1406                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1407                                         $wgLinkCache->addCategoryLinkObj( $nt, $sortkey );
1408                                         $this->mOutput->addCategoryLink( $t );
1409
1410                                         /**
1411                                          * Strip the whitespace Category links produce, see bug 87
1412                                          * @todo We might want to use trim($tmp, "\n") here.
1413                                          */
1414                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1415
1416                                         wfProfileOut( "$fname-category" );
1417                                         continue;
1418                                 }
1419                         }
1420
1421                         if( ( $nt->getPrefixedText() === $selflink ) &&
1422                             ( $nt->getFragment() === '' ) ) {
1423                                 # Self-links are handled specially; generally de-link and change to bold.
1424                                 $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1425                                 continue;
1426                         }
1427
1428                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1429                         if( $ns == NS_MEDIA ) {
1430                                 $s .= $prefix . $sk->makeMediaLinkObj( $nt, $text, true ) . $trail;
1431                                 $wgLinkCache->addImageLinkObj( $nt );
1432                                 continue;
1433                         } elseif( $ns == NS_SPECIAL ) {
1434                                 $s .= $prefix . $sk->makeKnownLinkObj( $nt, $text, '', $trail );
1435                                 continue;
1436                         }
1437                         if( !$nt->isExternal() && $nt->isAlwaysKnown() ) {
1438                                 /**
1439                                  * Skip lookups for special pages and self-links.
1440                                  * External interwiki links are not included here because
1441                                  * the HTTP urls would break output in the next parse step;
1442                                  * they will have placeholders kept.
1443                                  */
1444                                 $s .= $sk->makeKnownLinkObj( $nt, $text, '', $trail, $prefix );
1445                         } else {
1446                                 /**
1447                                  * Add a link placeholder
1448                                  * Later, this will be replaced by a real link, after the existence or
1449                                  * non-existence of all the links is known
1450                                  */
1451                                 $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1452                         }
1453                 }
1454                 wfProfileOut( $fname );
1455                 return $s;
1456         }
1457
1458         /**
1459          * Make a link placeholder. The text returned can be later resolved to a real link with
1460          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1461          * parsing of interwiki links, and secondly to allow all extistence checks and
1462          * article length checks (for stub links) to be bundled into a single query.
1463          *
1464          */
1465         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1466                 if ( ! is_object($nt) ) {
1467                         # Fail gracefully
1468                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1469                 } else {
1470                         # Separate the link trail from the rest of the link
1471                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1472
1473                         if ( $nt->isExternal() ) {
1474                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1475                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1476                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1477                         } else {
1478                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1479                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1480                                 $this->mLinkHolders['queries'][] = $query;
1481                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1482                                 $this->mLinkHolders['titles'][] = $nt;
1483
1484                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1485                         }
1486                 }
1487                 return $retVal;
1488         }
1489
1490         /**
1491          * Return true if subpage links should be expanded on this page.
1492          * @return bool
1493          */
1494         function areSubpagesAllowed() {
1495                 # Some namespaces don't allow subpages
1496                 global $wgNamespacesWithSubpages;
1497                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1498         }
1499
1500         /**
1501          * Handle link to subpage if necessary
1502          * @param string $target the source of the link
1503          * @param string &$text the link text, modified as necessary
1504          * @return string the full name of the link
1505          * @access private
1506          */
1507         function maybeDoSubpageLink($target, &$text) {
1508                 # Valid link forms:
1509                 # Foobar -- normal
1510                 # :Foobar -- override special treatment of prefix (images, language links)
1511                 # /Foobar -- convert to CurrentPage/Foobar
1512                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1513                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1514                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1515
1516                 $fname = 'Parser::maybeDoSubpageLink';
1517                 wfProfileIn( $fname );
1518                 $ret = $target; # default return value is no change
1519
1520                 # Some namespaces don't allow subpages,
1521                 # so only perform processing if subpages are allowed
1522                 if( $this->areSubpagesAllowed() ) {
1523                         # Look at the first character
1524                         if( $target != '' && $target{0} == '/' ) {
1525                                 # / at end means we don't want the slash to be shown
1526                                 if( substr( $target, -1, 1 ) == '/' ) {
1527                                         $target = substr( $target, 1, -1 );
1528                                         $noslash = $target;
1529                                 } else {
1530                                         $noslash = substr( $target, 1 );
1531                                 }
1532
1533                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash);
1534                                 if( '' === $text ) {
1535                                         $text = $target;
1536                                 } # this might be changed for ugliness reasons
1537                         } else {
1538                                 # check for .. subpage backlinks
1539                                 $dotdotcount = 0;
1540                                 $nodotdot = $target;
1541                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1542                                         ++$dotdotcount;
1543                                         $nodotdot = substr( $nodotdot, 3 );
1544                                 }
1545                                 if($dotdotcount > 0) {
1546                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1547                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1548                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1549                                                 # / at the end means don't show full path
1550                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1551                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1552                                                         if( '' === $text ) {
1553                                                                 $text = $nodotdot;
1554                                                         }
1555                                                 }
1556                                                 $nodotdot = trim( $nodotdot );
1557                                                 if( $nodotdot != '' ) {
1558                                                         $ret .= '/' . $nodotdot;
1559                                                 }
1560                                         }
1561                                 }
1562                         }
1563                 }
1564
1565                 wfProfileOut( $fname );
1566                 return $ret;
1567         }
1568
1569         /**#@+
1570          * Used by doBlockLevels()
1571          * @access private
1572          */
1573         /* private */ function closeParagraph() {
1574                 $result = '';
1575                 if ( '' != $this->mLastSection ) {
1576                         $result = '</' . $this->mLastSection  . ">\n";
1577                 }
1578                 $this->mInPre = false;
1579                 $this->mLastSection = '';
1580                 return $result;
1581         }
1582         # getCommon() returns the length of the longest common substring
1583         # of both arguments, starting at the beginning of both.
1584         #
1585         /* private */ function getCommon( $st1, $st2 ) {
1586                 $fl = strlen( $st1 );
1587                 $shorter = strlen( $st2 );
1588                 if ( $fl < $shorter ) { $shorter = $fl; }
1589
1590                 for ( $i = 0; $i < $shorter; ++$i ) {
1591                         if ( $st1{$i} != $st2{$i} ) { break; }
1592                 }
1593                 return $i;
1594         }
1595         # These next three functions open, continue, and close the list
1596         # element appropriate to the prefix character passed into them.
1597         #
1598         /* private */ function openList( $char ) {
1599                 $result = $this->closeParagraph();
1600
1601                 if ( '*' == $char ) { $result .= '<ul><li>'; }
1602                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
1603                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
1604                 else if ( ';' == $char ) {
1605                         $result .= '<dl><dt>';
1606                         $this->mDTopen = true;
1607                 }
1608                 else { $result = '<!-- ERR 1 -->'; }
1609
1610                 return $result;
1611         }
1612
1613         /* private */ function nextItem( $char ) {
1614                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
1615                 else if ( ':' == $char || ';' == $char ) {
1616                         $close = '</dd>';
1617                         if ( $this->mDTopen ) { $close = '</dt>'; }
1618                         if ( ';' == $char ) {
1619                                 $this->mDTopen = true;
1620                                 return $close . '<dt>';
1621                         } else {
1622                                 $this->mDTopen = false;
1623                                 return $close . '<dd>';
1624                         }
1625                 }
1626                 return '<!-- ERR 2 -->';
1627         }
1628
1629         /* private */ function closeList( $char ) {
1630                 if ( '*' == $char ) { $text = '</li></ul>'; }
1631                 else if ( '#' == $char ) { $text = '</li></ol>'; }
1632                 else if ( ':' == $char ) {
1633                         if ( $this->mDTopen ) {
1634                                 $this->mDTopen = false;
1635                                 $text = '</dt></dl>';
1636                         } else {
1637                                 $text = '</dd></dl>';
1638                         }
1639                 }
1640                 else {  return '<!-- ERR 3 -->'; }
1641                 return $text."\n";
1642         }
1643         /**#@-*/
1644
1645         /**
1646          * Make lists from lines starting with ':', '*', '#', etc.
1647          *
1648          * @access private
1649          * @return string the lists rendered as HTML
1650          */
1651         function doBlockLevels( $text, $linestart ) {
1652                 $fname = 'Parser::doBlockLevels';
1653                 wfProfileIn( $fname );
1654
1655                 # Parsing through the text line by line.  The main thing
1656                 # happening here is handling of block-level elements p, pre,
1657                 # and making lists from lines starting with * # : etc.
1658                 #
1659                 $textLines = explode( "\n", $text );
1660
1661                 $lastPrefix = $output = '';
1662                 $this->mDTopen = $inBlockElem = false;
1663                 $prefixLength = 0;
1664                 $paragraphStack = false;
1665
1666                 if ( !$linestart ) {
1667                         $output .= array_shift( $textLines );
1668                 }
1669                 foreach ( $textLines as $oLine ) {
1670                         $lastPrefixLength = strlen( $lastPrefix );
1671                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
1672                         $preOpenMatch = preg_match('/<pre/i', $oLine );
1673                         if ( !$this->mInPre ) {
1674                                 # Multiple prefixes may abut each other for nested lists.
1675                                 $prefixLength = strspn( $oLine, '*#:;' );
1676                                 $pref = substr( $oLine, 0, $prefixLength );
1677
1678                                 # eh?
1679                                 $pref2 = str_replace( ';', ':', $pref );
1680                                 $t = substr( $oLine, $prefixLength );
1681                                 $this->mInPre = !empty($preOpenMatch);
1682                         } else {
1683                                 # Don't interpret any other prefixes in preformatted text
1684                                 $prefixLength = 0;
1685                                 $pref = $pref2 = '';
1686                                 $t = $oLine;
1687                         }
1688
1689                         # List generation
1690                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
1691                                 # Same as the last item, so no need to deal with nesting or opening stuff
1692                                 $output .= $this->nextItem( substr( $pref, -1 ) );
1693                                 $paragraphStack = false;
1694
1695                                 if ( substr( $pref, -1 ) == ';') {
1696                                         # The one nasty exception: definition lists work like this:
1697                                         # ; title : definition text
1698                                         # So we check for : in the remainder text to split up the
1699                                         # title and definition, without b0rking links.
