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 require_once( 'HttpFunctions.php' );
  12
  13 /**
  14  * Update this version number when the ParserOutput format
  15  * changes in an incompatible way, so the parser cache
  16  * can automatically discard old data.
  17  */
  18 define( 'MW_PARSER_VERSION', '1.5.0' );
  19
  20 /**
  21  * Variable substitution O(N^2) attack
  22  *
  23  * Without countermeasures, it would be possible to attack the parser by saving
  24  * a page filled with a large number of inclusions of large pages. The size of
  25  * the generated page would be proportional to the square of the input size.
  26  * Hence, we limit the number of inclusions of any given page, thus bringing any
  27  * attack back to O(N).
  28  */
  29
  30 define( 'MAX_INCLUDE_REPEAT', 100 );
  31 define( 'MAX_INCLUDE_SIZE', 1000000 ); // 1 Million
  32
  33 define( 'RLH_FOR_UPDATE', 1 );
  34
  35 # Allowed values for $mOutputType
  36 define( 'OT_HTML', 1 );
  37 define( 'OT_WIKI', 2 );
  38 define( 'OT_MSG' , 3 );
  39
  40 # string parameter for extractTags which will cause it
  41 # to strip HTML comments in addition to regular
  42 # <XML>-style tags. This should not be anything we
  43 # may want to use in wikisyntax
  44 define( 'STRIP_COMMENTS', 'HTMLCommentStrip' );
  45
  46 # Constants needed for external link processing
  47 define( 'HTTP_PROTOCOLS', 'http:\/\/|https:\/\/' );
  48 # Everything except bracket, space, or control characters
  49 define( 'EXT_LINK_URL_CLASS', '[^][<>"\\x00-\\x20\\x7F]' );
  50 # Including space
  51 define( 'EXT_LINK_TEXT_CLASS', '[^\]\\x00-\\x1F\\x7F]' );
  52 define( 'EXT_IMAGE_FNAME_CLASS', '[A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]' );
  53 define( 'EXT_IMAGE_EXTENSIONS', 'gif|png|jpg|jpeg' );
  54 define( 'EXT_LINK_BRACKETED',  '/\[(\b(' . wfUrlProtocols() . ')'.EXT_LINK_URL_CLASS.'+) *('.EXT_LINK_TEXT_CLASS.'*?)\]/S' );
  55 define( 'EXT_IMAGE_REGEX',
  56         '/^('.HTTP_PROTOCOLS.')'.  # Protocol
  57         '('.EXT_LINK_URL_CLASS.'+)\\/'.  # Hostname and path
  58         '('.EXT_IMAGE_FNAME_CLASS.'+)\\.((?i)'.EXT_IMAGE_EXTENSIONS.')$/S' # Filename
  59 );
  60
  61 /**
  62  * PHP Parser
  63  *
  64  * Processes wiki markup
  65  *
  66  * <pre>
  67  * There are three main entry points into the Parser class:
  68  * parse()
  69  *   produces HTML output
  70  * preSaveTransform().
  71  *   produces altered wiki markup.
  72  * transformMsg()
  73  *   performs brace substitution on MediaWiki messages
  74  *
  75  * Globals used:
  76  *    objects:   $wgLang, $wgLinkCache
  77  *
  78  * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
  79  *
  80  * settings:
  81  *  $wgUseTex*, $wgUseDynamicDates*, $wgInterwikiMagic*,
  82  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  83  *  $wgLocaltimezone, $wgAllowSpecialInclusion*
  84  *
  85  *  * only within ParserOptions
  86  * </pre>
  87  *
  88  * @package MediaWiki
  89  */
  90 class Parser
  91 {
  92         /**#@+
  93          * @access private
  94          */
  95         # Persistent:
  96         var $mTagHooks;
  97
  98         # Cleared with clearState():
  99         var $mOutput, $mAutonumber, $mDTopen, $mStripState = array();
 100         var $mVariables, $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 101         var $mInterwikiLinkHolders, $mLinkHolders, $mUniqPrefix;
 102
 103         # Temporary:
 104         var $mOptions, $mTitle, $mOutputType,
 105             $mTemplates,        // cache of already loaded templates, avoids
 106                                 // multiple SQL queries for the same string
 107             $mTemplatePath;     // stores an unsorted hash of all the templates already loaded
 108                                 // in this path. Used for loop detection.
 109
 110         var $mIWTransData = array();
 111
 112         /**#@-*/
 113
 114         /**
 115          * Constructor
 116          *
 117          * @access public
 118          */
 119         function Parser() {
 120                 $this->mTemplates = array();
 121                 $this->mTemplatePath = array();
 122                 $this->mTagHooks = array();
 123                 $this->clearState();
 124         }
 125
 126         /**
 127          * Clear Parser state
 128          *
 129          * @access private
 130          */
 131         function clearState() {
 132                 $this->mOutput = new ParserOutput;
 133                 $this->mAutonumber = 0;
 134                 $this->mLastSection = '';
 135                 $this->mDTopen = false;
 136                 $this->mVariables = false;
 137                 $this->mIncludeCount = array();
 138                 $this->mStripState = array();
 139                 $this->mArgStack = array();
 140                 $this->mInPre = false;
 141                 $this->mInterwikiLinkHolders = array(
 142                         'texts' => array(),
 143                         'titles' => array()
 144                 );
 145                 $this->mLinkHolders = array(
 146                         'namespaces' => array(),
 147                         'dbkeys' => array(),
 148                         'queries' => array(),
 149                         'texts' => array(),
 150                         'titles' => array()
 151                 );
 152                 $this->mRevisionId = null;
 153                 $this->mUniqPrefix = 'UNIQ' . Parser::getRandomString();
 154
 155                 wfRunHooks( 'ParserClearState', array( &$this ) );
 156         }
 157
 158         /**
 159          * First pass--just handle <nowiki> sections, pass the rest off
 160          * to internalParse() which does all the real work.
 161          *
 162          * @access private
 163          * @param string $text Text we want to parse
 164          * @param Title &$title A title object
 165          * @param array $options
 166          * @param boolean $linestart
 167          * @param boolean $clearState
 168          * @param int $revid number to pass in {{REVISIONID}}
 169          * @return ParserOutput a ParserOutput
 170          */
 171         function parse( $text, &$title, $options, $linestart = true, $clearState = true, $revid = null ) {
 172                 global $wgUseTidy, $wgContLang;
 173                 $fname = 'Parser::parse';
 174                 wfProfileIn( $fname );
 175
 176                 if ( $clearState ) {
 177                         $this->clearState();
 178                 }
 179
 180                 $this->mOptions = $options;
 181                 $this->mTitle =& $title;
 182                 $this->mRevisionId = $revid;
 183                 $this->mOutputType = OT_HTML;
 184
 185                 $this->mStripState = NULL;
 186
 187                 //$text = $this->strip( $text, $this->mStripState );
 188                 // VOODOO MAGIC FIX! Sometimes the above segfaults in PHP5.
 189                 $x =& $this->mStripState;
 190
 191                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$x ) );
 192                 $text = $this->strip( $text, $x );
 193                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$x ) );
 194
 195                 # Hook to suspend the parser in this state
 196                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$x ) ) )
 197                         return $text ;
 198
 199                 $text = $this->internalParse( $text );
 200
 201                 $text = $this->unstrip( $text, $this->mStripState );
 202
 203                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 204                 $fixtags = array(
 205                         # french spaces, last one Guillemet-left
 206                         # only if there is something before the space
 207                         '/(.) (?=\\?|:|;|!|\\302\\273)/' => '\\1&nbsp;\\2',
 208                         # french spaces, Guillemet-right
 209                         '/(\\302\\253) /' => '\\1&nbsp;',
 210                         '/<center *>(.*)<\\/center *>/i' => '<div class="center">\\1</div>',
 211                 );
 212                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 213
 214                 # only once and last
 215                 $text = $this->doBlockLevels( $text, $linestart );
 216
 217                 $this->replaceLinkHolders( $text );
 218
 219                 # the position of the convert() call should not be changed. it
 220                 # assumes that the links are all replaces and the only thing left
 221                 # is the <nowiki> mark.
 222                 $text = $wgContLang->convert($text);
 223                 $this->mOutput->setTitleText($wgContLang->getParsedTitle());
 224
 225                 $text = $this->unstripNoWiki( $text, $this->mStripState );
 226
 227                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 228
 229                 $text = Sanitizer::normalizeCharReferences( $text );
 230
 231                 if ($wgUseTidy) {
 232                         $text = Parser::tidy($text);
 233                 }
 234
 235                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 236
 237                 $this->mOutput->setText( $text );
 238                 wfProfileOut( $fname );
 239                 return $this->mOutput;
 240         }
 241
 242         /**
 243          * Get a random string
 244          *
 245          * @access private
 246          * @static
 247          */
 248         function getRandomString() {
 249                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 250         }
 251
 252         /**
 253          * Replaces all occurrences of <$tag>content</$tag> in the text
 254          * with a random marker and returns the new text. the output parameter
 255          * $content will be an associative array filled with data on the form
 256          * $unique_marker => content.
 257          *
 258          * If $content is already set, the additional entries will be appended
 259          * If $tag is set to STRIP_COMMENTS, the function will extract
 260          * <!-- HTML comments -->
 261          *
 262          * @access private
 263          * @static
 264          */
 265         function extractTagsAndParams($tag, $text, &$content, &$tags, &$params, $uniq_prefix = ''){
 266                 $rnd = $uniq_prefix . '-' . $tag . Parser::getRandomString();
 267                 if ( !$content ) {
 268                         $content = array( );
 269                 }
 270                 $n = 1;
 271                 $stripped = '';
 272
 273                 if ( !$tags ) {
 274                         $tags = array( );
 275                 }
 276
 277                 if ( !$params ) {
 278                         $params = array( );
 279                 }
 280
 281                 if( $tag == STRIP_COMMENTS ) {
 282                         $start = '/<!--()()/';
 283                         $end   = '/-->/';
 284                 } else {
 285                         $start = "/<$tag(\\s+[^\\/>]*|\\s*)(\\/?)>/i";
 286                         $end   = "/<\\/$tag\\s*>/i";
 287                 }
 288
 289                 while ( '' != $text ) {
 290                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 291                         $stripped .= $p[0];
 292                         if( count( $p ) < 4 ) {
 293                                 break;
 294                         }
 295                         $attributes = $p[1];
 296                         $empty      = $p[2];
 297                         $inside     = $p[3];
 298
 299                         $marker = $rnd . sprintf('%08X', $n++);
 300                         $stripped .= $marker;
 301
 302                         $tags[$marker] = "<$tag$attributes$empty>";
 303                         $params[$marker] = Sanitizer::decodeTagAttributes( $attributes );
 304
 305                         if ( $empty === '/' ) {
 306                                 // Empty element tag, <tag />
 307                                 $content[$marker] = null;
 308                                 $text = $inside;
 309                         } else {
 310                                 $q = preg_split( $end, $inside, 2 );
 311                                 $content[$marker] = $q[0];
 312                                 if( count( $q ) < 2 ) {
 313                                         # No end tag -- let it run out to the end of the text.
 314                                         break;
 315                                 } else {
 316                                         $text = $q[1];
 317                                 }
 318                         }
 319                 }
 320                 return $stripped;
 321         }
 322
 323         /**
 324          * Wrapper function for extractTagsAndParams
 325          * for cases where $tags and $params isn't needed
 326          * i.e. where tags will never have params, like <nowiki>
 327          *
 328          * @access private
 329          * @static
 330          */
 331         function extractTags( $tag, $text, &$content, $uniq_prefix = '' ) {
 332                 $dummy_tags = array();
 333                 $dummy_params = array();
 334
 335                 return Parser::extractTagsAndParams( $tag, $text, $content,
 336                         $dummy_tags, $dummy_params, $uniq_prefix );
 337         }
 338
 339         /**
 340          * Strips and renders nowiki, pre, math, hiero
 341          * If $render is set, performs necessary rendering operations on plugins
 342          * Returns the text, and fills an array with data needed in unstrip()
 343          * If the $state is already a valid strip state, it adds to the state
 344          *
 345          * @param bool $stripcomments when set, HTML comments <!-- like this -->
 346          *  will be stripped in addition to other tags. This is important
 347          *  for section editing, where these comments cause confusion when
 348          *  counting the sections in the wikisource
 349          *
 350          * @access private
 351          */
 352         function strip( $text, &$state, $stripcomments = false ) {
 353                 $render = ($this->mOutputType == OT_HTML);
 354                 $html_content = array();
 355                 $nowiki_content = array();
 356                 $math_content = array();
 357                 $pre_content = array();
 358                 $comment_content = array();
 359                 $ext_content = array();
 360                 $ext_tags = array();
 361                 $ext_params = array();
 362                 $gallery_content = array();
 363
 364                 # Replace any instances of the placeholders
 365                 $uniq_prefix = $this->mUniqPrefix;
 366                 #$text = str_replace( $uniq_prefix, wfHtmlEscapeFirst( $uniq_prefix ), $text );
 367
 368                 # html
 369                 global $wgRawHtml;
 370                 if( $wgRawHtml ) {
 371                         $text = Parser::extractTags('html', $text, $html_content, $uniq_prefix);
 372                         foreach( $html_content as $marker => $content ) {
 373                                 if ($render ) {
 374                                         # Raw and unchecked for validity.
 375                                         $html_content[$marker] = $content;
 376                                 } else {
 377                                         $html_content[$marker] = '<html>'.$content.'</html>';
 378                                 }
 379                         }
 380                 }
 381
 382                 # nowiki
 383                 $text = Parser::extractTags('nowiki', $text, $nowiki_content, $uniq_prefix);
 384                 foreach( $nowiki_content as $marker => $content ) {
 385                         if( $render ){
 386                                 $nowiki_content[$marker] = wfEscapeHTMLTagsOnly( $content );
 387                         } else {
 388                                 $nowiki_content[$marker] = '<nowiki>'.$content.'</nowiki>';
 389                         }
 390                 }
 391
 392                 # math
 393                 if( $this->mOptions->getUseTeX() ) {
 394                         $text = Parser::extractTags('math', $text, $math_content, $uniq_prefix);
 395                         foreach( $math_content as $marker => $content ){
 396                                 if( $render ) {
 397                                         $math_content[$marker] = renderMath( $content );
 398                                 } else {
 399                                         $math_content[$marker] = '<math>'.$content.'</math>';
 400                                 }
 401                         }
 402                 }
 403
 404                 # pre
 405                 $text = Parser::extractTags('pre', $text, $pre_content, $uniq_prefix);
 406                 foreach( $pre_content as $marker => $content ){
 407                         if( $render ){
 408                                 $pre_content[$marker] = '<pre>' . wfEscapeHTMLTagsOnly( $content ) . '</pre>';
 409                         } else {
 410                                 $pre_content[$marker] = '<pre>'.$content.'</pre>';
 411                         }
 412                 }
 413
 414                 # gallery
 415                 $text = Parser::extractTags('gallery', $text, $gallery_content, $uniq_prefix);
 416                 foreach( $gallery_content as $marker => $content ) {
 417                         require_once( 'ImageGallery.php' );
 418                         if ( $render ) {
 419                                 $gallery_content[$marker] = Parser::renderImageGallery( $content );
 420                         } else {
 421                                 $gallery_content[$marker] = '<gallery>'.$content.'</gallery>';
 422                         }
 423                 }
 424
 425                 # Comments
 426                 if($stripcomments) {
 427                         $text = Parser::extractTags(STRIP_COMMENTS, $text, $comment_content, $uniq_prefix);
 428                         foreach( $comment_content as $marker => $content ){
 429                                 $comment_content[$marker] = '<!--'.$content.'-->';
 430                         }
 431                 }
 432
 433                 # Extensions
 434                 foreach ( $this->mTagHooks as $tag => $callback ) {
 435                         $ext_content[$tag] = array();
 436                         $text = Parser::extractTagsAndParams( $tag, $text, $ext_content[$tag],
 437                                 $ext_tags[$tag], $ext_params[$tag], $uniq_prefix );
 438                         foreach( $ext_content[$tag] as $marker => $content ) {
 439                                 $full_tag = $ext_tags[$tag][$marker];
 440                                 $params = $ext_params[$tag][$marker];
 441                                 if ( $render )
 442                                         $ext_content[$tag][$marker] = call_user_func_array( $callback, array( $content, $params, &$this ) );
 443                                 else {
 444                                         if ( is_null( $content ) ) {
 445                                                 // Empty element tag
 446                                                 $ext_content[$tag][$marker] = $full_tag;
 447                                         } else {
 448                                                 $ext_content[$tag][$marker] = "$full_tag$content</$tag>";
 449                                         }
 450                                 }
 451                         }
 452                 }
 453
 454                 # Merge state with the pre-existing state, if there is one
 455                 if ( $state ) {
 456                         $state['html'] = $state['html'] + $html_content;
 457                         $state['nowiki'] = $state['nowiki'] + $nowiki_content;
 458                         $state['math'] = $state['math'] + $math_content;
 459                         $state['pre'] = $state['pre'] + $pre_content;
 460                         $state['comment'] = $state['comment'] + $comment_content;
 461                         $state['gallery'] = $state['gallery'] + $gallery_content;
 462
 463                         foreach( $ext_content as $tag => $array ) {
 464                                 if ( array_key_exists( $tag, $state ) ) {
 465                                         $state[$tag] = $state[$tag] + $array;
 466                                 }
 467                         }
 468                 } else {
 469                         $state = array(
 470                           'html' => $html_content,
 471                           'nowiki' => $nowiki_content,
 472                           'math' => $math_content,
 473                           'pre' => $pre_content,
 474                           'comment' => $comment_content,
 475                           'gallery' => $gallery_content,
 476                         ) + $ext_content;
 477                 }
 478                 return $text;
 479         }
 480
 481         /**
 482          * restores pre, math, and hiero removed by strip()
 483          *
 484          * always call unstripNoWiki() after this one
 485          * @access private
 486          */
 487         function unstrip( $text, &$state ) {
 488                 if ( !is_array( $state ) ) {
 489                         return $text;
 490                 }
 491
 492                 # Must expand in reverse order, otherwise nested tags will be corrupted
 493                 foreach( array_reverse( $state, true ) as $tag => $contentDict ) {
 494                         if( $tag != 'nowiki' && $tag != 'html' ) {
 495                                 foreach( array_reverse( $contentDict, true ) as $uniq => $content ) {
 496                                         $text = str_replace( $uniq, $content, $text );
 497                                 }
 498                         }
 499                 }
 500
 501                 return $text;
 502         }
 503
 504         /**
 505          * always call this after unstrip() to preserve the order
 506          *
 507          * @access private
 508          */
 509         function unstripNoWiki( $text, &$state ) {
 510                 if ( !is_array( $state ) ) {
 511                         return $text;
 512                 }
 513
 514                 # Must expand in reverse order, otherwise nested tags will be corrupted
 515                 for ( $content = end($state['nowiki']); $content !== false; $content = prev( $state['nowiki'] ) ) {
 516                         $text = str_replace( key( $state['nowiki'] ), $content, $text );
 517                 }
 518
 519                 global $wgRawHtml;
 520                 if ($wgRawHtml) {
 521                         for ( $content = end($state['html']); $content !== false; $content = prev( $state['html'] ) ) {
 522                                 $text = str_replace( key( $state['html'] ), $content, $text );
 523                         }
 524                 }
 525
 526                 return $text;
 527         }
 528
 529         /**
 530          * Add an item to the strip state
 531          * Returns the unique tag which must be inserted into the stripped text
 532          * The tag will be replaced with the original text in unstrip()
 533          *
 534          * @access private
 535          */
 536         function insertStripItem( $text, &$state ) {
 537                 $rnd = $this->mUniqPrefix . '-item' . Parser::getRandomString();
 538                 if ( !$state ) {
 539                         $state = array(
 540                           'html' => array(),
 541                           'nowiki' => array(),
 542                           'math' => array(),
 543                           'pre' => array(),
 544                           'comment' => array(),
 545                           'gallery' => array(),
 546                         );
 547                 }
 548                 $state['item'][$rnd] = $text;
 549                 return $rnd;
 550         }
 551
 552         /**
 553          * Interface with html tidy, used if $wgUseTidy = true.
 554          * If tidy isn't able to correct the markup, the original will be
 555          * returned in all its glory with a warning comment appended.
 556          *
 557          * Either the external tidy program or the in-process tidy extension
 558          * will be used depending on availability. Override the default
 559          * $wgTidyInternal setting to disable the internal if it's not working.
 560          *
 561          * @param string $text Hideous HTML input
 562          * @return string Corrected HTML output
 563          * @access public
 564          * @static
 565          */
 566         function tidy( $text ) {
 567                 global $wgTidyInternal;
 568                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 569 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 570 '<head><title>test</title></head><body>'.$text.'</body></html>';
 571                 if( $wgTidyInternal ) {
 572                         $correctedtext = Parser::internalTidy( $wrappedtext );
 573                 } else {
 574                         $correctedtext = Parser::externalTidy( $wrappedtext );
 575                 }
 576                 if( is_null( $correctedtext ) ) {
 577                         wfDebug( "Tidy error detected!\n" );
 578                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 579                 }
 580                 return $correctedtext;
 581         }
 582
 583         /**
 584          * Spawn an external HTML tidy process and get corrected markup back from it.
 585          *
 586          * @access private
 587          * @static
 588          */
 589         function externalTidy( $text ) {
 590                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 591                 $fname = 'Parser::externalTidy';
 592                 wfProfileIn( $fname );
 593
 594                 $cleansource = '';
 595                 $opts = ' -utf8';
 596
 597                 $descriptorspec = array(
 598                         0 => array('pipe', 'r'),
 599                         1 => array('pipe', 'w'),
 600                         2 => array('file', '/dev/null', 'a')
 601                 );
 602                 $pipes = array();
 603                 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 604                 if (is_resource($process)) {
 605                         fwrite($pipes[0], $text);
 606                         fclose($pipes[0]);
 607                         while (!feof($pipes[1])) {
 608                                 $cleansource .= fgets($pipes[1], 1024);
 609                         }
 610                         fclose($pipes[1]);
 611                         proc_close($process);
 612                 }
 613
 614                 wfProfileOut( $fname );
 615
 616                 if( $cleansource == '' && $text != '') {
 617                         // Some kind of error happened, so we couldn't get the corrected text.
 618                         // Just give up; we'll use the source text and append a warning.
 619                         return null;
 620                 } else {
 621                         return $cleansource;
 622                 }
 623         }
 624
 625         /**
 626          * Use the HTML tidy PECL extension to use the tidy library in-process,
 627          * saving the overhead of spawning a new process. Currently written to
 628          * the PHP 4.3.x version of the extension, may not work on PHP 5.
 629          *
 630          * 'pear install tidy' should be able to compile the extension module.
 631          *
 632          * @access private
 633          * @static
 634          */
 635         function internalTidy( $text ) {
 636                 global $wgTidyConf;
 637                 $fname = 'Parser::internalTidy';
 638                 wfProfileIn( $fname );
 639
 640                 tidy_load_config( $wgTidyConf );
 641                 tidy_set_encoding( 'utf8' );
 642                 tidy_parse_string( $text );
 643                 tidy_clean_repair();
 644                 if( tidy_get_status() == 2 ) {
 645                         // 2 is magic number for fatal error
 646                         // http://www.php.net/manual/en/function.tidy-get-status.php
 647                         $cleansource = null;
 648                 } else {
 649                         $cleansource = tidy_get_output();
 650                 }
 651                 wfProfileOut( $fname );
 652                 return $cleansource;
 653         }
 654
 655         /**
 656          * parse the wiki syntax used to render tables
 657          *
 658          * @access private
 659          */
 660         function doTableStuff ( $t ) {
 661                 $fname = 'Parser::doTableStuff';
 662                 wfProfileIn( $fname );
 663
 664                 $t = explode ( "\n" , $t ) ;
 665                 $td = array () ; # Is currently a td tag open?
 666                 $ltd = array () ; # Was it TD or TH?
 667                 $tr = array () ; # Is currently a tr tag open?
 668                 $ltr = array () ; # tr attributes
 669                 $indent_level = 0; # indent level of the table
 670                 foreach ( $t AS $k => $x )
 671                 {
 672                         $x = trim ( $x ) ;
 673                         $fc = substr ( $x , 0 , 1 ) ;
 674                         if ( preg_match( '/^(:*)\{\|(.*)$/', $x, $matches ) ) {
 675                                 $indent_level = strlen( $matches[1] );
 676
 677                                 $attributes = $this->unstripForHTML( $matches[2] );
 678
 679                                 $t[$k] = str_repeat( '<dl><dd>', $indent_level ) .
 680                                         '<table' . Sanitizer::fixTagAttributes ( $attributes, 'table' ) . '>' ;
 681                                 array_push ( $td , false ) ;
 682                                 array_push ( $ltd , '' ) ;
 683                                 array_push ( $tr , false ) ;
 684                                 array_push ( $ltr , '' ) ;
 685                         }
 686                         else if ( count ( $td ) == 0 ) { } # Don't do any of the following
 687                         else if ( '|}' == substr ( $x , 0 , 2 ) ) {
 688                                 $z = "</table>" . substr ( $x , 2);
 689                                 $l = array_pop ( $ltd ) ;
 690                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 691                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 692                                 array_pop ( $ltr ) ;
 693                                 $t[$k] = $z . str_repeat( '</dd></dl>', $indent_level );
 694                         }
 695                         else if ( '|-' == substr ( $x , 0 , 2 ) ) { # Allows for |---------------
 696                                 $x = substr ( $x , 1 ) ;
 697                                 while ( $x != '' && substr ( $x , 0 , 1 ) == '-' ) $x = substr ( $x , 1 ) ;
 698                                 $z = '' ;
 699                                 $l = array_pop ( $ltd ) ;
 700                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 701                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 702                                 array_pop ( $ltr ) ;
 703                                 $t[$k] = $z ;
 704                                 array_push ( $tr , false ) ;
 705                                 array_push ( $td , false ) ;
 706                                 array_push ( $ltd , '' ) ;
 707                                 $attributes = $this->unstripForHTML( $x );
 708                                 array_push ( $ltr , Sanitizer::fixTagAttributes ( $attributes, 'tr' ) ) ;
 709                         }
 710                         else if ( '|' == $fc || '!' == $fc || '|+' == substr ( $x , 0 , 2 ) ) { # Caption
 711                                 # $x is a table row
 712                                 if ( '|+' == substr ( $x , 0 , 2 ) ) {
 713                                         $fc = '+' ;
 714                                         $x = substr ( $x , 1 ) ;
 715                                 }
 716                                 $after = substr ( $x , 1 ) ;
 717                                 if ( $fc == '!' ) $after = str_replace ( '!!' , '||' , $after ) ;
 718                                 $after = explode ( '||' , $after ) ;
 719                                 $t[$k] = '' ;
 720
 721                                 # Loop through each table cell
 722                                 foreach ( $after AS $theline )
 723                                 {
 724                                         $z = '' ;
 725                                         if ( $fc != '+' )
 726                                         {
 727                                                 $tra = array_pop ( $ltr ) ;
 728                                                 if ( !array_pop ( $tr ) ) $z = '<tr'.$tra.">\n" ;
 729                                                 array_push ( $tr , true ) ;
 730                                                 array_push ( $ltr , '' ) ;
 731                                         }
 732
 733                                         $l = array_pop ( $ltd ) ;
 734                                         if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 735                                         if ( $fc == '|' ) $l = 'td' ;
 736                                         else if ( $fc == '!' ) $l = 'th' ;
 737                                         else if ( $fc == '+' ) $l = 'caption' ;
 738                                         else $l = '' ;
 739                                         array_push ( $ltd , $l ) ;
 740
 741                                         # Cell parameters
 742                                         $y = explode ( '|' , $theline , 2 ) ;
 743                                         # Note that a '|' inside an invalid link should not
 744                                         # be mistaken as delimiting cell parameters
 745                                         if ( strpos( $y[0], '[[' ) !== false ) {
 746                                                 $y = array ($theline);
 747                                         }
 748                                         if ( count ( $y ) == 1 )
 749                                                 $y = "{$z}<{$l}>{$y[0]}" ;
 750                                         else {
 751                                                 $attributes = $this->unstripForHTML( $y[0] );
 752                                                 $y = "{$z}<{$l}".Sanitizer::fixTagAttributes($attributes, $l).">{$y[1]}" ;
 753                                         }
 754                                         $t[$k] .= $y ;
 755                                         array_push ( $td , true ) ;
 756                                 }
 757                         }
 758                 }
 759
 760                 # Closing open td, tr && table
 761                 while ( count ( $td ) > 0 )
 762                 {
 763                         if ( array_pop ( $td ) ) $t[] = '</td>' ;
 764                         if ( array_pop ( $tr ) ) $t[] = '</tr>' ;
 765                         $t[] = '</table>' ;
 766                 }
 767
 768                 $t = implode ( "\n" , $t ) ;
 769                 wfProfileOut( $fname );
 770                 return $t ;
 771         }
 772
 773         /**
 774          * Helper function for parse() that transforms wiki markup into
 775          * HTML. Only called for $mOutputType == OT_HTML.
 776          *
 777          * @access private
 778          */
 779         function internalParse( $text ) {
 780                 global $wgContLang;
 781                 $args = array();
 782                 $isMain = true;
 783                 $fname = 'Parser::internalParse';
 784                 wfProfileIn( $fname );
 785
 786                 # Remove <noinclude> tags and <includeonly> sections
 787                 $text = strtr( $text, array( '<onlyinclude>' => '' , '</onlyinclude>' => '' ) );
 788                 $text = strtr( $text, array( '<noinclude>' => '', '</noinclude>' => '') );
 789                 $text = preg_replace( '/<includeonly>.*?<\/includeonly>/s', '', $text );
 790
 791                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ) );
 792                 $text = $this->replaceVariables( $text, $args );
 793
 794                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 795
 796                 $text = $this->doHeadings( $text );
 797                 if($this->mOptions->getUseDynamicDates()) {
 798                         $df =& DateFormatter::getInstance();
 799                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 800                 }
 801                 $text = $this->doAllQuotes( $text );
 802                 $text = $this->replaceInternalLinks( $text );
 803                 $text = $this->replaceExternalLinks( $text );
 804
 805                 # replaceInternalLinks may sometimes leave behind
 806                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 807                 $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
 808
 809                 $text = $this->doMagicLinks( $text );
 810                 $text = $this->doTableStuff( $text );
 811                 $text = $this->formatHeadings( $text, $isMain );
 812
 813                 $regex = '/<!--IW_TRANSCLUDE (\d+)-->/';
 814                 $text = preg_replace_callback($regex, array(&$this, 'scarySubstitution'), $text);
 815
 816                 wfProfileOut( $fname );
 817                 return $text;
 818         }
 819
 820         function scarySubstitution($matches) {
 821 #               return "[[".$matches[0]."]]";
 822                 return $this->mIWTransData[(int)$matches[0]];
 823         }
 824
 825         /**
 826          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 827          * magic external links.
 828          *
 829          * @access private
 830          */
 831         function &doMagicLinks( &$text ) {
 832                 $text = $this->magicISBN( $text );
 833                 $text = $this->magicRFC( $text, 'RFC ', 'rfcurl' );
 834                 $text = $this->magicRFC( $text, 'PMID ', 'pubmedurl' );
 835                 return $text;
 836         }
 837
 838         /**
 839          * Parse headers and return html
 840          *
 841          * @access private
 842          */
 843         function doHeadings( $text ) {
 844                 $fname = 'Parser::doHeadings';
 845                 wfProfileIn( $fname );
 846                 for ( $i = 6; $i >= 1; --$i ) {
 847                         $h = str_repeat( '=', $i );
 848                         $text = preg_replace( "/^{$h}(.+){$h}(\\s|$)/m",
 849                           "<h{$i}>\\1</h{$i}>\\2", $text );
 850                 }
 851                 wfProfileOut( $fname );
 852                 return $text;
 853         }
 854
 855         /**
 856          * Replace single quotes with HTML markup
 857          * @access private
 858          * @return string the altered text
 859          */
 860         function doAllQuotes( $text ) {
 861                 $fname = 'Parser::doAllQuotes';
 862                 wfProfileIn( $fname );
 863                 $outtext = '';
 864                 $lines = explode( "\n", $text );
 865                 foreach ( $lines as $line ) {
 866                         $outtext .= $this->doQuotes ( $line ) . "\n";
 867                 }
 868                 $outtext = substr($outtext, 0,-1);
 869                 wfProfileOut( $fname );
 870                 return $outtext;
 871         }
 872
 873         /**
 874          * Helper function for doAllQuotes()
 875          * @access private
 876          */
 877         function doQuotes( $text ) {
 878                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
 879                 if ( count( $arr ) == 1 )
 880                         return $text;
 881                 else
 882                 {
 883                         # First, do some preliminary work. This may shift some apostrophes from
 884                         # being mark-up to being text. It also counts the number of occurrences
 885                         # of bold and italics mark-ups.