1700                                         $term = $t2 = '';
1701                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
1702                                                 $t = $t2;
1703                                                 $output .= $term . $this->nextItem( ':' );
1704                                         }
1705                                 }
1706                         } elseif( $prefixLength || $lastPrefixLength ) {
1707                                 # Either open or close a level...
1708                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
1709                                 $paragraphStack = false;
1710
1711                                 while( $commonPrefixLength < $lastPrefixLength ) {
1712                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
1713                                         --$lastPrefixLength;
1714                                 }
1715                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
1716                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
1717                                 }
1718                                 while ( $prefixLength > $commonPrefixLength ) {
1719                                         $char = substr( $pref, $commonPrefixLength, 1 );
1720                                         $output .= $this->openList( $char );
1721
1722                                         if ( ';' == $char ) {
1723                                                 # FIXME: This is dupe of code above
1724                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
1725                                                         $t = $t2;
1726                                                         $output .= $term . $this->nextItem( ':' );
1727                                                 }
1728                                         }
1729                                         ++$commonPrefixLength;
1730                                 }
1731                                 $lastPrefix = $pref2;
1732                         }
1733                         if( 0 == $prefixLength ) {
1734                                 wfProfileIn( "$fname-paragraph" );
1735                                 # No prefix (not in list)--go to paragraph mode
1736                                 $uniq_prefix = UNIQ_PREFIX;
1737                                 // XXX: use a stack for nestable elements like span, table and div
1738                                 $openmatch = preg_match('/(<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
1739                                 $closematch = preg_match(
1740                                         '/(<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
1741                                         '<td|<th|<div|<\\/div|<hr|<\\/pre|<\\/p|'.$uniq_prefix.'-pre|<\\/li|<\\/ul)/iS', $t );
1742                                 if ( $openmatch or $closematch ) {
1743                                         $paragraphStack = false;
1744                                         $output .= $this->closeParagraph();
1745                                         if($preOpenMatch and !$preCloseMatch) {
1746                                                 $this->mInPre = true;
1747                                         }
1748                                         if ( $closematch ) {
1749                                                 $inBlockElem = false;
1750                                         } else {
1751                                                 $inBlockElem = true;
1752                                         }
1753                                 } else if ( !$inBlockElem && !$this->mInPre ) {
1754                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
1755                                                 // pre
1756                                                 if ($this->mLastSection != 'pre') {
1757                                                         $paragraphStack = false;
1758                                                         $output .= $this->closeParagraph().'<pre>';
1759                                                         $this->mLastSection = 'pre';
1760                                                 }
1761                                                 $t = substr( $t, 1 );
1762                                         } else {
1763                                                 // paragraph
1764                                                 if ( '' == trim($t) ) {
1765                                                         if ( $paragraphStack ) {
1766                                                                 $output .= $paragraphStack.'<br />';
1767                                                                 $paragraphStack = false;
1768                                                                 $this->mLastSection = 'p';
1769                                                         } else {
1770                                                                 if ($this->mLastSection != 'p' ) {
1771                                                                         $output .= $this->closeParagraph();
1772                                                                         $this->mLastSection = '';
1773                                                                         $paragraphStack = '<p>';
1774                                                                 } else {
1775                                                                         $paragraphStack = '</p><p>';
1776                                                                 }
1777                                                         }
1778                                                 } else {
1779                                                         if ( $paragraphStack ) {
1780                                                                 $output .= $paragraphStack;
1781                                                                 $paragraphStack = false;
1782                                                                 $this->mLastSection = 'p';
1783                                                         } else if ($this->mLastSection != 'p') {
1784                                                                 $output .= $this->closeParagraph().'<p>';
1785                                                                 $this->mLastSection = 'p';
1786                                                         }
1787                                                 }
1788                                         }
1789                                 }
1790                                 wfProfileOut( "$fname-paragraph" );
1791                         }
1792                         if ($paragraphStack === false) {
1793                                 $output .= $t."\n";
1794                         }
1795                 }
1796                 while ( $prefixLength ) {
1797                         $output .= $this->closeList( $pref2{$prefixLength-1} );
1798                         --$prefixLength;
1799                 }
1800                 if ( '' != $this->mLastSection ) {
1801                         $output .= '</' . $this->mLastSection . '>';
1802                         $this->mLastSection = '';
1803                 }
1804
1805                 wfProfileOut( $fname );
1806                 return $output;
1807         }
1808
1809         /**
1810          * Split up a string on ':', ignoring any occurences inside
1811          * <a>..</a> or <span>...</span>
1812          * @param string $str the string to split
1813          * @param string &$before set to everything before the ':'
1814          * @param string &$after set to everything after the ':'
1815          * return string the position of the ':', or false if none found
1816          */
1817         function findColonNoLinks($str, &$before, &$after) {
1818                 # I wonder if we should make this count all tags, not just <a>
1819                 # and <span>. That would prevent us from matching a ':' that
1820                 # comes in the middle of italics other such formatting....
1821                 # -- Wil
1822                 $fname = 'Parser::findColonNoLinks';
1823                 wfProfileIn( $fname );
1824                 $pos = 0;
1825                 do {
1826                         $colon = strpos($str, ':', $pos);
1827
1828                         if ($colon !== false) {
1829                                 $before = substr($str, 0, $colon);
1830                                 $after = substr($str, $colon + 1);
1831
1832                                 # Skip any ':' within <a> or <span> pairs
1833                                 $a = substr_count($before, '<a');
1834                                 $s = substr_count($before, '<span');
1835                                 $ca = substr_count($before, '</a>');
1836                                 $cs = substr_count($before, '</span>');
1837
1838                                 if ($a <= $ca and $s <= $cs) {
1839                                         # Tags are balanced before ':'; ok
1840                                         break;
1841                                 }
1842                                 $pos = $colon + 1;
1843                         }
1844                 } while ($colon !== false);
1845                 wfProfileOut( $fname );
1846                 return $colon;
1847         }
1848
1849         /**
1850          * Return value of a magic variable (like PAGENAME)
1851          *
1852          * @access private
1853          */
1854         function getVariableValue( $index ) {
1855                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgArticle, $wgScriptPath;
1856
1857                 /**
1858                  * Some of these require message or data lookups and can be
1859                  * expensive to check many times.
1860                  */
1861                 static $varCache = array();
1862                 if( isset( $varCache[$index] ) ) return $varCache[$index];
1863
1864                 switch ( $index ) {
1865                         case MAG_CURRENTMONTH:
1866                                 return $varCache[$index] = $wgContLang->formatNum( date( 'm' ) );
1867                         case MAG_CURRENTMONTHNAME:
1868                                 return $varCache[$index] = $wgContLang->getMonthName( date('n') );
1869                         case MAG_CURRENTMONTHNAMEGEN:
1870                                 return $varCache[$index] = $wgContLang->getMonthNameGen( date('n') );
1871                         case MAG_CURRENTMONTHABBREV:
1872                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( date('n') );
1873                         case MAG_CURRENTDAY:
1874                                 return $varCache[$index] = $wgContLang->formatNum( date('j') );
1875                         case MAG_PAGENAME:
1876                                 return $this->mTitle->getText();
1877                         case MAG_PAGENAMEE:
1878                                 return $this->mTitle->getPartialURL();
1879                         case MAG_REVISIONID:
1880                                 return $wgArticle->getRevIdFetched();
1881                         case MAG_NAMESPACE:
1882                                 # return Namespace::getCanonicalName($this->mTitle->getNamespace());
1883                                 return $wgContLang->getNsText($this->mTitle->getNamespace()); # Patch by Dori
1884                         case MAG_CURRENTDAYNAME:
1885                                 return $varCache[$index] = $wgContLang->getWeekdayName( date('w')+1 );
1886                         case MAG_CURRENTYEAR:
1887                                 return $varCache[$index] = $wgContLang->formatNum( date( 'Y' ), true );
1888                         case MAG_CURRENTTIME:
1889                                 return $varCache[$index] = $wgContLang->time( wfTimestampNow(), false );
1890                         case MAG_CURRENTWEEK:
1891                                 return $varCache[$index] = $wgContLang->formatNum( date('W') );
1892                         case MAG_CURRENTDOW:
1893                                 return $varCache[$index] = $wgContLang->formatNum( date('w') );
1894                         case MAG_NUMBEROFARTICLES:
1895                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfArticles() );
1896                         case MAG_NUMBEROFFILES:
1897                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfFiles() );
1898                         case MAG_SITENAME:
1899                                 return $wgSitename;
1900                         case MAG_SERVER:
1901                                 return $wgServer;
1902                         case MAG_SERVERNAME:
1903                                 return $wgServerName;
1904                         case MAG_SCRIPTPATH:
1905                                 return $wgScriptPath;
1906                         default:
1907                                 return NULL;
1908                 }
1909         }
1910
1911         /**
1912          * initialise the magic variables (like CURRENTMONTHNAME)
1913          *
1914          * @access private
1915          */
1916         function initialiseVariables() {
1917                 $fname = 'Parser::initialiseVariables';
1918                 wfProfileIn( $fname );
1919                 global $wgVariableIDs;
1920                 $this->mVariables = array();
1921                 foreach ( $wgVariableIDs as $id ) {
1922                         $mw =& MagicWord::get( $id );
1923                         $mw->addToArray( $this->mVariables, $id );
1924                 }
1925                 wfProfileOut( $fname );
1926         }
1927
1928         /**
1929          * Replace magic variables, templates, and template arguments
1930          * with the appropriate text. Templates are substituted recursively,
1931          * taking care to avoid infinite loops.