 886                         $i = 0;
 887                         $numbold = 0;
 888                         $numitalics = 0;
 889                         foreach ( $arr as $r )
 890                         {
 891                                 if ( ( $i % 2 ) == 1 )
 892                                 {
 893                                         # If there are ever four apostrophes, assume the first is supposed to
 894                                         # be text, and the remaining three constitute mark-up for bold text.
 895                                         if ( strlen( $arr[$i] ) == 4 )
 896                                         {
 897                                                 $arr[$i-1] .= "'";
 898                                                 $arr[$i] = "'''";
 899                                         }
 900                                         # If there are more than 5 apostrophes in a row, assume they're all
 901                                         # text except for the last 5.
 902                                         else if ( strlen( $arr[$i] ) > 5 )
 903                                         {
 904                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
 905                                                 $arr[$i] = "'''''";
 906                                         }
 907                                         # Count the number of occurrences of bold and italics mark-ups.
 908                                         # We are not counting sequences of five apostrophes.
 909                                         if ( strlen( $arr[$i] ) == 2 ) $numitalics++;  else
 910                                         if ( strlen( $arr[$i] ) == 3 ) $numbold++;     else
 911                                         if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
 912                                 }
 913                                 $i++;
 914                         }
 915
 916                         # If there is an odd number of both bold and italics, it is likely
 917                         # that one of the bold ones was meant to be an apostrophe followed
 918                         # by italics. Which one we cannot know for certain, but it is more
 919                         # likely to be one that has a single-letter word before it.
 920                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
 921                         {
 922                                 $i = 0;
 923                                 $firstsingleletterword = -1;
 924                                 $firstmultiletterword = -1;
 925                                 $firstspace = -1;
 926                                 foreach ( $arr as $r )
 927                                 {
 928                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
 929                                         {
 930                                                 $x1 = substr ($arr[$i-1], -1);
 931                                                 $x2 = substr ($arr[$i-1], -2, 1);
 932                                                 if ($x1 == ' ') {
 933                                                         if ($firstspace == -1) $firstspace = $i;
 934                                                 } else if ($x2 == ' ') {
 935                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
 936                                                 } else {
 937                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
 938                                                 }
 939                                         }
 940                                         $i++;
 941                                 }
 942
 943                                 # If there is a single-letter word, use it!
 944                                 if ($firstsingleletterword > -1)
 945                                 {
 946                                         $arr [ $firstsingleletterword ] = "''";
 947                                         $arr [ $firstsingleletterword-1 ] .= "'";
 948                                 }
 949                                 # If not, but there's a multi-letter word, use that one.
 950                                 else if ($firstmultiletterword > -1)
 951                                 {
 952                                         $arr [ $firstmultiletterword ] = "''";
 953                                         $arr [ $firstmultiletterword-1 ] .= "'";
 954                                 }
 955                                 # ... otherwise use the first one that has neither.
 956                                 # (notice that it is possible for all three to be -1 if, for example,
 957                                 # there is only one pentuple-apostrophe in the line)
 958                                 else if ($firstspace > -1)
 959                                 {
 960                                         $arr [ $firstspace ] = "''";
 961                                         $arr [ $firstspace-1 ] .= "'";
 962                                 }
 963                         }
 964
 965                         # Now let's actually convert our apostrophic mush to HTML!
 966                         $output = '';
 967                         $buffer = '';
 968                         $state = '';
 969                         $i = 0;
 970                         foreach ($arr as $r)
 971                         {
 972                                 if (($i % 2) == 0)
 973                                 {
 974                                         if ($state == 'both')
 975                                                 $buffer .= $r;
 976                                         else
 977                                                 $output .= $r;
 978                                 }
 979                                 else
 980                                 {
 981                                         if (strlen ($r) == 2)
 982                                         {
 983                                                 if ($state == 'i')
 984                                                 { $output .= '</i>'; $state = ''; }
 985                                                 else if ($state == 'bi')
 986                                                 { $output .= '</i>'; $state = 'b'; }
 987                                                 else if ($state == 'ib')
 988                                                 { $output .= '</b></i><b>'; $state = 'b'; }
 989                                                 else if ($state == 'both')
 990                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
 991                                                 else # $state can be 'b' or ''
 992                                                 { $output .= '<i>'; $state .= 'i'; }
 993                                         }
 994                                         else if (strlen ($r) == 3)
 995                                         {
 996                                                 if ($state == 'b')
 997                                                 { $output .= '</b>'; $state = ''; }
 998                                                 else if ($state == 'bi')
 999                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1000                                                 else if ($state == 'ib')
1001                                                 { $output .= '</b>'; $state = 'i'; }
1002                                                 else if ($state == 'both')
1003                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1004                                                 else # $state can be 'i' or ''
1005                                                 { $output .= '<b>'; $state .= 'b'; }
1006                                         }
1007                                         else if (strlen ($r) == 5)
1008                                         {
1009                                                 if ($state == 'b')
1010                                                 { $output .= '</b><i>'; $state = 'i'; }
1011                                                 else if ($state == 'i')
1012                                                 { $output .= '</i><b>'; $state = 'b'; }
1013                                                 else if ($state == 'bi')
1014                                                 { $output .= '</i></b>'; $state = ''; }
1015                                                 else if ($state == 'ib')
1016                                                 { $output .= '</b></i>'; $state = ''; }
1017                                                 else if ($state == 'both')
1018                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1019                                                 else # ($state == '')
1020                                                 { $buffer = ''; $state = 'both'; }
1021                                         }
1022                                 }
1023                                 $i++;
1024                         }
1025                         # Now close all remaining tags.  Notice that the order is important.
1026                         if ($state == 'b' || $state == 'ib')
1027                                 $output .= '</b>';
1028                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
1029                                 $output .= '</i>';
1030                         if ($state == 'bi')
1031                                 $output .= '</b>';
1032                         if ($state == 'both')
1033                                 $output .= '<b><i>'.$buffer.'</i></b>';
1034                         return $output;
1035                 }
1036         }
1037
1038         /**
1039          * Replace external links
1040          *
1041          * Note: this is all very hackish and the order of execution matters a lot.
1042          * Make sure to run maintenance/parserTests.php if you change this code.
1043          *
1044          * @access private
1045          */
1046         function replaceExternalLinks( $text ) {
1047                 global $wgContLang;
1048                 $fname = 'Parser::replaceExternalLinks';
1049                 wfProfileIn( $fname );
1050
1051                 $sk =& $this->mOptions->getSkin();
1052
1053                 $bits = preg_split( EXT_LINK_BRACKETED, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1054
1055                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
1056
1057                 $i = 0;
1058                 while ( $i<count( $bits ) ) {
1059                         $url = $bits[$i++];
1060                         $protocol = $bits[$i++];
1061                         $text = $bits[$i++];
1062                         $trail = $bits[$i++];
1063
1064                         # The characters '<' and '>' (which were escaped by
1065                         # removeHTMLtags()) should not be included in
1066                         # URLs, per RFC 2396.
1067                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1068                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1069                                 $url = substr($url, 0, $m2[0][1]);
1070                         }
1071
1072                         # If the link text is an image URL, replace it with an <img> tag
1073                         # This happened by accident in the original parser, but some people used it extensively
1074                         $img = $this->maybeMakeExternalImage( $text );
1075                         if ( $img !== false ) {
1076                                 $text = $img;
1077                         }
1078
1079                         $dtrail = '';
1080
1081                         # Set linktype for CSS - if URL==text, link is essentially free
1082                         $linktype = ($text == $url) ? 'free' : 'text';
1083
1084                         # No link text, e.g. [http://domain.tld/some.link]
1085                         if ( $text == '' ) {
1086                                 # Autonumber if allowed
1087                                 if ( strpos( HTTP_PROTOCOLS, str_replace('/','\/', $protocol) ) !== false ) {
1088                                         $text = '[' . ++$this->mAutonumber . ']';
1089                                         $linktype = 'autonumber';
1090                                 } else {
1091                                         # Otherwise just use the URL
1092                                         $text = htmlspecialchars( $url );
1093                                         $linktype = 'free';
1094                                 }
1095                         } else {
1096                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1097                                 # Check for trail
1098                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1099                         }
1100
1101                         $text = $wgContLang->markNoConversion($text);
1102
1103                         # Replace &amp; from obsolete syntax with &.
1104                         # All HTML entities will be escaped by makeExternalLink()
1105                         # or maybeMakeExternalImage()
1106                         $url = str_replace( '&amp;', '&', $url );
1107
1108                         # Process the trail (i.e. everything after this link up until start of the next link),
1109                         # replacing any non-bracketed links
1110                         $trail = $this->replaceFreeExternalLinks( $trail );
1111
1112
1113                         # Use the encoded URL
1114                         # This means that users can paste URLs directly into the text
1115                         # Funny characters like &ouml; aren't valid in URLs anyway
1116                         # This was changed in August 2004
1117                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype ) . $dtrail . $trail;
1118                 }
1119
1120                 wfProfileOut( $fname );
1121                 return $s;
1122         }
1123
1124         /**
1125          * Replace anything that looks like a URL with a link
1126          * @access private
1127          */
1128         function replaceFreeExternalLinks( $text ) {
1129                 global $wgContLang;
1130                 $fname = 'Parser::replaceFreeExternalLinks';
1131                 wfProfileIn( $fname );
1132
1133                 $bits = preg_split( '/(\b(?:' . wfUrlProtocols() . '))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1134                 $s = array_shift( $bits );
1135                 $i = 0;
1136
1137                 $sk =& $this->mOptions->getSkin();
1138
1139                 while ( $i < count( $bits ) ){
1140                         $protocol = $bits[$i++];
1141                         $remainder = $bits[$i++];
1142
1143                         if ( preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1144                                 # Found some characters after the protocol that look promising
1145                                 $url = $protocol . $m[1];
1146                                 $trail = $m[2];
1147
1148                                 # The characters '<' and '>' (which were escaped by
1149                                 # removeHTMLtags()) should not be included in
1150                                 # URLs, per RFC 2396.
1151                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1152                                         $trail = substr($url, $m2[0][1]) . $trail;
1153                                         $url = substr($url, 0, $m2[0][1]);
1154                                 }
1155
1156                                 # Move trailing punctuation to $trail
1157                                 $sep = ',;\.:!?';
1158                                 # If there is no left bracket, then consider right brackets fair game too
1159                                 if ( strpos( $url, '(' ) === false ) {
1160                                         $sep .= ')';
1161                                 }
1162
1163                                 $numSepChars = strspn( strrev( $url ), $sep );
1164                                 if ( $numSepChars ) {
1165                                         $trail = substr( $url, -$numSepChars ) . $trail;
1166                                         $url = substr( $url, 0, -$numSepChars );
1167                                 }
1168
1169                                 # Replace &amp; from obsolete syntax with &.
1170                                 # All HTML entities will be escaped by makeExternalLink()
1171                                 # or maybeMakeExternalImage()
1172                                 $url = str_replace( '&amp;', '&', $url );
1173
1174                                 # Is this an external image?
1175                                 $text = $this->maybeMakeExternalImage( $url );
1176                                 if ( $text === false ) {
1177                                         # Not an image, make a link
1178                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free' );
1179                                 }
1180                                 $s .= $text . $trail;
1181                         } else {
1182                                 $s .= $protocol . $remainder;
1183                         }
1184                 }
1185                 wfProfileOut( $fname );
1186                 return $s;
1187         }
1188
1189         /**
1190          * make an image if it's allowed, either through the global
1191          * option or through the exception
1192          * @access private
1193          */
1194         function maybeMakeExternalImage( $url ) {
1195                 $sk =& $this->mOptions->getSkin();
1196                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1197                 $imagesexception = !empty($imagesfrom);
1198                 $text = false;
1199                 if ( $this->mOptions->getAllowExternalImages()
1200                      || ( $imagesexception && strpos( $url, $imagesfrom ) === 0 ) ) {
1201                         if ( preg_match( EXT_IMAGE_REGEX, $url ) ) {
1202                                 # Image found
1203                                 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
1204                         }
1205                 }
1206                 return $text;
1207         }
1208
1209         /**
1210          * Process [[ ]] wikilinks
1211          *
1212          * @access private
1213          */
1214         function replaceInternalLinks( $s ) {
1215                 global $wgContLang, $wgLinkCache;
1216                 static $fname = 'Parser::replaceInternalLinks' ;
1217
1218                 wfProfileIn( $fname );
1219
1220                 wfProfileIn( $fname.'-setup' );
1221                 static $tc = FALSE;
1222                 # the % is needed to support urlencoded titles as well
1223                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1224
1225                 $sk =& $this->mOptions->getSkin();
1226
1227                 #split the entire text string on occurences of [[
1228                 $a = explode( '[[', ' ' . $s );
1229                 #get the first element (all text up to first [[), and remove the space we added
1230                 $s = array_shift( $a );
1231                 $s = substr( $s, 1 );
1232
1233                 # Match a link having the form [[namespace:link|alternate]]trail
1234                 static $e1 = FALSE;
1235                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1236                 # Match cases where there is no "]]", which might still be images
1237                 static $e1_img = FALSE;
1238                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1239                 # Match the end of a line for a word that's not followed by whitespace,
1240                 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1241                 $e2 = wfMsgForContent( 'linkprefix' );
1242
1243                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1244
1245                 if( is_null( $this->mTitle ) ) {
1246                         wfDebugDieBacktrace( 'nooo' );
1247                 }
1248                 $nottalk = !$this->mTitle->isTalkPage();
1249
1250                 if ( $useLinkPrefixExtension ) {
1251                         if ( preg_match( $e2, $s, $m ) ) {
1252                                 $first_prefix = $m[2];
1253                         } else {
1254                                 $first_prefix = false;
1255                         }
1256                 } else {
1257                         $prefix = '';
1258                 }
1259
1260                 $selflink = $this->mTitle->getPrefixedText();
1261                 wfProfileOut( $fname.'-setup' );
1262
1263                 $checkVariantLink = sizeof($wgContLang->getVariants())>1;
1264                 $useSubpages = $this->areSubpagesAllowed();
1265
1266                 # Loop for each link
1267                 for ($k = 0; isset( $a[$k] ); $k++) {
1268                         $line = $a[$k];
1269                         if ( $useLinkPrefixExtension ) {
1270                                 wfProfileIn( $fname.'-prefixhandling' );
1271                                 if ( preg_match( $e2, $s, $m ) ) {
1272                                         $prefix = $m[2];
1273                                         $s = $m[1];
1274                                 } else {
1275                                         $prefix='';
1276                                 }
1277                                 # first link
1278                                 if($first_prefix) {
1279                                         $prefix = $first_prefix;
1280                                         $first_prefix = false;
1281                                 }
1282                                 wfProfileOut( $fname.'-prefixhandling' );
1283                         }
1284
1285                         $might_be_img = false;
1286
1287                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1288                                 $text = $m[2];
1289                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1290                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1291                                 # the real problem is with the $e1 regex
1292                                 # See bug 1300.
1293                                 #
1294                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1295                                 # and no image is in sight. See bug 2095.