1932          *
1933          * Note that the substitution depends on value of $mOutputType:
1934          *  OT_WIKI: only {{subst:}} templates
1935          *  OT_MSG: only magic variables
1936          *  OT_HTML: all templates and magic variables
1937          *
1938          * @param string $tex The text to transform
1939          * @param array $args Key-value pairs representing template parameters to substitute
1940          * @access private
1941          */
1942         function replaceVariables( $text, $args = array() ) {
1943
1944                 # Prevent too big inclusions
1945                 if( strlen( $text ) > MAX_INCLUDE_SIZE ) {
1946                         return $text;
1947                 }
1948
1949                 $fname = 'Parser::replaceVariables';
1950                 wfProfileIn( $fname );
1951
1952                 $titleChars = Title::legalChars();
1953
1954                 # This function is called recursively. To keep track of arguments we need a stack:
1955                 array_push( $this->mArgStack, $args );
1956
1957                 # Variable substitution
1958                 $text = preg_replace_callback( "/{{([$titleChars]*?)}}/", array( &$this, 'variableSubstitution' ), $text );
1959
1960                 if ( $this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI ) {
1961                         # Argument substitution
1962                         $text = preg_replace_callback( "/{{{([$titleChars]*?)}}}/", array( &$this, 'argSubstitution' ), $text );
1963                 }
1964                 # Template substitution
1965                 $regex = '/(\\n|{)?{{(['.$titleChars.']*)(\\|.*?|)}}/s';
1966                 $text = preg_replace_callback( $regex, array( &$this, 'braceSubstitution' ), $text );
1967
1968                 array_pop( $this->mArgStack );
1969
1970                 wfProfileOut( $fname );
1971                 return $text;
1972         }
1973
1974         /**
1975          * Replace magic variables
1976          * @access private
1977          */
1978         function variableSubstitution( $matches ) {
1979                 $fname = 'parser::variableSubstitution';
1980                 $varname = $matches[1];
1981                 wfProfileIn( $fname );
1982                 if ( !$this->mVariables ) {
1983                         $this->initialiseVariables();
1984                 }
1985                 $skip = false;
1986                 if ( $this->mOutputType == OT_WIKI ) {
1987                         # Do only magic variables prefixed by SUBST
1988                         $mwSubst =& MagicWord::get( MAG_SUBST );
1989                         if (!$mwSubst->matchStartAndRemove( $varname ))
1990                                 $skip = true;
1991                         # Note that if we don't substitute the variable below,
1992                         # we don't remove the {{subst:}} magic word, in case
1993                         # it is a template rather than a magic variable.
1994                 }
1995                 if ( !$skip && array_key_exists( $varname, $this->mVariables ) ) {
1996                         $id = $this->mVariables[$varname];
1997                         $text = $this->getVariableValue( $id );
1998                         $this->mOutput->mContainsOldMagic = true;
1999                 } else {
2000                         $text = $matches[0];
2001                 }
2002                 wfProfileOut( $fname );
2003                 return $text;
2004         }
2005
2006         # Split template arguments
2007         function getTemplateArgs( $argsString ) {
2008                 if ( $argsString === '' ) {
2009                         return array();
2010                 }
2011
2012                 $args = explode( '|', substr( $argsString, 1 ) );
2013
2014                 # If any of the arguments contains a '[[' but no ']]', it needs to be
2015                 # merged with the next arg because the '|' character between belongs
2016                 # to the link syntax and not the template parameter syntax.
2017                 $argc = count($args);
2018
2019                 for ( $i = 0; $i < $argc-1; $i++ ) {
2020                         if ( substr_count ( $args[$i], '[[' ) != substr_count ( $args[$i], ']]' ) ) {
2021                                 $args[$i] .= '|'.$args[$i+1];
2022                                 array_splice($args, $i+1, 1);
2023                                 $i--;
2024                                 $argc--;
2025                         }
2026                 }
2027
2028                 return $args;
2029         }
2030
2031         /**
2032          * Return the text of a template, after recursively
2033          * replacing any variables or templates within the template.
2034          *
2035          * @param array $matches The parts of the template
2036          *  $matches[1]: the title, i.e. the part before the |
2037          *  $matches[2]: the parameters (including a leading |), if  any
2038          * @return string the text of the template
2039          * @access private
2040          */
2041         function braceSubstitution( $matches ) {
2042                 global $wgLinkCache, $wgContLang;
2043                 $fname = 'Parser::braceSubstitution';
2044                 wfProfileIn( $fname );
2045
2046                 $found = false;
2047                 $nowiki = false;
2048                 $noparse = false;
2049
2050                 $title = NULL;
2051
2052                 # Need to know if the template comes at the start of a line,
2053                 # to treat the beginning of the template like the beginning
2054                 # of a line for tables and block-level elements.
2055                 $linestart = $matches[1];
2056
2057                 # $part1 is the bit before the first |, and must contain only title characters
2058                 # $args is a list of arguments, starting from index 0, not including $part1
2059
2060                 $part1 = $matches[2];
2061                 # If the third subpattern matched anything, it will start with |
2062
2063                 $args = $this->getTemplateArgs($matches[3]);
2064                 $argc = count( $args );
2065
2066                 # Don't parse {{{}}} because that's only for template arguments
2067                 if ( $linestart === '{' ) {
2068                         $text = $matches[0];
2069                         $found = true;
2070                         $noparse = true;
2071                 }
2072
2073                 # SUBST
2074                 if ( !$found ) {
2075                         $mwSubst =& MagicWord::get( MAG_SUBST );
2076                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor ($this->mOutputType == OT_WIKI) ) {
2077                                 # One of two possibilities is true:
2078                                 # 1) Found SUBST but not in the PST phase
2079                                 # 2) Didn't find SUBST and in the PST phase
2080                                 # In either case, return without further processing
2081                                 $text = $matches[0];
2082                                 $found = true;
2083                                 $noparse = true;
2084                         }
2085                 }
2086
2087                 # MSG, MSGNW and INT
2088                 if ( !$found ) {
2089                         # Check for MSGNW:
2090                         $mwMsgnw =& MagicWord::get( MAG_MSGNW );
2091                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2092                                 $nowiki = true;
2093                         } else {
2094                                 # Remove obsolete MSG:
2095                                 $mwMsg =& MagicWord::get( MAG_MSG );
2096                                 $mwMsg->matchStartAndRemove( $part1 );
2097                         }
2098
2099                         # Check if it is an internal message
2100                         $mwInt =& MagicWord::get( MAG_INT );
2101                         if ( $mwInt->matchStartAndRemove( $part1 ) ) {
2102                                 if ( $this->incrementIncludeCount( 'int:'.$part1 ) ) {
2103                                         $text = $linestart . wfMsgReal( $part1, $args, true );
2104                                         $found = true;
2105                                 }
2106                         }
2107                 }
2108
2109                 # NS
2110                 if ( !$found ) {
2111                         # Check for NS: (namespace expansion)
2112                         $mwNs = MagicWord::get( MAG_NS );
2113                         if ( $mwNs->matchStartAndRemove( $part1 ) ) {
2114                                 if ( intval( $part1 ) ) {
2115                                         $text = $linestart . $wgContLang->getNsText( intval( $part1 ) );
2116                                         $found = true;
2117                                 } else {
2118                                         $index = Namespace::getCanonicalIndex( strtolower( $part1 ) );
2119                                         if ( !is_null( $index ) ) {
2120                                                 $text = $linestart . $wgContLang->getNsText( $index );
2121                                                 $found = true;
2122                                         }
2123                                 }
2124                         }
2125                 }
2126
2127                 # LOCALURL and LOCALURLE
2128                 if ( !$found ) {
2129                         $mwLocal = MagicWord::get( MAG_LOCALURL );
2130                         $mwLocalE = MagicWord::get( MAG_LOCALURLE );
2131
2132                         if ( $mwLocal->matchStartAndRemove( $part1 ) ) {
2133                                 $func = 'getLocalURL';
2134                         } elseif ( $mwLocalE->matchStartAndRemove( $part1 ) ) {
2135                                 $func = 'escapeLocalURL';
2136                         } else {
2137                                 $func = '';
2138                         }
2139
2140                         if ( $func !== '' ) {
2141                                 $title = Title::newFromText( $part1 );
2142                                 if ( !is_null( $title ) ) {
2143                                         if ( $argc > 0 ) {
2144                                                 $text = $linestart . $title->$func( $args[0] );
2145                                         } else {
2146                                                 $text = $linestart . $title->$func();
2147                                         }
2148                                         $found = true;
2149                                 }
2150                         }
2151                 }
2152
2153                 # GRAMMAR
2154                 if ( !$found && $argc == 1 ) {
2155                         $mwGrammar =& MagicWord::get( MAG_GRAMMAR );
2156                         if ( $mwGrammar->matchStartAndRemove( $part1 ) ) {
2157                                 $text = $linestart . $wgContLang->convertGrammar( $args[0], $part1 );
2158                                 $found = true;
2159                         }
2160                 }
2161
2162                 # Template table test
2163
2164                 # Did we encounter this template already? If yes, it is in the cache
2165                 # and we need to check for loops.
2166                 if ( !$found && isset( $this->mTemplates[$part1] ) ) {
2167                         $found = true;
2168
2169                         # Infinite loop test
2170                         if ( isset( $this->mTemplatePath[$part1] ) ) {
2171                                 $noparse = true;
2172                                 $found = true;
2173                                 $text = $linestart .
2174                                         "\{\{$part1}}" .
2175                                         '<!-- WARNING: template loop detected -->';
2176                                 wfDebug( "$fname: template loop broken at '$part1'\n" );
2177                         } else {
2178                                 # set $text to cached message.