1296                                 #
1297                                 if( $text !== '' && preg_match( "/^\](.*)/s", $m[3], $n ) ) {
1298                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1299                                         $m[3] = $n[1];
1300                                 }
1301                                 # fix up urlencoded title texts
1302                                 if(preg_match('/%/', $m[1] )) $m[1] = urldecode($m[1]);
1303                                 $trail = $m[3];
1304                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1305                                 $might_be_img = true;
1306                                 $text = $m[2];
1307                                 if(preg_match('/%/', $m[1] )) $m[1] = urldecode($m[1]);
1308                                 $trail = "";
1309                         } else { # Invalid form; output directly
1310                                 $s .= $prefix . '[[' . $line ;
1311                                 continue;
1312                         }
1313
1314                         # Don't allow internal links to pages containing
1315                         # PROTO: where PROTO is a valid URL protocol; these
1316                         # should be external links.
1317                         if (preg_match('/^(\b(?:' . wfUrlProtocols() . '))/', $m[1])) {
1318                                 $s .= $prefix . '[[' . $line ;
1319                                 continue;
1320                         }
1321
1322                         # Make subpage if necessary
1323                         if( $useSubpages ) {
1324                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1325                         } else {
1326                                 $link = $m[1];
1327                         }
1328
1329                         $noforce = (substr($m[1], 0, 1) != ':');
1330                         if (!$noforce) {
1331                                 # Strip off leading ':'
1332                                 $link = substr($link, 1);
1333                         }
1334
1335                         $nt = Title::newFromText( $this->unstripNoWiki($link, $this->mStripState) );
1336                         if( !$nt ) {
1337                                 $s .= $prefix . '[[' . $line;
1338                                 continue;
1339                         }
1340
1341                         #check other language variants of the link
1342                         #if the article does not exist
1343                         if( $checkVariantLink
1344                             && $nt->getArticleID() == 0 ) {
1345                                 $wgContLang->findVariantLink($link, $nt);
1346                         }
1347
1348                         $ns = $nt->getNamespace();
1349                         $iw = $nt->getInterWiki();
1350
1351                         if ($might_be_img) { # if this is actually an invalid link
1352                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1353                                         $found = false;
1354                                         while (isset ($a[$k+1]) ) {
1355                                                 #look at the next 'line' to see if we can close it there
1356                                                 $spliced = array_splice( $a, $k + 1, 1 );
1357                                                 $next_line = array_shift( $spliced );
1358                                                 if( preg_match("/^(.*?]].*?)]](.*)$/sD", $next_line, $m) ) {
1359                                                 # the first ]] closes the inner link, the second the image
1360                                                         $found = true;
1361                                                         $text .= '[[' . $m[1];
1362                                                         $trail = $m[2];
1363                                                         break;
1364                                                 } elseif( preg_match("/^.*?]].*$/sD", $next_line, $m) ) {
1365                                                         #if there's exactly one ]] that's fine, we'll keep looking
1366                                                         $text .= '[[' . $m[0];
1367                                                 } else {
1368                                                         #if $next_line is invalid too, we need look no further
1369                                                         $text .= '[[' . $next_line;
1370                                                         break;
1371                                                 }
1372                                         }
1373                                         if ( !$found ) {
1374                                                 # we couldn't find the end of this imageLink, so output it raw
1375                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1376                                                 $text = $this->replaceInternalLinks($text);
1377                                                 $s .= $prefix . '[[' . $link . '|' . $text;
1378                                                 # note: no $trail, because without an end, there *is* no trail
1379                                                 continue;
1380                                         }
1381                                 } else { #it's not an image, so output it raw
1382                                         $s .= $prefix . '[[' . $link . '|' . $text;
1383                                         # note: no $trail, because without an end, there *is* no trail
1384                                         continue;
1385                                 }
1386                         }
1387
1388                         $wasblank = ( '' == $text );
1389                         if( $wasblank ) $text = $link;
1390
1391
1392                         # Link not escaped by : , create the various objects
1393                         if( $noforce ) {
1394
1395                                 # Interwikis
1396                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1397                                         array_push( $this->mOutput->mLanguageLinks, $nt->getFullText() );
1398                                         $s = rtrim($s . "\n");
1399                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1400                                         continue;
1401                                 }
1402
1403                                 if ( $ns == NS_IMAGE ) {
1404                                         wfProfileIn( "$fname-image" );
1405                                         if ( !wfIsBadImage( $nt->getDBkey() ) ) {
1406                                                 # recursively parse links inside the image caption
1407                                                 # actually, this will parse them in any other parameters, too,
1408                                                 # but it might be hard to fix that, and it doesn't matter ATM
1409                                                 $text = $this->replaceExternalLinks($text);
1410                                                 $text = $this->replaceInternalLinks($text);
1411
1412                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1413                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text ) ) . $trail;
1414                                                 $wgLinkCache->addImageLinkObj( $nt );
1415
1416                                                 wfProfileOut( "$fname-image" );
1417                                                 continue;
1418                                         }
1419                                         wfProfileOut( "$fname-image" );
1420
1421                                 }
1422
1423                                 if ( $ns == NS_CATEGORY ) {
1424                                         wfProfileIn( "$fname-category" );
1425                                         $t = $wgContLang->convertHtml( $nt->getText() );
1426                                         $s = rtrim($s . "\n"); # bug 87
1427
1428                                         $wgLinkCache->suspend(); # Don't save in links/brokenlinks
1429                                         $t = $sk->makeLinkObj( $nt, $t, '', '' , $prefix );
1430                                         $wgLinkCache->resume();
1431
1432                                         if ( $wasblank ) {
1433                                                 if ( $this->mTitle->getNamespace() == NS_CATEGORY ) {
1434                                                         $sortkey = $this->mTitle->getText();
1435                                                 } else {
1436                                                         $sortkey = $this->mTitle->getPrefixedText();
1437                                                 }
1438                                         } else {
1439                                                 $sortkey = $text;
1440                                         }
1441                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1442                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1443                                         $wgLinkCache->addCategoryLinkObj( $nt, $sortkey );
1444                                         $this->mOutput->addCategoryLink( $t );
1445
1446                                         /**
1447                                          * Strip the whitespace Category links produce, see bug 87
1448                                          * @todo We might want to use trim($tmp, "\n") here.
1449                                          */
1450                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1451
1452                                         wfProfileOut( "$fname-category" );
1453                                         continue;
1454                                 }
1455                         }
1456
1457                         if( ( $nt->getPrefixedText() === $selflink ) &&
1458                             ( $nt->getFragment() === '' ) ) {
1459                                 # Self-links are handled specially; generally de-link and change to bold.
1460                                 $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1461                                 continue;
1462                         }
1463
1464                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1465                         if( $ns == NS_MEDIA ) {
1466                                 $link = $sk->makeMediaLinkObj( $nt, $text );
1467                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1468                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1469                                 $wgLinkCache->addImageLinkObj( $nt );
1470                                 continue;
1471                         } elseif( $ns == NS_SPECIAL ) {
1472                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1473                                 continue;
1474                         } elseif( $ns == NS_IMAGE ) {
1475                                 $img = Image::newFromTitle( $nt );
1476                                 if( $img->exists() ) {
1477                                         // Force a blue link if the file exists; may be a remote
1478                                         // upload on the shared repository, and we want to see its
1479                                         // auto-generated page.
1480                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1481                                         continue;
1482                                 }
1483                         }
1484                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1485                 }
1486                 wfProfileOut( $fname );
1487                 return $s;
1488         }
1489
1490         /**
1491          * Make a link placeholder. The text returned can be later resolved to a real link with
1492          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1493          * parsing of interwiki links, and secondly to allow all extistence checks and
1494          * article length checks (for stub links) to be bundled into a single query.
1495          *
1496          */
1497         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1498                 if ( ! is_object($nt) ) {
1499                         # Fail gracefully
1500                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1501                 } else {
1502                         # Separate the link trail from the rest of the link
1503                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1504
1505                         if ( $nt->isExternal() ) {
1506                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1507                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1508                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1509                         } else {
1510                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1511                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1512                                 $this->mLinkHolders['queries'][] = $query;
1513                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1514                                 $this->mLinkHolders['titles'][] = $nt;
1515
1516                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1517                         }
1518                 }
1519                 return $retVal;
1520         }
1521
1522         /**
1523          * Render a forced-blue link inline; protect against double expansion of
1524          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1525          * Since this little disaster has to split off the trail text to avoid
1526          * breaking URLs in the following text without breaking trails on the
1527          * wiki links, it's been made into a horrible function.
1528          *
1529          * @param Title $nt
1530          * @param string $text
1531          * @param string $query
1532          * @param string $trail
1533          * @param string $prefix
1534          * @return string HTML-wikitext mix oh yuck
1535          */
1536         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1537                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1538                 $sk =& $this->mOptions->getSkin();
1539                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1540                 return $this->armorLinks( $link ) . $trail;
1541         }
1542
1543         /**
1544          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1545          * going to go through further parsing steps before inline URL expansion.
1546          *
1547          * In particular this is important when using action=render, which causes
1548          * full URLs to be included.
1549          *
1550          * Oh man I hate our multi-layer parser!
1551          *
1552          * @param string more-or-less HTML
1553          * @return string less-or-more HTML with NOPARSE bits
1554          */
1555         function armorLinks( $text ) {
1556                 return preg_replace( "/\b(" . wfUrlProtocols() . ')/',
1557                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1558         }
1559
1560         /**
1561          * Return true if subpage links should be expanded on this page.
1562          * @return bool
1563          */
1564         function areSubpagesAllowed() {
1565                 # Some namespaces don't allow subpages
1566                 global $wgNamespacesWithSubpages;
1567                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1568         }
1569
1570         /**
1571          * Handle link to subpage if necessary
1572          * @param string $target the source of the link
1573          * @param string &$text the link text, modified as necessary
1574          * @return string the full name of the link
1575          * @access private
1576          */
1577         function maybeDoSubpageLink($target, &$text) {
1578                 # Valid link forms:
1579                 # Foobar -- normal
1580                 # :Foobar -- override special treatment of prefix (images, language links)
1581                 # /Foobar -- convert to CurrentPage/Foobar
1582                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1583                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1584                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1585
1586                 $fname = 'Parser::maybeDoSubpageLink';
1587                 wfProfileIn( $fname );
1588                 $ret = $target; # default return value is no change
1589
1590                 # Some namespaces don't allow subpages,
1591                 # so only perform processing if subpages are allowed
1592                 if( $this->areSubpagesAllowed() ) {
1593                         # Look at the first character
1594                         if( $target != '' && $target{0} == '/' ) {
1595                                 # / at end means we don't want the slash to be shown
1596                                 if( substr( $target, -1, 1 ) == '/' ) {
1597                                         $target = substr( $target, 1, -1 );
1598                                         $noslash = $target;
1599                                 } else {
1600                                         $noslash = substr( $target, 1 );
1601                                 }
1602
1603                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash);
1604                                 if( '' === $text ) {
1605                                         $text = $target;
1606                                 } # this might be changed for ugliness reasons
1607                         } else {
1608                                 # check for .. subpage backlinks
1609                                 $dotdotcount = 0;
1610                                 $nodotdot = $target;
1611                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1612                                         ++$dotdotcount;
1613                                         $nodotdot = substr( $nodotdot, 3 );
1614                                 }
1615                                 if($dotdotcount > 0) {
1616                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1617                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1618                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1619                                                 # / at the end means don't show full path
1620                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1621                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1622                                                         if( '' === $text ) {
1623                                                                 $text = $nodotdot;
1624                                                         }
1625                                                 }
1626                                                 $nodotdot = trim( $nodotdot );
1627                                                 if( $nodotdot != '' ) {
1628                                                         $ret .= '/' . $nodotdot;
1629                                                 }
1630                                         }
1631                                 }
1632                         }
1633                 }
1634
1635                 wfProfileOut( $fname );
1636                 return $ret;
1637         }
1638
1639         /**#@+
1640          * Used by doBlockLevels()
1641          * @access private
1642          */
1643         /* private */ function closeParagraph() {
1644                 $result = '';
1645                 if ( '' != $this->mLastSection ) {
1646                         $result = '</' . $this->mLastSection  . ">\n";
1647                 }
1648                 $this->mInPre = false;
1649                 $this->mLastSection = '';
1650                 return $result;
1651         }
1652         # getCommon() returns the length of the longest common substring
1653         # of both arguments, starting at the beginning of both.
1654         #
1655         /* private */ function getCommon( $st1, $st2 ) {
1656                 $fl = strlen( $st1 );
1657                 $shorter = strlen( $st2 );
1658                 if ( $fl < $shorter ) { $shorter = $fl; }
1659
1660                 for ( $i = 0; $i < $shorter; ++$i ) {
1661                         if ( $st1{$i} != $st2{$i} ) { break; }
1662                 }
1663                 return $i;
1664         }
1665         # These next three functions open, continue, and close the list
1666         # element appropriate to the prefix character passed into them.
1667         #
1668         /* private */ function openList( $char ) {
1669                 $result = $this->closeParagraph();
1670
1671                 if ( '*' == $char ) { $result .= '<ul><li>'; }
1672                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
1673                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
1674                 else if ( ';' == $char ) {
1675                         $result .= '<dl><dt>';
1676                         $this->mDTopen = true;
1677                 }
1678                 else { $result = '<!-- ERR 1 -->'; }
1679
1680                 return $result;
1681         }
1682
1683         /* private */ function nextItem( $char ) {
1684                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
1685                 else if ( ':' == $char || ';' == $char ) {
1686                         $close = '</dd>';
1687                         if ( $this->mDTopen ) { $close = '</dt>'; }
1688                         if ( ';' == $char ) {
1689                                 $this->mDTopen = true;
1690                                 return $close . '<dt>';
1691                         } else {
1692                                 $this->mDTopen = false;
1693                                 return $close . '<dd>';
1694                         }
1695                 }
1696                 return '<!-- ERR 2 -->';
1697         }
1698
1699         /* private */ function closeList( $char ) {
1700                 if ( '*' == $char ) { $text = '</li></ul>'; }
1701                 else if ( '#' == $char ) { $text = '</li></ol>'; }
1702                 else if ( ':' == $char ) {
1703                         if ( $this->mDTopen ) {
1704                                 $this->mDTopen = false;
1705                                 $text = '</dt></dl>';
1706                         } else {
1707                                 $text = '</dd></dl>';
1708                         }
1709                 }
1710                 else {  return '<!-- ERR 3 -->'; }
1711                 return $text."\n";
1712         }
1713         /**#@-*/
1714
1715         /**
1716          * Make lists from lines starting with ':', '*', '#', etc.
1717          *
1718          * @access private
1719          * @return string the lists rendered as HTML
1720          */
1721         function doBlockLevels( $text, $linestart ) {
1722                 $fname = 'Parser::doBlockLevels';
1723                 wfProfileIn( $fname );
1724
1725                 # Parsing through the text line by line.  The main thing
1726                 # happening here is handling of block-level elements p, pre,
1727                 # and making lists from lines starting with * # : etc.
1728                 #
1729                 $textLines = explode( "\n", $text );
1730
1731                 $lastPrefix = $output = '';
1732                 $this->mDTopen = $inBlockElem = false;
1733                 $prefixLength = 0;
1734                 $paragraphStack = false;
1735
1736                 if ( !$linestart ) {
1737                         $output .= array_shift( $textLines );
1738                 }
1739                 foreach ( $textLines as $oLine ) {
1740                         $lastPrefixLength = strlen( $lastPrefix );
1741                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
1742                         $preOpenMatch = preg_match('/<pre/i', $oLine );
1743                         if ( !$this->mInPre ) {
1744                                 # Multiple prefixes may abut each other for nested lists.
1745                                 $prefixLength = strspn( $oLine, '*#:;' );
1746                                 $pref = substr( $oLine, 0, $prefixLength );
1747
1748                                 # eh?
1749                                 $pref2 = str_replace( ';', ':', $pref );
1750                                 $t = substr( $oLine, $prefixLength );
1751                                 $this->mInPre = !empty($preOpenMatch);
1752                         } else {
1753                                 # Don't interpret any other prefixes in preformatted text
1754                                 $prefixLength = 0;
1755                                 $pref = $pref2 = '';
1756                                 $t = $oLine;
1757                         }
1758
1759                         # List generation
1760                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
1761                                 # Same as the last item, so no need to deal with nesting or opening stuff
1762                                 $output .= $this->nextItem( substr( $pref, -1 ) );
1763                                 $paragraphStack = false;
1764
1765                                 if ( substr( $pref, -1 ) == ';') {
1766                                         # The one nasty exception: definition lists work like this:
1767                                         # ; title : definition text
1768                                         # So we check for : in the remainder text to split up the
1769                                         # title and definition, without b0rking links.