2179                                 $text = $linestart . $this->mTemplates[$part1];
2180                         }
2181                 }
2182
2183                 # Load from database
2184                 $replaceHeadings = false;
2185                 $isHTML = false;
2186                 $lastPathLevel = $this->mTemplatePath;
2187                 if ( !$found ) {
2188                         $ns = NS_TEMPLATE;
2189                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage='' );
2190                         if ($subpage !== '') {
2191                                 $ns = $this->mTitle->getNamespace();
2192                         }
2193                         $title = Title::newFromText( $part1, $ns );
2194                         if ( !is_null( $title ) && !$title->isExternal() ) {
2195                                 # Check for excessive inclusion
2196                                 $dbk = $title->getPrefixedDBkey();
2197                                 if ( $this->incrementIncludeCount( $dbk ) ) {
2198                                         if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() ) {
2199                                                 # Capture special page output
2200                                                 $text = SpecialPage::capturePath( $title );
2201                                                 if ( $text && !is_object( $text ) ) {
2202                                                         $found = true;
2203                                                         $noparse = true;
2204                                                         $isHTML = true;
2205                                                         $this->mOutput->setCacheTime( -1 );
2206                                                 }
2207                                         } else {
2208                                                 $article = new Article( $title );
2209                                                 $articleContent = $article->getContentWithoutUsingSoManyDamnGlobals();
2210                                                 if ( $articleContent !== false ) {
2211                                                         $found = true;
2212                                                         $text = $articleContent;
2213                                                         $replaceHeadings = true;
2214                                                 }
2215                                         }
2216                                 }
2217
2218                                 # If the title is valid but undisplayable, make a link to it
2219                                 if ( $this->mOutputType == OT_HTML && !$found ) {
2220                                         $text = '[['.$title->getPrefixedText().']]';
2221                                         $found = true;
2222                                 }
2223
2224                                 # Template cache array insertion
2225                                 if( $found ) {
2226                                         $this->mTemplates[$part1] = $text;
2227                                         $text = $linestart . $text;
2228                                 }
2229                         }
2230                 }
2231
2232                 # Recursive parsing, escaping and link table handling
2233                 # Only for HTML output
2234                 if ( $nowiki && $found && $this->mOutputType == OT_HTML ) {
2235                         $text = wfEscapeWikiText( $text );
2236                 } elseif ( ($this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI) && $found && !$noparse) {
2237                         # Clean up argument array
2238                         $assocArgs = array();
2239                         $index = 1;
2240                         foreach( $args as $arg ) {
2241                                 $eqpos = strpos( $arg, '=' );
2242                                 if ( $eqpos === false ) {
2243                                         $assocArgs[$index++] = $arg;
2244                                 } else {
2245                                         $name = trim( substr( $arg, 0, $eqpos ) );
2246                                         $value = trim( substr( $arg, $eqpos+1 ) );
2247                                         if ( $value === false ) {
2248                                                 $value = '';
2249                                         }
2250                                         if ( $name !== false ) {
2251                                                 $assocArgs[$name] = $value;
2252                                         }
2253                                 }
2254                         }
2255
2256                         # Add a new element to the templace recursion path
2257                         $this->mTemplatePath[$part1] = 1;
2258
2259                         if( $this->mOutputType == OT_HTML ) {
2260                                 $text = $this->strip( $text, $this->mStripState );
2261                                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'replaceVariables' ), $assocArgs );
2262                         }
2263                         $text = $this->replaceVariables( $text, $assocArgs );
2264
2265                         # Resume the link cache and register the inclusion as a link
2266                         if ( $this->mOutputType == OT_HTML && !is_null( $title ) ) {
2267                                 $wgLinkCache->addLinkObj( $title );
2268                         }
2269
2270                         # If the template begins with a table or block-level
2271                         # element, it should be treated as beginning a new line.
2272                         if ($linestart !== '\n' && preg_match('/^({\\||:|;|#|\*)/', $text)) {
2273                                 $text = "\n" . $text;
2274                         }
2275                 }
2276                 # Prune lower levels off the recursion check path
2277                 $this->mTemplatePath = $lastPathLevel;
2278
2279                 if ( !$found ) {
2280                         wfProfileOut( $fname );
2281                         return $matches[0];
2282                 } else {
2283                         if ( $isHTML ) {
2284                                 # Replace raw HTML by a placeholder
2285                                 # Add a blank line preceding, to prevent it from mucking up
2286                                 # immediately preceding headings
2287                                 $text = "\n\n" . $this->insertStripItem( $text, $this->mStripState );
2288                         } else {
2289                                 # replace ==section headers==
2290                                 # XXX this needs to go away once we have a better parser.
2291                                 if ( $this->mOutputType != OT_WIKI && $replaceHeadings ) {
2292                                         if( !is_null( $title ) )
2293                                                 $encodedname = base64_encode($title->getPrefixedDBkey());
2294                                         else
2295                                                 $encodedname = base64_encode("");
2296                                         $m = preg_split('/(^={1,6}.*?={1,6}\s*?$)/m', $text, -1,
2297                                                 PREG_SPLIT_DELIM_CAPTURE);
2298                                         $text = '';
2299                                         $nsec = 0;
2300                                         for( $i = 0; $i < count($m); $i += 2 ) {
2301                                                 $text .= $m[$i];
2302                                                 if (!isset($m[$i + 1]) || $m[$i + 1] == "") continue;
2303                                                 $hl = $m[$i + 1];
2304                                                 if( strstr($hl, "<!--MWTEMPLATESECTION") ) {
2305                                                         $text .= $hl;
2306                                                         continue;
2307                                                 }
2308                                                 preg_match('/^(={1,6})(.*?)(={1,6})\s*?$/m', $hl, $m2);
2309                                                 $text .= $m2[1] . $m2[2] . "<!--MWTEMPLATESECTION="
2310                                                         . $encodedname . "&" . base64_encode("$nsec") . "-->" . $m2[3];
2311
2312                                                 $nsec++;
2313                                         }
2314                                 }
2315                         }
2316                 }
2317
2318                 # Prune lower levels off the recursion check path
2319                 $this->mTemplatePath = $lastPathLevel;
2320
2321                 if ( !$found ) {
2322                         wfProfileOut( $fname );
2323                         return $matches[0];
2324                 } else {
2325                         wfProfileOut( $fname );
2326                         return $text;
2327                 }
2328         }
2329
2330         /**
2331          * Triple brace replacement -- used for template arguments
2332          * @access private
2333          */
2334         function argSubstitution( $matches ) {
2335                 $arg = trim( $matches[1] );
2336                 $text = $matches[0];
2337                 $inputArgs = end( $this->mArgStack );
2338
2339                 if ( array_key_exists( $arg, $inputArgs ) ) {
2340                         $text = $inputArgs[$arg];
2341                 }
2342
2343                 return $text;
2344         }
2345
2346         /**
2347          * Returns true if the function is allowed to include this entity
2348          * @access private
2349          */
2350         function incrementIncludeCount( $dbk ) {
2351                 if ( !array_key_exists( $dbk, $this->mIncludeCount ) ) {
2352                         $this->mIncludeCount[$dbk] = 0;
2353                 }
2354                 if ( ++$this->mIncludeCount[$dbk] <= MAX_INCLUDE_REPEAT ) {
2355                         return true;
2356                 } else {
2357                         return false;
2358                 }
2359         }
2360
2361         /**
2362          * This function accomplishes several tasks:
2363          * 1) Auto-number headings if that option is enabled
2364          * 2) Add an [edit] link to sections for logged in users who have enabled the option
2365          * 3) Add a Table of contents on the top for users who have enabled the option
2366          * 4) Auto-anchor headings
2367          *
2368          * It loops through all headlines, collects the necessary data, then splits up the
2369          * string and re-inserts the newly formatted headlines.
2370          *
2371          * @param string $text
2372          * @param boolean $isMain
2373          * @access private
2374          */
2375         function formatHeadings( $text, $isMain=true ) {
2376                 global $wgMaxTocLevel, $wgContLang, $wgLinkHolders, $wgInterwikiLinkHolders;
2377
2378                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
2379                 $doShowToc = true;
2380                 $forceTocHere = false;
2381                 if( !$this->mTitle->userCanEdit() ) {
2382                         $showEditLink = 0;
2383                 } else {
2384                         $showEditLink = $this->mOptions->getEditSection();
2385                 }
2386
2387                 # Inhibit editsection links if requested in the page
2388                 $esw =& MagicWord::get( MAG_NOEDITSECTION );
2389                 if( $esw->matchAndRemove( $text ) ) {
2390                         $showEditLink = 0;
2391                 }
2392                 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
2393                 # do not add TOC
2394                 $mw =& MagicWord::get( MAG_NOTOC );
2395                 if( $mw->matchAndRemove( $text ) ) {
2396                         $doShowToc = false;
2397                 }
2398
2399                 # Get all headlines for numbering them and adding funky stuff like [edit]
2400                 # links - this is for later, but we need the number of headlines right now
2401                 $numMatches = preg_match_all( '/<H([1-6])(.*?'.'>)(.*?)<\/H[1-6] *>/i', $text, $matches );
2402
2403                 # if there are fewer than 4 headlines in the article, do not show TOC
2404                 if( $numMatches < 4 ) {
2405                         $doShowToc = false;
2406                 }
2407
2408                 # if the string __TOC__ (not case-sensitive) occurs in the HTML,
2409                 # override above conditions and always show TOC at that place
2410
2411                 $mw =& MagicWord::get( MAG_TOC );
2412                 if($mw->match( $text ) ) {
2413                         $doShowToc = true;
2414                         $forceTocHere = true;
2415                 } else {
2416                         # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
2417                         # override above conditions and always show TOC above first header
2418                         $mw =& MagicWord::get( MAG_FORCETOC );
2419                         if ($mw->matchAndRemove( $text ) ) {
2420                                 $doShowToc = true;
2421                         }
2422                 }
2423
2424                 # Never ever show TOC if no headers
2425                 if( $numMatches < 1 ) {
2426                         $doShowToc = false;
2427                 }
2428
2429                 # We need this to perform operations on the HTML
2430                 $sk =& $this->mOptions->getSkin();
2431
2432                 # headline counter
2433                 $headlineCount = 0;
2434                 $sectionCount = 0; # headlineCount excluding template sections
2435
2436                 # Ugh .. the TOC should have neat indentation levels which can be
2437                 # passed to the skin functions. These are determined here
2438                 $toc = '';
2439                 $full = '';
2440                 $head = array();
2441                 $sublevelCount = array();
2442                 $levelCount = array();
2443                 $toclevel = 0;
2444                 $level = 0;
2445                 $prevlevel = 0;
2446                 $toclevel = 0;
2447                 $prevtoclevel = 0;
2448
2449                 foreach( $matches[3] as $headline ) {
2450                         $istemplate = 0;
2451                         $templatetitle = '';
2452                         $templatesection = 0;
2453                         $numbering = '';
2454
2455                         if (preg_match("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", $headline, $mat)) {
2456                                 $istemplate = 1;
2457                                 $templatetitle = base64_decode($mat[1]);
2458                                 $templatesection = 1 + (int)base64_decode($mat[2]);
2459                                 $headline = preg_replace("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", "", $headline);
2460                         }
2461
2462                         if( $toclevel ) {
2463                                 $prevlevel = $level;
2464                                 $prevtoclevel = $toclevel;
2465                         }
2466                         $level = $matches[1][$headlineCount];
2467
2468                         if( $doNumberHeadings || $doShowToc ) {
2469
2470                                 if ( $level > $prevlevel ) {
2471                                         # Increase TOC level
2472                                         $toclevel++;
2473                                         $sublevelCount[$toclevel] = 0;
2474                                         $toc .= $sk->tocIndent();
2475                                 }
2476                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
2477                                         # Decrease TOC level, find level to jump to
2478
2479                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
2480                                                 # Can only go down to level 1
2481                                                 $toclevel = 1;
2482                                         } else {
2483                                                 for ($i = $toclevel; $i > 0; $i--) {
2484                                                         if ( $levelCount[$i] == $level ) {
2485                                                                 # Found last matching level
2486                                                                 $toclevel = $i;
2487                                                                 break;
2488                                                         }
2489                                                         elseif ( $levelCount[$i] < $level ) {
2490                                                                 # Found first matching level below current level
2491                                                                 $toclevel = $i + 1;
2492                                                                 break;
2493                                                         }
2494                                                 }
2495                                         }
2496
2497                                         $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
2498                                 }
2499                                 else {
2500                                         # No change in level, end TOC line
2501                                         $toc .= $sk->tocLineEnd();
2502                                 }
2503
2504                                 $levelCount[$toclevel] = $level;
2505
2506                                 # count number of headlines for each level
2507                                 @$sublevelCount[$toclevel]++;
2508                                 $dot = 0;
2509                                 for( $i = 1; $i <= $toclevel; $i++ ) {
2510                                         if( !empty( $sublevelCount[$i] ) ) {
2511                                                 if( $dot ) {
2512                                                         $numbering .= '.';
2513                                                 }
2514                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
2515                                                 $dot = 1;
2516                                         }
2517                                 }
2518                         }
2519
2520                         # The canonized header is a version of the header text safe to use for links
2521                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
2522                         $canonized_headline = $this->unstrip( $headline, $this->mStripState );
2523                         $canonized_headline = $this->unstripNoWiki( $canonized_headline, $this->mStripState );
2524
2525                         # Remove link placeholders by the link text.