1770                                         $term = $t2 = '';
1771                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
1772                                                 $t = $t2;
1773                                                 $output .= $term . $this->nextItem( ':' );
1774                                         }
1775                                 }
1776                         } elseif( $prefixLength || $lastPrefixLength ) {
1777                                 # Either open or close a level...
1778                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
1779                                 $paragraphStack = false;
1780
1781                                 while( $commonPrefixLength < $lastPrefixLength ) {
1782                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
1783                                         --$lastPrefixLength;
1784                                 }
1785                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
1786                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
1787                                 }
1788                                 while ( $prefixLength > $commonPrefixLength ) {
1789                                         $char = substr( $pref, $commonPrefixLength, 1 );
1790                                         $output .= $this->openList( $char );
1791
1792                                         if ( ';' == $char ) {
1793                                                 # FIXME: This is dupe of code above
1794                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
1795                                                         $t = $t2;
1796                                                         $output .= $term . $this->nextItem( ':' );
1797                                                 }
1798                                         }
1799                                         ++$commonPrefixLength;
1800                                 }
1801                                 $lastPrefix = $pref2;
1802                         }
1803                         if( 0 == $prefixLength ) {
1804                                 wfProfileIn( "$fname-paragraph" );
1805                                 # No prefix (not in list)--go to paragraph mode
1806                                 // XXX: use a stack for nestable elements like span, table and div
1807                                 $openmatch = preg_match('/(<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
1808                                 $closematch = preg_match(
1809                                         '/(<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
1810                                         '<td|<th|<div|<\\/div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul)/iS', $t );
1811                                 if ( $openmatch or $closematch ) {
1812                                         $paragraphStack = false;
1813                                         $output .= $this->closeParagraph();
1814                                         if ( $preOpenMatch and !$preCloseMatch ) {
1815                                                 $this->mInPre = true;
1816                                         }
1817                                         if ( $closematch ) {
1818                                                 $inBlockElem = false;
1819                                         } else {
1820                                                 $inBlockElem = true;
1821                                         }
1822                                 } else if ( !$inBlockElem && !$this->mInPre ) {
1823                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
1824                                                 // pre
1825                                                 if ($this->mLastSection != 'pre') {
1826                                                         $paragraphStack = false;
1827                                                         $output .= $this->closeParagraph().'<pre>';
1828                                                         $this->mLastSection = 'pre';
1829                                                 }
1830                                                 $t = substr( $t, 1 );
1831                                         } else {
1832                                                 // paragraph
1833                                                 if ( '' == trim($t) ) {
1834                                                         if ( $paragraphStack ) {
1835                                                                 $output .= $paragraphStack.'<br />';
1836                                                                 $paragraphStack = false;
1837                                                                 $this->mLastSection = 'p';
1838                                                         } else {
1839                                                                 if ($this->mLastSection != 'p' ) {
1840                                                                         $output .= $this->closeParagraph();
1841                                                                         $this->mLastSection = '';
1842                                                                         $paragraphStack = '<p>';
1843                                                                 } else {
1844                                                                         $paragraphStack = '</p><p>';
1845                                                                 }
1846                                                         }
1847                                                 } else {
1848                                                         if ( $paragraphStack ) {
1849                                                                 $output .= $paragraphStack;
1850                                                                 $paragraphStack = false;
1851                                                                 $this->mLastSection = 'p';
1852                                                         } else if ($this->mLastSection != 'p') {
1853                                                                 $output .= $this->closeParagraph().'<p>';
1854                                                                 $this->mLastSection = 'p';
1855                                                         }
1856                                                 }
1857                                         }
1858                                 }
1859                                 wfProfileOut( "$fname-paragraph" );
1860                         }
1861                         // somewhere above we forget to get out of pre block (bug 785)
1862                         if($preCloseMatch && $this->mInPre) {
1863                                 $this->mInPre = false;
1864                         }
1865                         if ($paragraphStack === false) {
1866                                 $output .= $t."\n";
1867                         }
1868                 }
1869                 while ( $prefixLength ) {
1870                         $output .= $this->closeList( $pref2{$prefixLength-1} );
1871                         --$prefixLength;
1872                 }
1873                 if ( '' != $this->mLastSection ) {
1874                         $output .= '</' . $this->mLastSection . '>';
1875                         $this->mLastSection = '';
1876                 }
1877
1878                 wfProfileOut( $fname );
1879                 return $output;
1880         }
1881
1882         /**
1883          * Split up a string on ':', ignoring any occurences inside
1884          * <a>..</a> or <span>...</span>
1885          * @param string $str the string to split
1886          * @param string &$before set to everything before the ':'
1887          * @param string &$after set to everything after the ':'
1888          * return string the position of the ':', or false if none found
1889          */
1890         function findColonNoLinks($str, &$before, &$after) {
1891                 # I wonder if we should make this count all tags, not just <a>
1892                 # and <span>. That would prevent us from matching a ':' that
1893                 # comes in the middle of italics other such formatting....
1894                 # -- Wil
1895                 $fname = 'Parser::findColonNoLinks';
1896                 wfProfileIn( $fname );
1897                 $pos = 0;
1898                 do {
1899                         $colon = strpos($str, ':', $pos);
1900
1901                         if ($colon !== false) {
1902                                 $before = substr($str, 0, $colon);
1903                                 $after = substr($str, $colon + 1);
1904
1905                                 # Skip any ':' within <a> or <span> pairs
1906                                 $a = substr_count($before, '<a');
1907                                 $s = substr_count($before, '<span');
1908                                 $ca = substr_count($before, '</a>');
1909                                 $cs = substr_count($before, '</span>');
1910
1911                                 if ($a <= $ca and $s <= $cs) {
1912                                         # Tags are balanced before ':'; ok
1913                                         break;
1914                                 }
1915                                 $pos = $colon + 1;
1916                         }
1917                 } while ($colon !== false);
1918                 wfProfileOut( $fname );
1919                 return $colon;
1920         }
1921
1922         /**
1923          * Return value of a magic variable (like PAGENAME)
1924          *
1925          * @access private
1926          */
1927         function getVariableValue( $index ) {
1928                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
1929
1930                 /**
1931                  * Some of these require message or data lookups and can be
1932                  * expensive to check many times.
1933                  */
1934                 static $varCache = array();
1935                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$varCache ) ) )
1936                         if ( isset( $varCache[$index] ) )
1937                                 return $varCache[$index];
1938
1939                 $ts = time();
1940                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
1941
1942                 switch ( $index ) {
1943                         case MAG_CURRENTMONTH:
1944                                 return $varCache[$index] = $wgContLang->formatNum( date( 'm', $ts ) );
1945                         case MAG_CURRENTMONTHNAME:
1946                                 return $varCache[$index] = $wgContLang->getMonthName( date( 'n', $ts ) );
1947                         case MAG_CURRENTMONTHNAMEGEN:
1948                                 return $varCache[$index] = $wgContLang->getMonthNameGen( date( 'n', $ts ) );
1949                         case MAG_CURRENTMONTHABBREV:
1950                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( date( 'n', $ts ) );
1951                         case MAG_CURRENTDAY:
1952                                 return $varCache[$index] = $wgContLang->formatNum( date( 'j', $ts ) );
1953                         case MAG_PAGENAME:
1954                                 return $this->mTitle->getText();
1955                         case MAG_PAGENAMEE:
1956                                 return $this->mTitle->getPartialURL();
1957                         case MAG_FULLPAGENAME:
1958                                 return $this->mTitle->getPrefixedText();
1959                         case MAG_FULLPAGENAMEE:
1960                                 return wfUrlencode( $this->mTitle->getPrefixedText() );
1961                         case MAG_REVISIONID:
1962                                 return $this->mRevisionId;
1963                         case MAG_NAMESPACE:
1964                                 return $wgContLang->getNsText( $this->mTitle->getNamespace() );
1965                         case MAG_NAMESPACEE:
1966                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
1967                         case MAG_CURRENTDAYNAME:
1968                                 return $varCache[$index] = $wgContLang->getWeekdayName( date( 'w', $ts ) + 1 );
1969                         case MAG_CURRENTYEAR:
1970                                 return $varCache[$index] = $wgContLang->formatNum( date( 'Y', $ts ), true );
1971                         case MAG_CURRENTTIME:
1972                                 return $varCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
1973                         case MAG_CURRENTWEEK:
1974                                 return $varCache[$index] = $wgContLang->formatNum( date( 'W', $ts ) );
1975                         case MAG_CURRENTDOW:
1976                                 return $varCache[$index] = $wgContLang->formatNum( date( 'w', $ts ) );
1977                         case MAG_NUMBEROFARTICLES:
1978                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfArticles() );
1979                         case MAG_NUMBEROFFILES:
1980                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfFiles() );
1981                         case MAG_SITENAME:
1982                                 return $wgSitename;
1983                         case MAG_SERVER:
1984                                 return $wgServer;
1985                         case MAG_SERVERNAME:
1986                                 return $wgServerName;
1987                         case MAG_SCRIPTPATH:
1988                                 return $wgScriptPath;
1989                         default:
1990                                 $ret = null;
1991                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$varCache, &$index, &$ret ) ) )
1992                                         return $ret;
1993                                 else
1994                                         return null;
1995                 }
1996         }
1997
1998         /**
1999          * initialise the magic variables (like CURRENTMONTHNAME)
2000          *
2001          * @access private
2002          */
2003         function initialiseVariables() {
2004                 $fname = 'Parser::initialiseVariables';
2005                 wfProfileIn( $fname );
2006                 global $wgVariableIDs;
2007                 $this->mVariables = array();
2008                 foreach ( $wgVariableIDs as $id ) {
2009                         $mw =& MagicWord::get( $id );
2010                         $mw->addToArray( $this->mVariables, $id );
2011                 }
2012                 wfProfileOut( $fname );
2013         }
2014
2015         /**
2016          * parse any parentheses in format ((title|part|part))
2017          * and call callbacks to get a replacement text for any found piece
2018          *
2019          * @param string $text The text to parse
2020          * @param array $callbacks rules in form:
2021          *     '{' => array(                            # opening parentheses
2022          *                                      'end' => '}',   # closing parentheses
2023          *                                      'cb' => array(2 => callback,    # replacement callback to call if {{..}} is found
2024          *                                                                4 => callback         # replacement callback to call if {{{{..}}}} is found
2025          *                                                                )
2026          *                                      )
2027          * @access private
2028          */
2029         function replace_callback ($text, $callbacks) {
2030                 $openingBraceStack = array();   # this array will hold a stack of parentheses which are not closed yet
2031                 $lastOpeningBrace = -1;                 # last not closed parentheses
2032
2033                 for ($i = 0; $i < strlen($text); $i++) {
2034                         # check for any opening brace
2035                         $rule = null;
2036                         $nextPos = -1;
2037                         foreach ($callbacks as $key => $value) {
2038                                 $pos = strpos ($text, $key, $i);
2039                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)) {
2040                                         $rule = $value;
2041                                         $nextPos = $pos;
2042                                 }
2043                         }
2044
2045                         if ($lastOpeningBrace >= 0) {
2046                                 $pos = strpos ($text, $openingBraceStack[$lastOpeningBrace]['braceEnd'], $i);
2047
2048                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)){
2049                                         $rule = null;
2050                                         $nextPos = $pos;
2051                                 }
2052
2053                                 $pos = strpos ($text, '|', $i);
2054
2055                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)){
2056                                         $rule = null;
2057                                         $nextPos = $pos;
2058                                 }
2059                         }
2060
2061                         if ($nextPos == -1)
2062                                 break;
2063
2064                         $i = $nextPos;
2065
2066                         # found openning brace, lets add it to parentheses stack
2067                         if (null != $rule) {
2068                                 $piece = array('brace' => $text[$i],
2069                                                            'braceEnd' => $rule['end'],
2070                                                            'count' => 1,
2071                                                            'title' => '',
2072                                                            'parts' => null);
2073
2074                                 # count openning brace characters
2075                                 while ($i+1 < strlen($text) && $text[$i+1] == $piece['brace']) {
2076                                         $piece['count']++;
2077                                         $i++;
2078                                 }
2079
2080                                 $piece['startAt'] = $i+1;
2081                                 $piece['partStart'] = $i+1;
2082
2083                                 # we need to add to stack only if openning brace count is enough for any given rule
2084                                 foreach ($rule['cb'] as $cnt => $fn) {
2085                                         if ($piece['count'] >= $cnt) {
2086                                                 $lastOpeningBrace ++;
2087                                                 $openingBraceStack[$lastOpeningBrace] = $piece;
2088                                                 break;
2089                                         }
2090                                 }
2091
2092                                 continue;
2093                         }
2094                         else if ($lastOpeningBrace >= 0) {
2095                                 # first check if it is a closing brace
2096                                 if ($openingBraceStack[$lastOpeningBrace]['braceEnd'] == $text[$i]) {
2097                                         # lets check if it is enough characters for closing brace
2098                                         $count = 1;
2099                                         while ($i+$count < strlen($text) && $text[$i+$count] == $text[$i])
2100                                                 $count++;
2101
2102                                         # if there are more closing parentheses than opening ones, we parse less
2103                                         if ($openingBraceStack[$lastOpeningBrace]['count'] < $count)
2104                                                 $count = $openingBraceStack[$lastOpeningBrace]['count'];
2105
2106                                         # check for maximum matching characters (if there are 5 closing characters, we will probably need only 3 - depending on the rules)
2107                                         $matchingCount = 0;
2108                                         $matchingCallback = null;
2109                                         foreach ($callbacks[$openingBraceStack[$lastOpeningBrace]['brace']]['cb'] as $cnt => $fn) {
2110                                                 if ($count >= $cnt && $matchingCount < $cnt) {
2111                                                         $matchingCount = $cnt;
2112                                                         $matchingCallback = $fn;
2113                                                 }
2114                                         }
2115
2116                                         if ($matchingCount == 0) {
2117                                                 $i += $count - 1;
2118                                                 continue;
2119                                         }
2120
2121                                         # lets set a title or last part (if '|' was found)
2122                                         if (null === $openingBraceStack[$lastOpeningBrace]['parts'])
2123                                                 $openingBraceStack[$lastOpeningBrace]['title'] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2124                                         else
2125                                                 $openingBraceStack[$lastOpeningBrace]['parts'][] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2126
2127                                         $pieceStart = $openingBraceStack[$lastOpeningBrace]['startAt'] - $matchingCount;
2128                                         $pieceEnd = $i + $matchingCount;
2129
2130                                         if( is_callable( $matchingCallback ) ) {
2131                                                 $cbArgs = array (
2132                                                                                  'text' => substr($text, $pieceStart, $pieceEnd - $pieceStart),
2133                                                                                  'title' => trim($openingBraceStack[$lastOpeningBrace]['title']),
2134                                                                                  'parts' => $openingBraceStack[$lastOpeningBrace]['parts'],
2135                                                                                  'lineStart' => (($pieceStart > 0) && ($text[$pieceStart-1] == '\n')),
2136                                                                                  );
2137                                                 # finally we can call a user callback and replace piece of text
2138                                                 $replaceWith = call_user_func( $matchingCallback, $cbArgs );
2139                                                 $text = substr($text, 0, $pieceStart) . $replaceWith . substr($text, $pieceEnd);
2140                                                 $i = $pieceStart + strlen($replaceWith) - 1;
2141                                         }
2142                                         else {
2143                                                 # null value for callback means that parentheses should be parsed, but not replaced
2144                                                 $i += $matchingCount - 1;
2145                                         }
2146
2147                                         # reset last openning parentheses, but keep it in case there are unused characters
2148                                         $piece = array('brace' => $openingBraceStack[$lastOpeningBrace]['brace'],
2149                                                                    'braceEnd' => $openingBraceStack[$lastOpeningBrace]['braceEnd'],
2150                                                                    'count' => $openingBraceStack[$lastOpeningBrace]['count'],
2151                                                                    'title' => '',
2152                                                                    'parts' => null,
2153                                                                    'startAt' => $openingBraceStack[$lastOpeningBrace]['startAt']);
2154                                         $openingBraceStack[$lastOpeningBrace--] = null;
2155
2156                                         if ($matchingCount < $piece['count']) {
2157                                                 $piece['count'] -= $matchingCount;
2158                                                 $piece['startAt'] -= $matchingCount;
2159                                                 $piece['partStart'] = $piece['startAt'];
2160                                                 # do we still qualify for any callback with remaining count?