2526                         #     <!--LINK number-->
2527                         # turns into
2528                         #     link text with suffix
2529                         $canonized_headline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
2530                                                             "\$this->mLinkHolders['texts'][\$1]",
2531                                                             $canonized_headline );
2532                         $canonized_headline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
2533                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
2534                                                             $canonized_headline );
2535
2536                         # strip out HTML
2537                         $canonized_headline = preg_replace( '/<.*?' . '>/','',$canonized_headline );
2538                         $tocline = trim( $canonized_headline );
2539                         $canonized_headline = urlencode( Sanitizer::decodeCharReferences( str_replace(' ', '_', $tocline) ) );
2540                         $replacearray = array(
2541                                 '%3A' => ':',
2542                                 '%' => '.'
2543                         );
2544                         $canonized_headline = str_replace(array_keys($replacearray),array_values($replacearray),$canonized_headline);
2545                         $refers[$headlineCount] = $canonized_headline;
2546
2547                         # count how many in assoc. array so we can track dupes in anchors
2548                         @$refers[$canonized_headline]++;
2549                         $refcount[$headlineCount]=$refers[$canonized_headline];
2550
2551                         # Don't number the heading if it is the only one (looks silly)
2552                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
2553                                 # the two are different if the line contains a link
2554                                 $headline=$numbering . ' ' . $headline;
2555                         }
2556
2557                         # Create the anchor for linking from the TOC to the section
2558                         $anchor = $canonized_headline;
2559                         if($refcount[$headlineCount] > 1 ) {
2560                                 $anchor .= '_' . $refcount[$headlineCount];
2561                         }
2562                         if( $doShowToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
2563                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
2564                         }
2565                         if( $showEditLink && ( !$istemplate || $templatetitle !== "" ) ) {
2566                                 if ( empty( $head[$headlineCount] ) ) {
2567                                         $head[$headlineCount] = '';
2568                                 }
2569                                 if( $istemplate )
2570                                         $head[$headlineCount] .= $sk->editSectionLinkForOther($templatetitle, $templatesection);
2571                                 else
2572                                         $head[$headlineCount] .= $sk->editSectionLink($this->mTitle, $sectionCount+1);
2573                         }
2574
2575                         # give headline the correct <h#> tag
2576                         @$head[$headlineCount] .= "<a name=\"$anchor\"></a><h".$level.$matches[2][$headlineCount] .$headline.'</h'.$level.'>';
2577
2578                         $headlineCount++;
2579                         if( !$istemplate )
2580                                 $sectionCount++;
2581                 }
2582
2583                 if( $doShowToc ) {
2584                         $toc .= $sk->tocUnindent( $toclevel - 1 );
2585                         $toc = $sk->tocList( $toc );
2586                 }
2587
2588                 # split up and insert constructed headlines
2589
2590                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
2591                 $i = 0;
2592
2593                 foreach( $blocks as $block ) {
2594                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
2595                                 # This is the [edit] link that appears for the top block of text when
2596                                 # section editing is enabled
2597
2598                                 # Disabled because it broke block formatting
2599                                 # For example, a bullet point in the top line
2600                                 # $full .= $sk->editSectionLink(0);
2601                         }
2602                         $full .= $block;
2603                         if( $doShowToc && !$i && $isMain && !$forceTocHere) {
2604                         # Top anchor now in skin
2605                                 $full = $full.$toc;
2606                         }
2607
2608                         if( !empty( $head[$i] ) ) {
2609                                 $full .= $head[$i];
2610                         }
2611                         $i++;
2612                 }
2613                 if($forceTocHere) {
2614                         $mw =& MagicWord::get( MAG_TOC );
2615                         return $mw->replace( $toc, $full );
2616                 } else {
2617                         return $full;
2618                 }
2619         }
2620
2621         /**
2622          * Return an HTML link for the "ISBN 123456" text
2623          * @access private
2624          */
2625         function magicISBN( $text ) {
2626                 $fname = 'Parser::magicISBN';
2627                 wfProfileIn( $fname );
2628
2629                 $a = split( 'ISBN ', ' '.$text );
2630                 if ( count ( $a ) < 2 ) {
2631                         wfProfileOut( $fname );
2632                         return $text;
2633                 }
2634                 $text = substr( array_shift( $a ), 1);
2635                 $valid = '0123456789-Xx';
2636
2637                 foreach ( $a as $x ) {
2638                         $isbn = $blank = '' ;
2639                         while ( ' ' == $x{0} ) {
2640                                 $blank .= ' ';
2641                                 $x = substr( $x, 1 );
2642                         }
2643                         if ( $x == '' ) { # blank isbn
2644                                 $text .= "ISBN $blank";
2645                                 continue;
2646                         }
2647                         while ( strstr( $valid, $x{0} ) != false ) {
2648                                 $isbn .= $x{0};
2649                                 $x = substr( $x, 1 );
2650                         }
2651                         $num = str_replace( '-', '', $isbn );
2652                         $num = str_replace( ' ', '', $num );
2653                         $num = str_replace( 'x', 'X', $num );
2654
2655                         if ( '' == $num ) {
2656                                 $text .= "ISBN $blank$x";
2657                         } else {
2658                                 $titleObj = Title::makeTitle( NS_SPECIAL, 'Booksources' );
2659                                 $text .= '<a href="' .
2660                                 $titleObj->escapeLocalUrl( 'isbn='.$num ) .
2661                                         "\" class=\"internal\">ISBN $isbn</a>";
2662                                 $text .= $x;
2663                         }
2664                 }
2665                 wfProfileOut( $fname );
2666                 return $text;
2667         }
2668
2669         /**
2670          * Return an HTML link for the "RFC 1234" text
2671          *
2672          * @access private
2673          * @param string $text     Text to be processed
2674          * @param string $keyword  Magic keyword to use (default RFC)
2675          * @param string $urlmsg   Interface message to use (default rfcurl)
2676          * @return string
2677          */
2678         function magicRFC( $text, $keyword='RFC ', $urlmsg='rfcurl'  ) {
2679
2680                 $valid = '0123456789';
2681                 $internal = false;
2682
2683                 $a = split( $keyword, ' '.$text );
2684                 if ( count ( $a ) < 2 ) {
2685                         return $text;
2686                 }
2687                 $text = substr( array_shift( $a ), 1);
2688
2689                 /* Check if keyword is preceed by [[.
2690                  * This test is made here cause of the array_shift above
2691                  * that prevent the test to be done in the foreach.
2692                  */
2693                 if ( substr( $text, -2 ) == '[[' ) {
2694                         $internal = true;
2695                 }
2696
2697                 foreach ( $a as $x ) {
2698                         /* token might be empty if we have RFC RFC 1234 */
2699                         if ( $x=='' ) {
2700                                 $text.=$keyword;
2701                                 continue;
2702                                 }
2703
2704                         $id = $blank = '' ;
2705
2706                         /** remove and save whitespaces in $blank */
2707                         while ( $x{0} == ' ' ) {
2708                                 $blank .= ' ';
2709                                 $x = substr( $x, 1 );
2710                         }
2711
2712                         /** remove and save the rfc number in $id */
2713                         while ( strstr( $valid, $x{0} ) != false ) {
2714                                 $id .= $x{0};
2715                                 $x = substr( $x, 1 );
2716                         }
2717
2718                         if ( $id == '' ) {
2719                                 /* call back stripped spaces*/
2720                                 $text .= $keyword.$blank.$x;
2721                         } elseif( $internal ) {
2722                                 /* normal link */
2723                                 $text .= $keyword.$id.$x;
2724                         } else {
2725                                 /* build the external link*/
2726                                 $url = wfMsg( $urlmsg, $id);
2727                                 $sk =& $this->mOptions->getSkin();
2728                                 $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
2729                                 $text .= "<a href='{$url}'{$la}>{$keyword}{$id}</a>{$x}";
2730                         }
2731
2732                         /* Check if the next RFC keyword is preceed by [[ */
2733                         $internal = ( substr($x,-2) == '[[' );
2734                 }
2735                 return $text;
2736         }
2737
2738         /**
2739          * Transform wiki markup when saving a page by doing \r\n -> \n
2740          * conversion, substitting signatures, {{subst:}} templates, etc.