2161                                                 foreach ($callbacks[$piece['brace']]['cb'] as $cnt => $fn) {
2162                                                         if ($piece['count'] >= $cnt) {
2163                                                                 $lastOpeningBrace ++;
2164                                                                 $openingBraceStack[$lastOpeningBrace] = $piece;
2165                                                                 break;
2166                                                         }
2167                                                 }
2168                                         }
2169                                         continue;
2170                                 }
2171
2172                                 # lets set a title if it is a first separator, or next part otherwise
2173                                 if ($text[$i] == '|') {
2174                                         if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2175                                                 $openingBraceStack[$lastOpeningBrace]['title'] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2176                                                 $openingBraceStack[$lastOpeningBrace]['parts'] = array();
2177                                         }
2178                                         else
2179                                                 $openingBraceStack[$lastOpeningBrace]['parts'][] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2180
2181                                         $openingBraceStack[$lastOpeningBrace]['partStart'] = $i + 1;
2182                                 }
2183                         }
2184                 }
2185
2186                 return $text;
2187         }
2188
2189         /**
2190          * Replace magic variables, templates, and template arguments
2191          * with the appropriate text. Templates are substituted recursively,
2192          * taking care to avoid infinite loops.
2193          *
2194          * Note that the substitution depends on value of $mOutputType:
2195          *  OT_WIKI: only {{subst:}} templates
2196          *  OT_MSG: only magic variables
2197          *  OT_HTML: all templates and magic variables
2198          *
2199          * @param string $tex The text to transform
2200          * @param array $args Key-value pairs representing template parameters to substitute
2201          * @access private
2202          */
2203         function replaceVariables( $text, $args = array() ) {
2204                 # Prevent too big inclusions
2205                 if( strlen( $text ) > MAX_INCLUDE_SIZE ) {
2206                         return $text;
2207                 }
2208
2209                 $fname = 'Parser::replaceVariables';
2210                 wfProfileIn( $fname );
2211
2212                 # This function is called recursively. To keep track of arguments we need a stack:
2213                 array_push( $this->mArgStack, $args );
2214
2215                 $braceCallbacks = array();
2216                 $braceCallbacks[2] = array( &$this, 'braceSubstitution' );
2217                 if ( $this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI ) {
2218                         $braceCallbacks[3] = array( &$this, 'argSubstitution' );
2219                 }
2220                 $callbacks = array();
2221                 $callbacks['{'] = array('end' => '}', 'cb' => $braceCallbacks);
2222                 $callbacks['['] = array('end' => ']', 'cb' => array(2=>null));
2223                 $text = $this->replace_callback ($text, $callbacks);
2224
2225                 array_pop( $this->mArgStack );
2226
2227                 wfProfileOut( $fname );
2228                 return $text;
2229         }
2230
2231         /**
2232          * Replace magic variables
2233          * @access private
2234          */
2235         function variableSubstitution( $matches ) {
2236                 $fname = 'Parser::variableSubstitution';
2237                 $varname = $matches[1];
2238                 wfProfileIn( $fname );
2239                 if ( !$this->mVariables ) {
2240                         $this->initialiseVariables();
2241                 }
2242                 $skip = false;
2243                 if ( $this->mOutputType == OT_WIKI ) {
2244                         # Do only magic variables prefixed by SUBST
2245                         $mwSubst =& MagicWord::get( MAG_SUBST );
2246                         if (!$mwSubst->matchStartAndRemove( $varname ))
2247                                 $skip = true;
2248                         # Note that if we don't substitute the variable below,
2249                         # we don't remove the {{subst:}} magic word, in case
2250                         # it is a template rather than a magic variable.
2251                 }
2252                 if ( !$skip && array_key_exists( $varname, $this->mVariables ) ) {
2253                         $id = $this->mVariables[$varname];
2254                         $text = $this->getVariableValue( $id );
2255                         $this->mOutput->mContainsOldMagic = true;
2256                 } else {
2257                         $text = $matches[0];
2258                 }
2259                 wfProfileOut( $fname );
2260                 return $text;
2261         }
2262
2263         # Split template arguments
2264         function getTemplateArgs( $argsString ) {
2265                 if ( $argsString === '' ) {
2266                         return array();
2267                 }
2268
2269                 $args = explode( '|', substr( $argsString, 1 ) );
2270
2271                 # If any of the arguments contains a '[[' but no ']]', it needs to be
2272                 # merged with the next arg because the '|' character between belongs
2273                 # to the link syntax and not the template parameter syntax.
2274                 $argc = count($args);
2275
2276                 for ( $i = 0; $i < $argc-1; $i++ ) {
2277                         if ( substr_count ( $args[$i], '[[' ) != substr_count ( $args[$i], ']]' ) ) {
2278                                 $args[$i] .= '|'.$args[$i+1];
2279                                 array_splice($args, $i+1, 1);
2280                                 $i--;
2281                                 $argc--;
2282                         }
2283                 }
2284
2285                 return $args;
2286         }
2287
2288         /**
2289          * Return the text of a template, after recursively
2290          * replacing any variables or templates within the template.
2291          *
2292          * @param array $piece The parts of the template
2293          *  $piece['text']: matched text
2294          *  $piece['title']: the title, i.e. the part before the |
2295          *  $piece['parts']: the parameter array
2296          * @return string the text of the template
2297          * @access private
2298          */
2299         function braceSubstitution( $piece ) {
2300                 global $wgLinkCache, $wgContLang;
2301                 $fname = 'Parser::braceSubstitution';
2302                 wfProfileIn( $fname );
2303
2304                 $found = false;
2305                 $nowiki = false;
2306                 $noparse = false;
2307
2308                 $title = NULL;
2309
2310                 $linestart = '';
2311
2312                 # $part1 is the bit before the first |, and must contain only title characters
2313                 # $args is a list of arguments, starting from index 0, not including $part1
2314
2315                 $part1 = $piece['title'];
2316                 # If the third subpattern matched anything, it will start with |
2317
2318                 if (null == $piece['parts']) {
2319                         $replaceWith = $this->variableSubstitution (array ($piece['text'], $piece['title']));
2320                         if ($replaceWith != $piece['text']) {
2321                                 $text = $replaceWith;
2322                                 $found = true;
2323                                 $noparse = true;
2324                         }
2325                 }
2326
2327                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2328                 $argc = count( $args );
2329
2330                 # SUBST
2331                 if ( !$found ) {
2332                         $mwSubst =& MagicWord::get( MAG_SUBST );
2333                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor ($this->mOutputType == OT_WIKI) ) {
2334                                 # One of two possibilities is true:
2335                                 # 1) Found SUBST but not in the PST phase
2336                                 # 2) Didn't find SUBST and in the PST phase
2337                                 # In either case, return without further processing
2338                                 $text = $piece['text'];
2339                                 $found = true;
2340                                 $noparse = true;
2341                         }
2342                 }
2343
2344                 # MSG, MSGNW and INT
2345                 if ( !$found ) {
2346                         # Check for MSGNW:
2347                         $mwMsgnw =& MagicWord::get( MAG_MSGNW );
2348                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2349                                 $nowiki = true;
2350                         } else {
2351                                 # Remove obsolete MSG:
2352                                 $mwMsg =& MagicWord::get( MAG_MSG );
2353                                 $mwMsg->matchStartAndRemove( $part1 );
2354                         }
2355
2356                         # Check if it is an internal message
2357                         $mwInt =& MagicWord::get( MAG_INT );
2358                         if ( $mwInt->matchStartAndRemove( $part1 ) ) {
2359                                 if ( $this->incrementIncludeCount( 'int:'.$part1 ) ) {
2360                                         $text = $linestart . wfMsgReal( $part1, $args, true );
2361                                         $found = true;
2362                                 }
2363                         }
2364                 }
2365
2366                 # NS
2367                 if ( !$found ) {
2368                         # Check for NS: (namespace expansion)
2369                         $mwNs = MagicWord::get( MAG_NS );
2370                         if ( $mwNs->matchStartAndRemove( $part1 ) ) {
2371                                 if ( intval( $part1 ) ) {
2372                                         $text = $linestart . $wgContLang->getNsText( intval( $part1 ) );
2373                                         $found = true;
2374                                 } else {
2375                                         $index = Namespace::getCanonicalIndex( strtolower( $part1 ) );
2376                                         if ( !is_null( $index ) ) {
2377                                                 $text = $linestart . $wgContLang->getNsText( $index );
2378                                                 $found = true;
2379                                         }
2380                                 }
2381                         }
2382                 }
2383
2384                 # LCFIRST, UCFIRST, LC and UC
2385                 if ( !$found ) {
2386                         $lcfirst =& MagicWord::get( MAG_LCFIRST );
2387                         $ucfirst =& MagicWord::get( MAG_UCFIRST );
2388                         $lc =& MagicWord::get( MAG_LC );
2389                         $uc =& MagicWord::get( MAG_UC );
2390                         if ( $lcfirst->matchStartAndRemove( $part1 ) ) {
2391                                 $text = $linestart . $wgContLang->lcfirst( $part1 );
2392                                 $found = true;
2393                         } else if ( $ucfirst->matchStartAndRemove( $part1 ) ) {
2394                                 $text = $linestart . $wgContLang->ucfirst( $part1 );
2395                                 $found = true;
2396                         } else if ( $lc->matchStartAndRemove( $part1 ) ) {
2397                                 $text = $linestart . $wgContLang->lc( $part1 );
2398                                 $found = true;
2399                         } else if ( $uc->matchStartAndRemove( $part1 ) ) {
2400                                  $text = $linestart . $wgContLang->uc( $part1 );
2401                                  $found = true;
2402                         }
2403                 }
2404
2405                 # LOCALURL and FULLURL
2406                 if ( !$found ) {
2407                         $mwLocal =& MagicWord::get( MAG_LOCALURL );
2408                         $mwLocalE =& MagicWord::get( MAG_LOCALURLE );
2409                         $mwFull =& MagicWord::get( MAG_FULLURL );
2410                         $mwFullE =& MagicWord::get( MAG_FULLURLE );
2411
2412
2413                         if ( $mwLocal->matchStartAndRemove( $part1 ) ) {
2414                                 $func = 'getLocalURL';
2415                         } elseif ( $mwLocalE->matchStartAndRemove( $part1 ) ) {
2416                                 $func = 'escapeLocalURL';
2417                         } elseif ( $mwFull->matchStartAndRemove( $part1 ) ) {
2418                                 $func = 'getFullURL';
2419                         } elseif ( $mwFullE->matchStartAndRemove( $part1 ) ) {
2420                                 $func = 'escapeFullURL';
2421                         } else {
2422                                 $func = false;
2423                         }
2424
2425                         if ( $func !== false ) {
2426                                 $title = Title::newFromText( $part1 );
2427                                 if ( !is_null( $title ) ) {
2428                                         if ( $argc > 0 ) {
2429                                                 $text = $linestart . $title->$func( $args[0] );
2430                                         } else {
2431                                                 $text = $linestart . $title->$func();
2432                                         }
2433                                         $found = true;
2434                                 }
2435                         }
2436                 }
2437
2438                 # GRAMMAR
2439                 if ( !$found && $argc == 1 ) {
2440                         $mwGrammar =& MagicWord::get( MAG_GRAMMAR );
2441                         if ( $mwGrammar->matchStartAndRemove( $part1 ) ) {
2442                                 $text = $linestart . $wgContLang->convertGrammar( $args[0], $part1 );
2443                                 $found = true;
2444                         }
2445                 }
2446
2447                 # PLURAL
2448                 if ( !$found && $argc >= 2 ) {
2449                         $mwPluralForm =& MagicWord::get( MAG_PLURAL );
2450                         if ( $mwPluralForm->matchStartAndRemove( $part1 ) ) {
2451                                 if ($argc==2) {$args[2]=$args[1];}
2452                                 $text = $linestart . $wgContLang->convertPlural( $part1, $args[0], $args[1], $args[2]);
2453                                 $found = true;
2454                         }
2455                 }
2456
2457                 # Template table test
2458
2459                 # Did we encounter this template already? If yes, it is in the cache
2460                 # and we need to check for loops.
2461                 if ( !$found && isset( $this->mTemplates[$part1] ) ) {
2462                         $found = true;
2463
2464                         # Infinite loop test
2465                         if ( isset( $this->mTemplatePath[$part1] ) ) {
2466                                 $noparse = true;
2467                                 $found = true;
2468                                 $text = $linestart .
2469                                         '{{' . $part1 . '}}' .
2470                                         '<!-- WARNING: template loop detected -->';
2471                                 wfDebug( "$fname: template loop broken at '$part1'\n" );
2472                         } else {
2473                                 # set $text to cached message.
2474                                 $text = $linestart . $this->mTemplates[$part1];
2475                         }
2476                 }
2477
2478                 # Load from database
2479                 $replaceHeadings = false;
2480                 $isHTML = false;
2481                 $lastPathLevel = $this->mTemplatePath;
2482                 if ( !$found ) {
2483                         $ns = NS_TEMPLATE;
2484                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage='' );
2485                         if ($subpage !== '') {
2486                                 $ns = $this->mTitle->getNamespace();
2487                         }
2488                         $title = Title::newFromText( $part1, $ns );
2489
2490                         if ($title) {
2491                                 $interwiki = Title::getInterwikiLink($title->getInterwiki());
2492                                 if ($interwiki != '' && $title->isTrans()) {
2493                                         return $this->scarytransclude($title, $interwiki);
2494                                 }
2495                         }
2496
2497                         if ( !is_null( $title ) && !$title->isExternal() ) {
2498                                 # Check for excessive inclusion
2499                                 $dbk = $title->getPrefixedDBkey();
2500                                 if ( $this->incrementIncludeCount( $dbk ) ) {
2501                                         if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() ) {
2502                                                 # Capture special page output
2503                                                 $text = SpecialPage::capturePath( $title );
2504                                                 if ( is_string( $text ) ) {
2505                                                         $found = true;
2506                                                         $noparse = true;
2507                                                         $isHTML = true;
2508                                                         $this->disableCache();
2509                                                 }
2510                                         } else {
2511                                                 $article = new Article( $title );
2512                                                 $articleContent = $article->fetchContent(0, false);
2513                                                 if ( $articleContent !== false ) {
2514                                                         $found = true;
2515                                                         $text = $articleContent;
2516                                                         $replaceHeadings = true;
2517                                                 }
2518                                         }
2519                                 }
2520
2521                                 # If the title is valid but undisplayable, make a link to it
2522                                 if ( $this->mOutputType == OT_HTML && !$found ) {
2523                                         $text = '[['.$title->getPrefixedText().']]';
2524                                         $found = true;
2525                                 }
2526
2527                                 # Template cache array insertion
2528                                 if( $found ) {
2529                                         $this->mTemplates[$part1] = $text;
2530                                         $text = $linestart . $text;
2531                                 }
2532                         }
2533                 }
2534
2535                 # Recursive parsing, escaping and link table handling
2536                 # Only for HTML output
2537                 if ( $nowiki && $found && $this->mOutputType == OT_HTML ) {
2538                         $text = wfEscapeWikiText( $text );
2539                 } elseif ( ($this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI) && $found && !$noparse) {
2540                         # Clean up argument array
2541                         $assocArgs = array();
2542                         $index = 1;
2543                         foreach( $args as $arg ) {
2544                                 $eqpos = strpos( $arg, '=' );
2545                                 if ( $eqpos === false ) {
2546                                         $assocArgs[$index++] = $arg;
2547                                 } else {
2548                                         $name = trim( substr( $arg, 0, $eqpos ) );
2549                                         $value = trim( substr( $arg, $eqpos+1 ) );
2550                                         if ( $value === false ) {
2551                                                 $value = '';
2552                                         }
2553                                         if ( $name !== false ) {
2554                                                 $assocArgs[$name] = $value;
2555                                         }
2556                                 }
2557                         }
2558
2559                         # Add a new element to the templace recursion path
2560                         $this->mTemplatePath[$part1] = 1;
2561
2562                         # If there are any <onlyinclude> tags, only include them
2563                         if ( in_string( '<onlyinclude>', $text ) && in_string( '</onlyinclude>', $text ) ) {
2564                                 preg_match_all( '/<onlyinclude>(.*?)\n?<\/onlyinclude>/s', $text, $m );
2565                                 $text = '';
2566                                 foreach ($m[1] as $piece)
2567                                         $text .= $piece;
2568                         }
2569                         # Remove <noinclude> sections and <includeonly> tags
2570                         $text = preg_replace( '/<noinclude>.*?<\/noinclude>/s', '', $text );
2571                         $text = strtr( $text, array( '<includeonly>' => '' , '</includeonly>' => '' ) );
2572
2573                         if( $this->mOutputType == OT_HTML ) {
2574                                 # Strip <nowiki>, <pre>, etc.
2575                                 $text = $this->strip( $text, $this->mStripState );
2576                                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'replaceVariables' ), $assocArgs );
2577                         }
2578                         $text = $this->replaceVariables( $text, $assocArgs );
2579
2580                         # Resume the link cache and register the inclusion as a link
2581                         if ( $this->mOutputType == OT_HTML && !is_null( $title ) ) {
2582                                 $wgLinkCache->addLinkObj( $title );
2583                         }
2584
2585                         # If the template begins with a table or block-level
2586                         # element, it should be treated as beginning a new line.
2587                         if (!$piece['lineStart'] && preg_match('/^({\\||:|;|#|\*)/', $text)) {
2588                                 $text = "\n" . $text;
2589                         }
2590                 }
2591                 # Prune lower levels off the recursion check path
2592                 $this->mTemplatePath = $lastPathLevel;
2593
2594                 if ( !$found ) {
2595                         wfProfileOut( $fname );
2596                         return $piece['text'];
2597                 } else {
2598                         if ( $isHTML ) {
2599                                 # Replace raw HTML by a placeholder
2600                                 # Add a blank line preceding, to prevent it from mucking up
2601                                 # immediately preceding headings
2602                                 $text = "\n\n" . $this->insertStripItem( $text, $this->mStripState );
2603                         } else {
2604                                 # replace ==section headers==
2605                                 # XXX this needs to go away once we have a better parser.
2606                                 if ( $this->mOutputType != OT_WIKI && $replaceHeadings ) {
2607                                         if( !is_null( $title ) )
2608                                                 $encodedname = base64_encode($title->getPrefixedDBkey());
2609                                         else
2610                                                 $encodedname = base64_encode("");
2611                                         $m = preg_split('/(^={1,6}.*?={1,6}\s*?$)/m', $text, -1,
2612                                                 PREG_SPLIT_DELIM_CAPTURE);
2613                                         $text = '';
2614                                         $nsec = 0;
2615                                         for( $i = 0; $i < count($m); $i += 2 ) {
2616                                                 $text .= $m[$i];
2617                                                 if (!isset($m[$i + 1]) || $m[$i + 1] == "") continue;
2618                                                 $hl = $m[$i + 1];
2619                                                 if( strstr($hl, "<!--MWTEMPLATESECTION") ) {
2620                                                         $text .= $hl;
2621                                                         continue;
2622                                                 }
2623                                                 preg_match('/^(={1,6})(.*?)(={1,6})\s*?$/m', $hl, $m2);
2624                                                 $text .= $m2[1] . $m2[2] . "<!--MWTEMPLATESECTION="
2625                                                         . $encodedname . "&" . base64_encode("$nsec") . "-->" . $m2[3];
2626
2627                                                 $nsec++;
2628                                         }
2629                                 }
2630                         }
2631                 }
2632
2633                 # Prune lower levels off the recursion check path
2634                 $this->mTemplatePath = $lastPathLevel;
2635
2636                 if ( !$found ) {
2637                         wfProfileOut( $fname );
2638                         return $piece['text'];
2639                 } else {
2640                         wfProfileOut( $fname );
2641                         return $text;
2642                 }
2643         }
2644
2645         /**
2646          * Translude an interwiki link.
2647          */
2648         function scarytransclude($title, $interwiki) {
2649                 global $wgEnableScaryTranscluding;
2650
2651                 if (!$wgEnableScaryTranscluding)
2652                         return wfMsg('scarytranscludedisabled');
2653
2654                 $articlename = "Template:" . $title->getDBkey();
2655                 $url = str_replace('$1', urlencode($articlename), $interwiki);
2656                 if (strlen($url) > 255)
2657                         return wfMsg('scarytranscludetoolong');
2658                 $text = $this->fetchScaryTemplateMaybeFromCache($url);
2659                 $this->mIWTransData[] = $text;
2660                 return "<!--IW_TRANSCLUDE ".(count($this->mIWTransData) - 1)."-->";
2661         }
2662
2663         function fetchScaryTemplateMaybeFromCache($url) {
2664                 $dbr =& wfGetDB(DB_SLAVE);
2665                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
2666                                 array('tc_url' => $url));
2667                 if ($obj) {
2668                         $time = $obj->tc_time;
2669                         $text = $obj->tc_contents;
2670                         if ($time && $time < (time() + (60*60))) {
2671                                 return $text;
2672                         }
2673                 }
2674
2675                 $text = wfGetHTTP($url . '?action=render');
2676                 if (!$text)
2677                         return wfMsg('scarytranscludefailed', $url);
2678
2679                 $dbw =& wfGetDB(DB_MASTER);
2680                 $dbw->replace('transcache', array(), array(
2681                         'tc_url' => $url,
2682                         'tc_time' => time(),
2683                         'tc_contents' => $text));
2684                 return $text;
2685         }
2686
2687
2688         /**
2689          * Triple brace replacement -- used for template arguments
2690          * @access private
2691          */
2692         function argSubstitution( $matches ) {
2693                 $arg = trim( $matches['title'] );
2694                 $text = $matches['text'];
2695                 $inputArgs = end( $this->mArgStack );
2696
2697                 if ( array_key_exists( $arg, $inputArgs ) ) {
2698                         $text = $inputArgs[$arg];
2699                 } else if ($this->mOutputType == OT_HTML && null != $matches['parts'] && count($matches['parts']) > 0) {
2700                         $text = $matches['parts'][0];
2701                 }
2702
2703                 return $text;
2704         }
2705
2706         /**
2707          * Returns true if the function is allowed to include this entity
2708          * @access private
2709          */
2710         function incrementIncludeCount( $dbk ) {
2711                 if ( !array_key_exists( $dbk, $this->mIncludeCount ) ) {
2712                         $this->mIncludeCount[$dbk] = 0;
2713                 }
2714                 if ( ++$this->mIncludeCount[$dbk] <= MAX_INCLUDE_REPEAT ) {
2715                         return true;
2716                 } else {
2717                         return false;
2718                 }
2719         }
2720
2721         /**
2722          * This function accomplishes several tasks:
2723          * 1) Auto-number headings if that option is enabled
2724          * 2) Add an [edit] link to sections for logged in users who have enabled the option
2725          * 3) Add a Table of contents on the top for users who have enabled the option
2726          * 4) Auto-anchor headings
2727          *
2728          * It loops through all headlines, collects the necessary data, then splits up the
2729          * string and re-inserts the newly formatted headlines.
2730          *
2731          * @param string $text
2732          * @param boolean $isMain
2733          * @access private
2734          */
2735         function formatHeadings( $text, $isMain=true ) {
2736                 global $wgMaxTocLevel, $wgContLang;
2737
2738                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
2739                 $doShowToc = true;
2740                 $forceTocHere = false;
2741                 if( !$this->mTitle->userCanEdit() ) {
2742                         $showEditLink = 0;
2743                 } else {
2744                         $showEditLink = $this->mOptions->getEditSection();
2745                 }
2746
2747                 # Inhibit editsection links if requested in the page
2748                 $esw =& MagicWord::get( MAG_NOEDITSECTION );
2749                 if( $esw->matchAndRemove( $text ) ) {
2750                         $showEditLink = 0;
2751                 }
2752                 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
2753                 # do not add TOC
2754                 $mw =& MagicWord::get( MAG_NOTOC );
2755                 if( $mw->matchAndRemove( $text ) ) {
2756                         $doShowToc = false;
2757                 }
2758
2759                 # Get all headlines for numbering them and adding funky stuff like [edit]
2760                 # links - this is for later, but we need the number of headlines right now
2761                 $numMatches = preg_match_all( '/<H([1-6])(.*?'.'>)(.*?)<\/H[1-6] *>/i', $text, $matches );
2762
2763                 # if there are fewer than 4 headlines in the article, do not show TOC
2764                 if( $numMatches < 4 ) {
2765                         $doShowToc = false;
2766                 }
2767
2768                 # if the string __TOC__ (not case-sensitive) occurs in the HTML,
2769                 # override above conditions and always show TOC at that place
2770
2771                 $mw =& MagicWord::get( MAG_TOC );
2772                 if($mw->match( $text ) ) {
2773                         $doShowToc = true;
2774                         $forceTocHere = true;
2775                 } else {
2776                         # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
2777                         # override above conditions and always show TOC above first header
2778                         $mw =& MagicWord::get( MAG_FORCETOC );
2779                         if ($mw->matchAndRemove( $text ) ) {
2780                                 $doShowToc = true;
2781                         }
2782                 }
2783
2784                 # Never ever show TOC if no headers
2785                 if( $numMatches < 1 ) {
2786                         $doShowToc = false;
2787                 }
2788
2789                 # We need this to perform operations on the HTML
2790                 $sk =& $this->mOptions->getSkin();
2791
2792                 # headline counter
2793                 $headlineCount = 0;
2794                 $sectionCount = 0; # headlineCount excluding template sections
2795
2796                 # Ugh .. the TOC should have neat indentation levels which can be
2797                 # passed to the skin functions. These are determined here
2798                 $toc = '';
2799                 $full = '';
2800                 $head = array();
2801                 $sublevelCount = array();
2802                 $levelCount = array();
2803                 $toclevel = 0;
2804                 $level = 0;
2805                 $prevlevel = 0;
2806                 $toclevel = 0;
2807                 $prevtoclevel = 0;
2808
2809                 foreach( $matches[3] as $headline ) {
2810                         $istemplate = 0;
2811                         $templatetitle = '';
2812                         $templatesection = 0;
2813                         $numbering = '';
2814
2815                         if (preg_match("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", $headline, $mat)) {
2816                                 $istemplate = 1;
2817                                 $templatetitle = base64_decode($mat[1]);
2818                                 $templatesection = 1 + (int)base64_decode($mat[2]);
2819                                 $headline = preg_replace("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", "", $headline);
2820                         }
2821
2822                         if( $toclevel ) {
2823                                 $prevlevel = $level;
2824                                 $prevtoclevel = $toclevel;
2825                         }
2826                         $level = $matches[1][$headlineCount];
2827
2828                         if( $doNumberHeadings || $doShowToc ) {
2829
2830                                 if ( $level > $prevlevel ) {
2831                                         # Increase TOC level
2832                                         $toclevel++;
2833                                         $sublevelCount[$toclevel] = 0;
2834                                         $toc .= $sk->tocIndent();
2835                                 }
2836                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
2837                                         # Decrease TOC level, find level to jump to
2838
2839                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
2840                                                 # Can only go down to level 1
2841                                                 $toclevel = 1;
2842                                         } else {
2843                                                 for ($i = $toclevel; $i > 0; $i--) {
2844                                                         if ( $levelCount[$i] == $level ) {
2845                                                                 # Found last matching level
2846                                                                 $toclevel = $i;
2847                                                                 break;
2848                                                         }
2849                                                         elseif ( $levelCount[$i] < $level ) {
2850                                                                 # Found first matching level below current level
2851                                                                 $toclevel = $i + 1;
2852                                                                 break;
2853                                                         }
2854                                                 }
2855                                         }
2856
2857                                         $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
2858                                 }
2859                                 else {
2860                                         # No change in level, end TOC line
2861                                         $toc .= $sk->tocLineEnd();
2862                                 }
2863
2864                                 $levelCount[$toclevel] = $level;
2865
2866                                 # count number of headlines for each level
2867                                 @$sublevelCount[$toclevel]++;
2868                                 $dot = 0;
2869                                 for( $i = 1; $i <= $toclevel; $i++ ) {
2870                                         if( !empty( $sublevelCount[$i] ) ) {
2871                                                 if( $dot ) {
2872                                                         $numbering .= '.';
2873                                                 }
2874                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
2875                                                 $dot = 1;
2876                                         }
2877                                 }
2878                         }
2879
2880                         # The canonized header is a version of the header text safe to use for links
2881                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
2882                         $canonized_headline = $this->unstrip( $headline, $this->mStripState );
2883                         $canonized_headline = $this->unstripNoWiki( $canonized_headline, $this->mStripState );
2884
2885                         # Remove link placeholders by the link text.
2886                         #     <!--LINK number-->
2887                         # turns into
2888                         #     link text with suffix
2889                         $canonized_headline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
2890                                                             "\$this->mLinkHolders['texts'][\$1]",
2891                                                             $canonized_headline );
2892                         $canonized_headline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
2893                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
2894                                                             $canonized_headline );
2895
2896                         # strip out HTML
2897                         $canonized_headline = preg_replace( '/<.*?' . '>/','',$canonized_headline );
2898                         $tocline = trim( $canonized_headline );
2899                         $canonized_headline = urlencode( Sanitizer::decodeCharReferences( str_replace(' ', '_', $tocline) ) );
2900                         $replacearray = array(
2901                                 '%3A' => ':',
2902                                 '%' => '.'
2903                         );
2904                         $canonized_headline = str_replace(array_keys($replacearray),array_values($replacearray),$canonized_headline);
2905                         $refers[$headlineCount] = $canonized_headline;
2906
2907                         # count how many in assoc. array so we can track dupes in anchors
2908                         @$refers[$canonized_headline]++;
2909                         $refcount[$headlineCount]=$refers[$canonized_headline];
2910
2911                         # Don't number the heading if it is the only one (looks silly)
2912                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
2913                                 # the two are different if the line contains a link
2914                                 $headline=$numbering . ' ' . $headline;
2915                         }
2916
2917                         # Create the anchor for linking from the TOC to the section
2918                         $anchor = $canonized_headline;
2919                         if($refcount[$headlineCount] > 1 ) {
2920                                 $anchor .= '_' . $refcount[$headlineCount];
2921                         }
2922                         if( $doShowToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
2923                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
2924                         }
2925                         if( $showEditLink && ( !$istemplate || $templatetitle !== "" ) ) {
2926                                 if ( empty( $head[$headlineCount] ) ) {
2927                                         $head[$headlineCount] = '';
2928                                 }
2929                                 if( $istemplate )
2930                                         $head[$headlineCount] .= $sk->editSectionLinkForOther($templatetitle, $templatesection);
2931                                 else
2932                                         $head[$headlineCount] .= $sk->editSectionLink($this->mTitle, $sectionCount+1);
2933                         }
2934
2935                         # give headline the correct <h#> tag
2936                         @$head[$headlineCount] .= "<a name=\"$anchor\"></a><h".$level.$matches[2][$headlineCount] .$headline.'</h'.$level.'>';
2937
2938                         $headlineCount++;
2939                         if( !$istemplate )
2940                                 $sectionCount++;
2941                 }
2942
2943                 if( $doShowToc ) {
2944                         $toc .= $sk->tocUnindent( $toclevel - 1 );
2945                         $toc = $sk->tocList( $toc );
2946                 }
2947
2948                 # split up and insert constructed headlines
2949
2950                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
2951                 $i = 0;
2952
2953                 foreach( $blocks as $block ) {
2954                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
2955                                 # This is the [edit] link that appears for the top block of text when
2956                                 # section editing is enabled
2957
2958                                 # Disabled because it broke block formatting
2959                                 # For example, a bullet point in the top line
2960                                 # $full .= $sk->editSectionLink(0);
2961                         }
2962                         $full .= $block;
2963                         if( $doShowToc && !$i && $isMain && !$forceTocHere) {
2964                         # Top anchor now in skin
2965                                 $full = $full.$toc;
2966                         }
2967
2968                         if( !empty( $head[$i] ) ) {
2969                                 $full .= $head[$i];
2970                         }
2971                         $i++;
2972                 }
2973                 if($forceTocHere) {
2974                         $mw =& MagicWord::get( MAG_TOC );
2975                         return $mw->replace( $toc, $full );
2976                 } else {
2977                         return $full;
2978                 }
2979         }
2980
2981         /**
2982          * Return an HTML link for the "ISBN 123456" text
2983          * @access private
2984          */
2985         function magicISBN( $text ) {
2986                 $fname = 'Parser::magicISBN';
2987                 wfProfileIn( $fname );
2988
2989                 $a = split( 'ISBN ', ' '.$text );
2990                 if ( count ( $a ) < 2 ) {
2991                         wfProfileOut( $fname );
2992                         return $text;
2993                 }
2994                 $text = substr( array_shift( $a ), 1);
2995                 $valid = '0123456789-Xx';
2996
2997                 foreach ( $a as $x ) {
2998                         $isbn = $blank = '' ;
2999                         while ( ' ' == $x{0} ) {
3000                                 $blank .= ' ';
3001                                 $x = substr( $x, 1 );
3002                         }
3003                         if ( $x == '' ) { # blank isbn
3004                                 $text .= "ISBN $blank";
3005                                 continue;
3006                         }
3007                         while ( strstr( $valid, $x{0} ) != false ) {
3008                                 $isbn .= $x{0};
3009                                 $x = substr( $x, 1 );
3010                         }
3011                         $num = str_replace( '-', '', $isbn );
3012                         $num = str_replace( ' ', '', $num );
3013                         $num = str_replace( 'x', 'X', $num );
3014
3015                         if ( '' == $num ) {
3016                                 $text .= "ISBN $blank$x";
3017                         } else {
3018                                 $titleObj = Title::makeTitle( NS_SPECIAL, 'Booksources' );
3019                                 $text .= '<a href="' .