2741          *
2742          * @param string $text the text to transform
2743          * @param Title &$title the Title object for the current article
2744          * @param User &$user the User object describing the current user
2745          * @param ParserOptions $options parsing options
2746          * @param bool $clearState whether to clear the parser state first
2747          * @return string the altered wiki markup
2748          * @access public
2749          */
2750         function preSaveTransform( $text, &$title, &$user, $options, $clearState = true ) {
2751                 $this->mOptions = $options;
2752                 $this->mTitle =& $title;
2753                 $this->mOutputType = OT_WIKI;
2754
2755                 if ( $clearState ) {
2756                         $this->clearState();
2757                 }
2758
2759                 $stripState = false;
2760                 $pairs = array(
2761                         "\r\n" => "\n",
2762                 );
2763                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
2764                 $text = $this->strip( $text, $stripState, false );
2765                 $text = $this->pstPass2( $text, $user );
2766                 $text = $this->unstrip( $text, $stripState );
2767                 $text = $this->unstripNoWiki( $text, $stripState );
2768                 return $text;
2769         }
2770
2771         /**
2772          * Pre-save transform helper function
2773          * @access private
2774          */
2775         function pstPass2( $text, &$user ) {
2776                 global $wgContLang, $wgLocaltimezone;
2777
2778                 # Variable replacement
2779                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
2780                 $text = $this->replaceVariables( $text );
2781
2782                 # Signatures
2783                 #
2784                 $n = $user->getName();
2785                 $k = $user->getOption( 'nickname' );
2786                 if ( '' == $k ) { $k = $n; }
2787                 if ( isset( $wgLocaltimezone ) ) {
2788                         $oldtz = getenv( 'TZ' );
2789                         putenv( 'TZ='.$wgLocaltimezone );
2790                 }
2791
2792                 /* Note: This is the timestamp saved as hardcoded wikitext to
2793                  * the database, we use $wgContLang here in order to give
2794                  * everyone the same signiture and use the default one rather
2795                  * than the one selected in each users preferences.
2796                  */
2797                 $d = $wgContLang->timeanddate( wfTimestampNow(), false, false) .
2798                   ' (' . date( 'T' ) . ')';
2799                 if ( isset( $wgLocaltimezone ) ) {
2800                         putenv( 'TZ='.$oldtz );
2801                 }
2802
2803                 if( $user->getOption( 'fancysig' ) ) {
2804                         $sigText = $k;
2805                 } else {
2806                         $sigText = '[[' . $wgContLang->getNsText( NS_USER ) . ":$n|$k]]";
2807                 }
2808                 $text = preg_replace( '/~~~~~/', $d, $text );
2809                 $text = preg_replace( '/~~~~/', "$sigText $d", $text );
2810                 $text = preg_replace( '/~~~/', $sigText, $text );
2811
2812                 # Context links: [[|name]] and [[name (context)|]]
2813                 #
2814                 $tc = "[&;%\\-,.\\(\\)' _0-9A-Za-z\\/:\\x80-\\xff]";
2815                 $np = "[&;%\\-,.' _0-9A-Za-z\\/:\\x80-\\xff]"; # No parens
2816                 $namespacechar = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
2817                 $conpat = "/^({$np}+) \\(({$tc}+)\\)$/";
2818
2819                 $p1 = "/\[\[({$np}+) \\(({$np}+)\\)\\|]]/";             # [[page (context)|]]
2820                 $p2 = "/\[\[\\|({$tc}+)]]/";                                    # [[|page]]
2821                 $p3 = "/\[\[(:*$namespacechar+):({$np}+)\\|]]/";                # [[namespace:page|]] and [[:namespace:page|]]
2822                 $p4 = "/\[\[(:*$namespacechar+):({$np}+) \\(({$np}+)\\)\\|]]/"; # [[ns:page (cont)|]] and [[:ns:page (cont)|]]
2823                 $context = '';
2824                 $t = $this->mTitle->getText();
2825                 if ( preg_match( $conpat, $t, $m ) ) {
2826                         $context = $m[2];
2827                 }
2828                 $text = preg_replace( $p4, '[[\\1:\\2 (\\3)|\\2]]', $text );
2829                 $text = preg_replace( $p1, '[[\\1 (\\2)|\\1]]', $text );
2830                 $text = preg_replace( $p3, '[[\\1:\\2|\\2]]', $text );
2831
2832                 if ( '' == $context ) {
2833                         $text = preg_replace( $p2, '[[\\1]]', $text );
2834                 } else {
2835                         $text = preg_replace( $p2, "[[\\1 ({$context})|\\1]]", $text );
2836                 }
2837
2838                 # Trim trailing whitespace
2839                 # MAG_END (__END__) tag allows for trailing
2840                 # whitespace to be deliberately included
2841                 $text = rtrim( $text );
2842                 $mw =& MagicWord::get( MAG_END );
2843                 $mw->matchAndRemove( $text );
2844
2845                 return $text;
2846         }
2847
2848         /**
2849          * Set up some variables which are usually set up in parse()
2850          * so that an external function can call some class members with confidence
2851          * @access public
2852          */
2853         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
2854                 $this->mTitle =& $title;
2855                 $this->mOptions = $options;
2856                 $this->mOutputType = $outputType;
2857                 if ( $clearState ) {
2858                         $this->clearState();
2859                 }
2860         }
2861
2862         /**
2863          * Transform a MediaWiki message by replacing magic variables.
2864          *
2865          * @param string $text the text to transform
2866          * @param ParserOptions $options  options
2867          * @return string the text with variables substituted
2868          * @access public
2869          */
2870         function transformMsg( $text, $options ) {
2871                 global $wgTitle;
2872                 static $executing = false;
2873
2874                 # Guard against infinite recursion
2875                 if ( $executing ) {
2876                         return $text;
2877                 }
2878                 $executing = true;
2879
2880                 $this->mTitle = $wgTitle;
2881                 $this->mOptions = $options;
2882                 $this->mOutputType = OT_MSG;
2883                 $this->clearState();
2884                 $text = $this->replaceVariables( $text );
2885
2886                 $executing = false;
2887                 return $text;
2888         }
2889
2890         /**
2891          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
2892          * Callback will be called with the text within
2893          * Transform and return the text within
2894          * @access public
2895          */
2896         function setHook( $tag, $callback ) {
2897                 $oldVal = @$this->mTagHooks[$tag];
2898                 $this->mTagHooks[$tag] = $callback;
2899                 return $oldVal;
2900         }
2901
2902         /**
2903          * Replace <!--LINK--> link placeholders with actual links, in the buffer
2904          * Placeholders created in Skin::makeLinkObj()
2905          * Returns an array of links found, indexed by PDBK:
2906          *  0 - broken
2907          *  1 - normal link
2908          *  2 - stub
2909          * $options is a bit field, RLH_FOR_UPDATE to select for update
2910          */
2911         function replaceLinkHolders( &$text, $options = 0 ) {
2912                 global $wgUser, $wgLinkCache;
2913                 global $wgOutputReplace;
2914
2915                 $fname = 'Parser::replaceLinkHolders';
2916                 wfProfileIn( $fname );
2917
2918                 $pdbks = array();
2919                 $colours = array();
2920                 $sk = $this->mOptions->getSkin();
2921
2922                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
2923                         wfProfileIn( $fname.'-check' );
2924                         $dbr =& wfGetDB( DB_SLAVE );
2925                         $page = $dbr->tableName( 'page' );
2926                         $threshold = $wgUser->getOption('stubthreshold');
2927
2928                         # Sort by namespace
2929                         asort( $this->mLinkHolders['namespaces'] );
2930
2931                         # Generate query
2932                         $query = false;
2933                         foreach ( $this->mLinkHolders['namespaces'] as $key => $val ) {
2934                                 # Make title object
2935                                 $title = $this->mLinkHolders['titles'][$key];
2936
2937                                 # Skip invalid entries.
2938                                 # Result will be ugly, but prevents crash.
2939                                 if ( is_null( $title ) ) {
2940                                         continue;
2941                                 }
2942                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
2943
2944                                 # Check if it's in the link cache already
2945                                 if ( $wgLinkCache->getGoodLinkID( $pdbk ) ) {
2946                                         $colours[$pdbk] = 1;
2947                                 } elseif ( $wgLinkCache->isBadLink( $pdbk ) ) {
2948                                         $colours[$pdbk] = 0;
2949                                 } else {
2950                                         # Not in the link cache, add it to the query
2951                                         if ( !isset( $current ) ) {
2952                                                 $current = $val;
2953                                                 $query =  "SELECT page_id, page_namespace, page_title";
2954                                                 if ( $threshold > 0 ) {
2955                                                         $query .= ', page_len, page_is_redirect';
2956                                                 }
2957                                                 $query .= " FROM $page WHERE (page_namespace=$val AND page_title IN(";
2958                                         } elseif ( $current != $val ) {
2959                                                 $current = $val;
2960                                                 $query .= ")) OR (page_namespace=$val AND page_title IN(";
2961                                         } else {
2962                                                 $query .= ', ';
2963                                         }
2964
2965                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
2966                                 }
2967                         }
2968                         if ( $query ) {
2969                                 $query .= '))';
2970                                 if ( $options & RLH_FOR_UPDATE ) {
2971                                         $query .= ' FOR UPDATE';
2972                                 }
2973
2974                                 $res = $dbr->query( $query, $fname );
2975
2976                                 # Fetch data and form into an associative array
2977                                 # non-existent = broken
2978                                 # 1 = known
2979                                 # 2 = stub
2980                                 while ( $s = $dbr->fetchObject($res) ) {
2981                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
2982                                         $pdbk = $title->getPrefixedDBkey();
2983                                         $wgLinkCache->addGoodLinkObj( $s->page_id, $title );
2984
2985                                         if ( $threshold >  0 ) {
2986                                                 $size = $s->page_len;
2987                                                 if ( $s->page_is_redirect || $s->page_namespace != 0 || $size >= $threshold ) {
2988                                                         $colours[$pdbk] = 1;
2989                                                 } else {
2990                                                         $colours[$pdbk] = 2;
2991                                                 }
2992                                         } else {
2993                                                 $colours[$pdbk] = 1;
2994                                         }
2995                                 }
2996                         }
2997                         wfProfileOut( $fname.'-check' );
2998
2999                         # Construct search and replace arrays
3000                         wfProfileIn( $fname.'-construct' );
3001                         $wgOutputReplace = array();
3002                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3003                                 $pdbk = $pdbks[$key];
3004                                 $searchkey = "<!--LINK $key-->";
3005                                 $title = $this->mLinkHolders['titles'][$key];
3006                                 if ( empty( $colours[$pdbk] ) ) {
3007                                         $wgLinkCache->addBadLinkObj( $title );
3008                                         $colours[$pdbk] = 0;
3009                                         $wgOutputReplace[$searchkey] = $sk->makeBrokenLinkObj( $title,
3010                                                                         $this->mLinkHolders['texts'][$key],
3011                                                                         $this->mLinkHolders['queries'][$key] );
3012                                 } elseif ( $colours[$pdbk] == 1 ) {
3013                                         $wgOutputReplace[$searchkey] = $sk->makeKnownLinkObj( $title,
3014                                                                         $this->mLinkHolders['texts'][$key],
3015                                                                         $this->mLinkHolders['queries'][$key] );
3016                                 } elseif ( $colours[$pdbk] == 2 ) {
3017                                         $wgOutputReplace[$searchkey] = $sk->makeStubLinkObj( $title,
3018                                                                         $this->mLinkHolders['texts'][$key],
3019                                                                         $this->mLinkHolders['queries'][$key] );
3020                                 }
3021                         }
3022                         wfProfileOut( $fname.'-construct' );
3023
3024                         # Do the thing
3025                         wfProfileIn( $fname.'-replace' );
3026
3027                         $text = preg_replace_callback(
3028                                 '/(<!--LINK .*?-->)/',
3029                                 "wfOutputReplaceMatches",
3030                                 $text);
3031
3032                         wfProfileOut( $fname.'-replace' );
3033                 }
3034
3035                 # Now process interwiki link holders
3036                 # This is quite a bit simpler than internal links
3037                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
3038                         wfProfileIn( $fname.'-interwiki' );
3039                         # Make interwiki link HTML
3040                         $wgOutputReplace = array();
3041                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
3042                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
3043                                 $wgOutputReplace[$key] = $sk->makeLinkObj( $title, $link );
3044                         }
3045
3046                         $text = preg_replace_callback(
3047                                 '/<!--IWLINK (.*?)-->/',
3048                                 "wfOutputReplaceMatches",
3049                                 $text );
3050                         wfProfileOut( $fname.'-interwiki' );
3051                 }
3052
3053                 wfProfileOut( $fname );
3054                 return $colours;
3055         }
3056
3057         /**
3058          * Replace <!--LINK--> link placeholders with plain text of links
3059          * (not HTML-formatted).
3060          * @param string $text
3061          * @return string
3062          */
3063         function replaceLinkHoldersText( $text ) {
3064                 global $wgUser, $wgLinkCache;
3065                 global $wgOutputReplace;
3066
3067                 $fname = 'Parser::replaceLinkHoldersText';
3068                 wfProfileIn( $fname );
3069
3070                 $text = preg_replace_callback(
3071                         '/<!--(LINK|IWLINK) (.*?)-->/',
3072                         array( &$this, 'replaceLinkHoldersTextCallback' ),
3073                         $text );
3074
3075                 wfProfileOut( $fname );
3076                 return $text;
3077         }
3078
3079         /**
3080          * @param array $matches
3081          * @return string
3082          * @access private
3083          */
3084         function replaceLinkHoldersTextCallback( $matches ) {
3085                 $type = $matches[1];
3086                 $key  = $matches[2];
3087                 if( $type == 'LINK' ) {
3088                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
3089                                 return $this->mLinkHolders['texts'][$key];
3090                         }
3091                 } elseif( $type == 'IWLINK' ) {
3092                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
3093                                 return $this->mInterwikiLinkHolders['texts'][$key];
3094                         }
3095                 }
3096                 return $matches[0];
3097         }
3098
3099         /**
3100          * Renders an image gallery from a text with one line per image.
3101          * text labels may be given by using |-style alternative text. E.g.
3102          *   Image:one.jpg|The number "1"
3103          *   Image:tree.jpg|A tree
3104          * given as text will return the HTML of a gallery with two images,
3105          * labeled 'The number "1"' and
3106          * 'A tree'.
3107          *
3108          * @static
3109          */
3110         function renderImageGallery( $text ) {
3111                 # Setup the parser
3112                 global $wgUser, $wgTitle;
3113                 $parserOptions = ParserOptions::newFromUser( $wgUser );
3114                 $localParser = new Parser();
3115
3116                 global $wgLinkCache;
3117                 $ig = new ImageGallery();
3118                 $ig->setShowBytes( false );
3119                 $ig->setShowFilename( false );
3120                 $lines = explode( "\n", $text );
3121
3122                 foreach ( $lines as $line ) {
3123                         # match lines like these:
3124                         # Image:someimage.jpg|This is some image
3125                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
3126                         # Skip empty lines
3127                         if ( count( $matches ) == 0 ) {
3128                                 continue;
3129                         }
3130                         $nt = Title::newFromURL( $matches[1] );
3131                         if( is_null( $nt ) ) {
3132                                 # Bogus title. Ignore these so we don't bomb out later.
3133                                 continue;
3134                         }
3135                         if ( isset( $matches[3] ) ) {
3136                                 $label = $matches[3];
3137                         } else {
3138                                 $label = '';
3139                         }
3140
3141                         $html = $localParser->parse( $label , $wgTitle, $parserOptions );
3142                         $html = $html->mText;
3143
3144                         $ig->add( new Image( $nt ), $html );
3145                         $wgLinkCache->addImageLinkObj( $nt );
3146                 }
3147                 return $ig->toHTML();
3148         }
3149
3150         /**
3151          * Parse image options text and use it to make an image
3152          */
3153         function makeImage( &$nt, $options ) {
3154                 global $wgContLang, $wgUseImageResize;
3155                 global $wgUser, $wgThumbLimits;
3156
3157                 $align = '';
3158
3159                 # Check if the options text is of the form "options|alt text"
3160                 # Options are:
3161                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
3162                 #  * left               no resizing, just left align. label is used for alt= only
3163                 #  * right              same, but right aligned
3164                 #  * none               same, but not aligned
3165                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
3166                 #  * center             center the image
3167                 #  * framed             Keep original image size, no magnify-button.
3168
3169                 $part = explode( '|', $options);
3170
3171                 $mwThumb  =& MagicWord::get( MAG_IMG_THUMBNAIL );
3172                 $mwLeft   =& MagicWord::get( MAG_IMG_LEFT );
3173                 $mwRight  =& MagicWord::get( MAG_IMG_RIGHT );
3174                 $mwNone   =& MagicWord::get( MAG_IMG_NONE );
3175                 $mwWidth  =& MagicWord::get( MAG_IMG_WIDTH );
3176                 $mwCenter =& MagicWord::get( MAG_IMG_CENTER );
3177                 $mwFramed =& MagicWord::get( MAG_IMG_FRAMED );
3178                 $caption = '';
3179
3180                 $width = $height = $framed = $thumb = false;
3181                 $manual_thumb = '' ;
3182
3183                 foreach( $part as $key => $val ) {
3184                         $val_parts = explode ( '=' , $val , 2 ) ;
3185                         $left_part = array_shift ( $val_parts ) ;
3186                         if ( $wgUseImageResize && ! is_null( $mwThumb->matchVariableStartToEnd($val) ) ) {
3187                                 $thumb=true;
3188                         } elseif ( $wgUseImageResize && count ( $val_parts ) == 1 && ! is_null( $mwThumb->matchVariableStartToEnd($left_part) ) ) {
3189                                 # use manually specified thumbnail
3190                                 $thumb=true;
3191                                 $manual_thumb = array_shift ( $val_parts ) ;
3192                         } elseif ( ! is_null( $mwRight->matchVariableStartToEnd($val) ) ) {
3193                                 # remember to set an alignment, don't render immediately
3194                                 $align = 'right';
3195                         } elseif ( ! is_null( $mwLeft->matchVariableStartToEnd($val) ) ) {
3196                                 # remember to set an alignment, don't render immediately
3197                                 $align = 'left';
3198                         } elseif ( ! is_null( $mwCenter->matchVariableStartToEnd($val) ) ) {
3199                                 # remember to set an alignment, don't render immediately
3200                                 $align = 'center';
3201                         } elseif ( ! is_null( $mwNone->matchVariableStartToEnd($val) ) ) {
3202                                 # remember to set an alignment, don't render immediately
3203                                 $align = 'none';
3204                         } elseif ( $wgUseImageResize && ! is_null( $match = $mwWidth->matchVariableStartToEnd($val) ) ) {
3205                                 wfDebug( "MAG_IMG_WIDTH match: $match\n" );
3206                                 # $match is the image width in pixels
3207                                 if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $match, $m ) ) {
3208                                         $width = intval( $m[1] );
3209                                         $height = intval( $m[2] );
3210                                 } else {
3211                                         $width = intval($match);
3212                                 }
3213                         } elseif ( ! is_null( $mwFramed->matchVariableStartToEnd($val) ) ) {
3214                                 $framed=true;
3215                         } else {
3216                                 $caption = $val;
3217                         }
3218                 }
3219                 # Strip bad stuff out of the alt text
3220                 $alt = $this->replaceLinkHoldersText( $caption );
3221                 $alt = Sanitizer::stripAllTags( $alt );
3222
3223                 # Linker does the rest
3224                 $sk =& $this->mOptions->getSkin();
3225                 return $sk->makeImageLinkObj( $nt, $caption, $alt, $align, $width, $height, $framed, $thumb, $manual_thumb );
3226         }
3227 }
3228
3229 /**
3230  * @todo document
3231  * @package MediaWiki
3232  */
3233 class ParserOutput
3234 {
3235         var $mText, $mLanguageLinks, $mCategoryLinks, $mContainsOldMagic;
3236         var $mCacheTime; # Timestamp on this article, or -1 for uncacheable. Used in ParserCache.
3237         var $mVersion;   # Compatibility check
3238         var $mTitleText; # title text of the chosen language variant
3239
3240         function ParserOutput( $text = '', $languageLinks = array(), $categoryLinks = array(),
3241                 $containsOldMagic = false, $titletext = '' )
3242         {
3243                 $this->mText = $text;
3244                 $this->mLanguageLinks = $languageLinks;
3245                 $this->mCategoryLinks = $categoryLinks;
3246                 $this->mContainsOldMagic = $containsOldMagic;
3247                 $this->mCacheTime = '';
3248                 $this->mVersion = MW_PARSER_VERSION;
3249                 $this->mTitleText = $titletext;
3250         }
3251
3252         function getText()                   { return $this->mText; }
3253         function getLanguageLinks()          { return $this->mLanguageLinks; }
3254         function getCategoryLinks()          { return array_keys( $this->mCategoryLinks ); }
3255         function getCacheTime()              { return $this->mCacheTime; }
3256         function getTitleText()              { return $this->mTitleText; }
3257         function containsOldMagic()          { return $this->mContainsOldMagic; }
3258         function setText( $text )            { return wfSetVar( $this->mText, $text ); }
3259         function setLanguageLinks( $ll )     { return wfSetVar( $this->mLanguageLinks, $ll ); }
3260         function setCategoryLinks( $cl )     { return wfSetVar( $this->mCategoryLinks, $cl ); }
3261         function setContainsOldMagic( $com ) { return wfSetVar( $this->mContainsOldMagic, $com ); }
3262         function setCacheTime( $t )          { return wfSetVar( $this->mCacheTime, $t ); }
3263         function setTitleText( $t )          { return wfSetVar ($this->mTitleText, $t); }
3264
3265         function addCategoryLink( $c )       { $this->mCategoryLinks[$c] = 1; }
3266
3267         function merge( $other ) {
3268                 $this->mLanguageLinks = array_merge( $this->mLanguageLinks, $other->mLanguageLinks );
3269                 $this->mCategoryLinks = array_merge( $this->mCategoryLinks, $this->mLanguageLinks );
3270                 $this->mContainsOldMagic = $this->mContainsOldMagic || $other->mContainsOldMagic;
3271         }
3272
3273         /**
3274          * Return true if this cached output object predates the global or
3275          * per-article cache invalidation timestamps, or if it comes from
3276          * an incompatible older version.
3277          *
3278          * @param string $touched the affected article's last touched timestamp
3279          * @return bool
3280          * @access public
3281          */
3282         function expired( $touched ) {
3283                 global $wgCacheEpoch;
3284                 return $this->getCacheTime() == -1 || // parser says it's uncacheable
3285                        $this->getCacheTime() <= $touched ||
3286                        $this->getCacheTime() <= $wgCacheEpoch ||
3287                        !isset( $this->mVersion ) ||
3288                        version_compare( $this->mVersion, MW_PARSER_VERSION, "lt" );
3289         }
3290 }
3291
3292 /**
3293  * Set options of the Parser
3294  * @todo document
3295  * @package MediaWiki
3296  */
3297 class ParserOptions
3298 {
3299         # All variables are private
3300         var $mUseTeX;                    # Use texvc to expand <math> tags
3301         var $mUseDynamicDates;           # Use DateFormatter to format dates
3302         var $mInterwikiMagic;            # Interlanguage links are removed and returned in an array
3303         var $mAllowExternalImages;       # Allow external images inline
3304         var $mSkin;                      # Reference to the preferred skin
3305         var $mDateFormat;                # Date format index
3306         var $mEditSection;               # Create "edit section" links
3307         var $mNumberHeadings;            # Automatically number headings
3308         var $mAllowSpecialInclusion;     # Allow inclusion of special pages
3309
3310         function getUseTeX()                        { return $this->mUseTeX; }
3311         function getUseDynamicDates()               { return $this->mUseDynamicDates; }
3312         function getInterwikiMagic()                { return $this->mInterwikiMagic; }
3313         function getAllowExternalImages()           { return $this->mAllowExternalImages; }
3314         function getSkin()                          { return $this->mSkin; }
3315         function getDateFormat()                    { return $this->mDateFormat; }
3316         function getEditSection()                   { return $this->mEditSection; }
3317         function getNumberHeadings()                { return $this->mNumberHeadings; }
3318         function getAllowSpecialInclusion()         { return $this->mAllowSpecialInclusion; }
3319
3320
3321         function setUseTeX( $x )                    { return wfSetVar( $this->mUseTeX, $x ); }
3322         function setUseDynamicDates( $x )           { return wfSetVar( $this->mUseDynamicDates, $x ); }
3323         function setInterwikiMagic( $x )            { return wfSetVar( $this->mInterwikiMagic, $x ); }
3324         function setAllowExternalImages( $x )       { return wfSetVar( $this->mAllowExternalImages, $x ); }
3325         function setDateFormat( $x )                { return wfSetVar( $this->mDateFormat, $x ); }
3326         function setEditSection( $x )               { return wfSetVar( $this->mEditSection, $x ); }
3327         function setNumberHeadings( $x )            { return wfSetVar( $this->mNumberHeadings, $x ); }
3328         function setAllowSpecialInclusion( $x )     { return wfSetVar( $this->mAllowSpecialInclusion, $x ); }
3329
3330         function setSkin( &$x ) { $this->mSkin =& $x; }
3331
3332         function ParserOptions() {
3333                 global $wgUser;
3334                 $this->initialiseFromUser( $wgUser );
3335         }
3336
3337         /**
3338          * Get parser options
3339          * @static
3340          */
3341         function newFromUser( &$user ) {
3342                 $popts = new ParserOptions;
3343                 $popts->initialiseFromUser( $user );
3344                 return $popts;
3345         }
3346
3347         /** Get user options */
3348         function initialiseFromUser( &$userInput ) {
3349                 global $wgUseTeX, $wgUseDynamicDates, $wgInterwikiMagic, $wgAllowExternalImages,
3350                        $wgAllowSpecialInclusion;
3351                 $fname = 'ParserOptions::initialiseFromUser';
3352                 wfProfileIn( $fname );
3353                 if ( !$userInput ) {
3354                         $user = new User;
3355                         $user->setLoaded( true );
3356                 } else {
3357                         $user =& $userInput;
3358                 }
3359
3360                 $this->mUseTeX = $wgUseTeX;
3361                 $this->mUseDynamicDates = $wgUseDynamicDates;
3362                 $this->mInterwikiMagic = $wgInterwikiMagic;
3363                 $this->mAllowExternalImages = $wgAllowExternalImages;
3364                 wfProfileIn( $fname.'-skin' );
3365                 $this->mSkin =& $user->getSkin();
3366                 wfProfileOut( $fname.'-skin' );
3367                 $this->mDateFormat = $user->getOption( 'date' );
3368                 $this->mEditSection = $user->getOption( 'editsection' );
3369                 $this->mNumberHeadings = $user->getOption( 'numberheadings' );
3370                 $this->mAllowSpecialInclusion = $wgAllowSpecialInclusion;
3371                 wfProfileOut( $fname );
3372         }
3373 }
3374
3375 /**
3376  * Callback function used by Parser::replaceLinkHolders()
3377  * to substitute link placeholders.
3378  */
3379 function &wfOutputReplaceMatches( $matches ) {
3380         global $wgOutputReplace;
3381         return $wgOutputReplace[$matches[1]];
3382 }
3383
3384 /**
3385  * Return the total number of articles
3386  */
3387 function wfNumberOfArticles() {
3388         global $wgNumberOfArticles;
3389
3390         wfLoadSiteStats();
3391         return $wgNumberOfArticles;
3392 }
3393
3394 /**
3395  * Return the number of files
3396  */
3397 function wfNumberOfFiles() {
3398         $fname = 'Parser::wfNumberOfFiles';
3399
3400         wfProfileIn( $fname );
3401         $dbr =& wfGetDB( DB_SLAVE );
3402         $res = $dbr->selectField('image', 'COUNT(*)', array(), $fname );
3403         wfProfileOut( $fname );
3404
3405         return $res;
3406 }
3407
3408 /**
3409  * Get various statistics from the database
3410  * @private
3411  */
3412 function wfLoadSiteStats() {
3413         global $wgNumberOfArticles, $wgTotalViews, $wgTotalEdits;
3414         $fname = 'wfLoadSiteStats';
3415
3416         if ( -1 != $wgNumberOfArticles ) return;
3417         $dbr =& wfGetDB( DB_SLAVE );
3418         $s = $dbr->selectRow( 'site_stats',
3419                 array( 'ss_total_views', 'ss_total_edits', 'ss_good_articles' ),
3420                 array( 'ss_row_id' => 1 ), $fname
3421         );
3422
3423         if ( $s === false ) {
3424                 return;
3425         } else {
3426                 $wgTotalViews = $s->ss_total_views;
3427                 $wgTotalEdits = $s->ss_total_edits;
3428                 $wgNumberOfArticles = $s->ss_good_articles;
3429         }
3430 }
3431
3432 /**
3433  * Escape html tags
3434  * Basicly replacing " > and < with HTML entities ( &quot;, &gt;, &lt;)
3435  *
3436  * @param string $in Text that might contain HTML tags
3437  * @return string Escaped string
3438  */
3439 function wfEscapeHTMLTagsOnly( $in ) {
3440         return str_replace(
3441                 array( '"', '>', '<' ),
3442                 array( '&quot;', '&gt;', '&lt;' ),
3443                 $in );
3444 }
3445
3446 ?>