3020                                 $titleObj->escapeLocalUrl( 'isbn='.$num ) .
3021                                         "\" class=\"internal\">ISBN $isbn</a>";
3022                                 $text .= $x;
3023                         }
3024                 }
3025                 wfProfileOut( $fname );
3026                 return $text;
3027         }
3028
3029         /**
3030          * Return an HTML link for the "RFC 1234" text
3031          *
3032          * @access private
3033          * @param string $text     Text to be processed
3034          * @param string $keyword  Magic keyword to use (default RFC)
3035          * @param string $urlmsg   Interface message to use (default rfcurl)
3036          * @return string
3037          */
3038         function magicRFC( $text, $keyword='RFC ', $urlmsg='rfcurl'  ) {
3039
3040                 $valid = '0123456789';
3041                 $internal = false;
3042
3043                 $a = split( $keyword, ' '.$text );
3044                 if ( count ( $a ) < 2 ) {
3045                         return $text;
3046                 }
3047                 $text = substr( array_shift( $a ), 1);
3048
3049                 /* Check if keyword is preceed by [[.
3050                  * This test is made here cause of the array_shift above
3051                  * that prevent the test to be done in the foreach.
3052                  */
3053                 if ( substr( $text, -2 ) == '[[' ) {
3054                         $internal = true;
3055                 }
3056
3057                 foreach ( $a as $x ) {
3058                         /* token might be empty if we have RFC RFC 1234 */
3059                         if ( $x=='' ) {
3060                                 $text.=$keyword;
3061                                 continue;
3062                                 }
3063
3064                         $id = $blank = '' ;
3065
3066                         /** remove and save whitespaces in $blank */
3067                         while ( $x{0} == ' ' ) {
3068                                 $blank .= ' ';
3069                                 $x = substr( $x, 1 );
3070                         }
3071
3072                         /** remove and save the rfc number in $id */
3073                         while ( strstr( $valid, $x{0} ) != false ) {
3074                                 $id .= $x{0};
3075                                 $x = substr( $x, 1 );
3076                         }
3077
3078                         if ( $id == '' ) {
3079                                 /* call back stripped spaces*/
3080                                 $text .= $keyword.$blank.$x;
3081                         } elseif( $internal ) {
3082                                 /* normal link */
3083                                 $text .= $keyword.$id.$x;
3084                         } else {
3085                                 /* build the external link*/
3086                                 $url = wfMsg( $urlmsg, $id);
3087                                 $sk =& $this->mOptions->getSkin();
3088                                 $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
3089                                 $text .= "<a href='{$url}'{$la}>{$keyword}{$id}</a>{$x}";
3090                         }
3091
3092                         /* Check if the next RFC keyword is preceed by [[ */
3093                         $internal = ( substr($x,-2) == '[[' );
3094                 }
3095                 return $text;
3096         }
3097
3098         /**
3099          * Transform wiki markup when saving a page by doing \r\n -> \n
3100          * conversion, substitting signatures, {{subst:}} templates, etc.
3101          *
3102          * @param string $text the text to transform
3103          * @param Title &$title the Title object for the current article
3104          * @param User &$user the User object describing the current user
3105          * @param ParserOptions $options parsing options
3106          * @param bool $clearState whether to clear the parser state first
3107          * @return string the altered wiki markup
3108          * @access public
3109          */
3110         function preSaveTransform( $text, &$title, &$user, $options, $clearState = true ) {
3111                 $this->mOptions = $options;
3112                 $this->mTitle =& $title;
3113                 $this->mOutputType = OT_WIKI;
3114
3115                 if ( $clearState ) {
3116                         $this->clearState();
3117                 }
3118
3119                 $stripState = false;
3120                 $pairs = array(
3121                         "\r\n" => "\n",
3122                 );
3123                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3124                 $text = $this->strip( $text, $stripState, true );
3125                 $text = $this->pstPass2( $text, $user );
3126                 $text = $this->unstrip( $text, $stripState );
3127                 $text = $this->unstripNoWiki( $text, $stripState );
3128                 return $text;
3129         }
3130
3131         /**
3132          * Pre-save transform helper function
3133          * @access private
3134          */
3135         function pstPass2( $text, &$user ) {
3136                 global $wgContLang, $wgLocaltimezone;
3137
3138                 # Variable replacement
3139                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3140                 $text = $this->replaceVariables( $text );
3141
3142                 # Signatures
3143                 #
3144                 $sigText = $this->getUserSig( $user );
3145
3146                 /* Note: This is the timestamp saved as hardcoded wikitext to
3147                  * the database, we use $wgContLang here in order to give
3148                  * everyone the same signiture and use the default one rather
3149                  * than the one selected in each users preferences.
3150                  */
3151                 if ( isset( $wgLocaltimezone ) ) {
3152                         $oldtz = getenv( 'TZ' );
3153                         putenv( 'TZ='.$wgLocaltimezone );
3154                 }
3155                 $d = $wgContLang->timeanddate( date( 'YmdHis' ), false, false) .
3156                   ' (' . date( 'T' ) . ')';
3157                 if ( isset( $wgLocaltimezone ) ) {
3158                         putenv( 'TZ='.$oldtz );
3159                 }
3160
3161                 $text = preg_replace( '/~~~~~/', $d, $text );
3162                 $text = preg_replace( '/~~~~/', "$sigText $d", $text );
3163                 $text = preg_replace( '/~~~/', $sigText, $text );
3164
3165                 # Context links: [[|name]] and [[name (context)|]]
3166                 #
3167                 global $wgLegalTitleChars;
3168                 $tc = "[$wgLegalTitleChars]";
3169                 $np = str_replace( array( '(', ')' ), array( '', '' ), $tc ); # No parens
3170
3171                 $namespacechar = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
3172                 $conpat = "/^({$np}+) \\(({$tc}+)\\)$/";
3173
3174                 $p1 = "/\[\[({$np}+) \\(({$np}+)\\)\\|]]/";             # [[page (context)|]]
3175                 $p2 = "/\[\[\\|({$tc}+)]]/";                                    # [[|page]]
3176                 $p3 = "/\[\[(:*$namespacechar+):({$np}+)\\|]]/";                # [[namespace:page|]] and [[:namespace:page|]]
3177                 $p4 = "/\[\[(:*$namespacechar+):({$np}+) \\(({$np}+)\\)\\|]]/"; # [[ns:page (cont)|]] and [[:ns:page (cont)|]]
3178                 $context = '';
3179                 $t = $this->mTitle->getText();
3180                 if ( preg_match( $conpat, $t, $m ) ) {
3181                         $context = $m[2];
3182                 }
3183                 $text = preg_replace( $p4, '[[\\1:\\2 (\\3)|\\2]]', $text );
3184                 $text = preg_replace( $p1, '[[\\1 (\\2)|\\1]]', $text );
3185                 $text = preg_replace( $p3, '[[\\1:\\2|\\2]]', $text );
3186
3187                 if ( '' == $context ) {
3188                         $text = preg_replace( $p2, '[[\\1]]', $text );
3189                 } else {
3190                         $text = preg_replace( $p2, "[[\\1 ({$context})|\\1]]", $text );
3191                 }
3192
3193                 # Trim trailing whitespace
3194                 # MAG_END (__END__) tag allows for trailing
3195                 # whitespace to be deliberately included
3196                 $text = rtrim( $text );
3197                 $mw =& MagicWord::get( MAG_END );
3198                 $mw->matchAndRemove( $text );
3199
3200                 return $text;
3201         }
3202
3203         /**
3204          * Fetch the user's signature text, if any, and normalize to
3205          * validated, ready-to-insert wikitext.
3206          *
3207          * @param User $user
3208          * @return string
3209          * @access private
3210          */
3211         function getUserSig( &$user ) {
3212                 $name = $user->getName();
3213                 $nick = trim( $user->getOption( 'nickname' ) );
3214                 if ( '' == $nick ) {
3215                         $nick = $name;
3216                 }
3217
3218                 if( $user->getOption( 'fancysig' ) ) {
3219                         // A wikitext signature.
3220                         $valid = $this->validateSig( $nick );
3221                         if( $valid === false ) {
3222                                 // Fall back to default sig
3223                                 $nick = $name;
3224                                 wfDebug( "Parser::getUserSig: $name has bad XML tags in signature.\n" );
3225                         } else {
3226                                 return $nick;
3227                         }
3228                 }
3229
3230                 // Plain text linking to the user's homepage
3231                 global $wgContLang;
3232                 $page = $user->getUserPage();
3233                 return '[[' .
3234                         $page->getPrefixedText() .
3235                         "|" .
3236                         wfEscapeWikIText( $nick ) .
3237                         "]]";
3238         }
3239
3240         /**
3241          * We want to enforce two rules on wikitext sigs here:
3242          * 1) Expand any templates at save time (forced subst:)
3243          * 2) Check for unbalanced XML tags, and reject if so.
3244          *
3245          * @param string $text
3246          * @return mixed An expanded string, or false if invalid.
3247          *
3248          * @todo Run brace substitutions
3249          * @todo ?? Check for unbalanced '' and ''' quotes, etc
3250          */
3251         function validateSig( $text ) {
3252                 if( wfIsWellFormedXmlFragment( $text ) ) {
3253                         return $text;
3254                 } else {
3255                         return false;
3256                 }
3257         }
3258
3259         /**
3260          * Set up some variables which are usually set up in parse()
3261          * so that an external function can call some class members with confidence
3262          * @access public
3263          */
3264         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
3265                 $this->mTitle =& $title;
3266                 $this->mOptions = $options;
3267                 $this->mOutputType = $outputType;
3268                 if ( $clearState ) {
3269                         $this->clearState();
3270                 }
3271         }
3272
3273         /**
3274          * Transform a MediaWiki message by replacing magic variables.
3275          *
3276          * @param string $text the text to transform
3277          * @param ParserOptions $options  options
3278          * @return string the text with variables substituted
3279          * @access public
3280          */
3281         function transformMsg( $text, $options ) {
3282                 global $wgTitle;
3283                 static $executing = false;
3284
3285                 # Guard against infinite recursion
3286                 if ( $executing ) {
3287                         return $text;
3288                 }
3289                 $executing = true;
3290
3291                 $this->mTitle = $wgTitle;
3292                 $this->mOptions = $options;
3293                 $this->mOutputType = OT_MSG;
3294                 $this->clearState();
3295                 $text = $this->replaceVariables( $text );
3296
3297                 $executing = false;
3298                 return $text;
3299         }
3300
3301         /**
3302          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
3303          * Callback will be called with the text within
3304          * Transform and return the text within
3305          *
3306          * @access public
3307          *
3308          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
3309          * @param mixed $callback The callback function (and object) to use for the tag
3310          *
3311          * @return The old value of the mTagHooks array associated with the hook
3312          */
3313         function setHook( $tag, $callback ) {
3314                 $oldVal = @$this->mTagHooks[$tag];
3315                 $this->mTagHooks[$tag] = $callback;
3316
3317                 return $oldVal;
3318         }
3319
3320         /**
3321          * Replace <!--LINK--> link placeholders with actual links, in the buffer
3322          * Placeholders created in Skin::makeLinkObj()
3323          * Returns an array of links found, indexed by PDBK:
3324          *  0 - broken
3325          *  1 - normal link
3326          *  2 - stub
3327          * $options is a bit field, RLH_FOR_UPDATE to select for update
3328          */
3329         function replaceLinkHolders( &$text, $options = 0 ) {
3330                 global $wgUser, $wgLinkCache;
3331                 global $wgOutputReplace;
3332
3333                 $fname = 'Parser::replaceLinkHolders';
3334                 wfProfileIn( $fname );
3335
3336                 $pdbks = array();
3337                 $colours = array();
3338                 $sk = $this->mOptions->getSkin();
3339
3340                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
3341                         wfProfileIn( $fname.'-check' );
3342                         $dbr =& wfGetDB( DB_SLAVE );
3343                         $page = $dbr->tableName( 'page' );
3344                         $threshold = $wgUser->getOption('stubthreshold');
3345
3346                         # Sort by namespace
3347                         asort( $this->mLinkHolders['namespaces'] );
3348
3349                         # Generate query
3350                         $query = false;
3351                         foreach ( $this->mLinkHolders['namespaces'] as $key => $val ) {
3352                                 # Make title object
3353                                 $title = $this->mLinkHolders['titles'][$key];
3354
3355                                 # Skip invalid entries.
3356                                 # Result will be ugly, but prevents crash.
3357                                 if ( is_null( $title ) ) {
3358                                         continue;
3359                                 }
3360                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
3361
3362                                 # Check if it's in the link cache already
3363                                 if ( $title->isAlwaysKnown() || $wgLinkCache->getGoodLinkID( $pdbk ) ) {
3364                                         $colours[$pdbk] = 1;
3365                                 } elseif ( $wgLinkCache->isBadLink( $pdbk ) ) {
3366                                         $colours[$pdbk] = 0;
3367                                 } else {
3368                                         # Not in the link cache, add it to the query
3369                                         if ( !isset( $current ) ) {
3370                                                 $current = $val;
3371                                                 $query =  "SELECT page_id, page_namespace, page_title";
3372                                                 if ( $threshold > 0 ) {
3373                                                         $query .= ', page_len, page_is_redirect';
3374                                                 }
3375                                                 $query .= " FROM $page WHERE (page_namespace=$val AND page_title IN(";
3376                                         } elseif ( $current != $val ) {
3377                                                 $current = $val;
3378                                                 $query .= ")) OR (page_namespace=$val AND page_title IN(";
3379                                         } else {
3380                                                 $query .= ', ';
3381                                         }
3382
3383                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
3384                                 }
3385                         }
3386                         if ( $query ) {
3387                                 $query .= '))';
3388                                 if ( $options & RLH_FOR_UPDATE ) {
3389                                         $query .= ' FOR UPDATE';
3390                                 }
3391
3392                                 $res = $dbr->query( $query, $fname );
3393
3394                                 # Fetch data and form into an associative array
3395                                 # non-existent = broken
3396                                 # 1 = known
3397                                 # 2 = stub
3398                                 while ( $s = $dbr->fetchObject($res) ) {
3399                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
3400                                         $pdbk = $title->getPrefixedDBkey();
3401                                         $wgLinkCache->addGoodLinkObj( $s->page_id, $title );
3402
3403                                         if ( $threshold >  0 ) {
3404                                                 $size = $s->page_len;
3405                                                 if ( $s->page_is_redirect || $s->page_namespace != 0 || $size >= $threshold ) {
3406                                                         $colours[$pdbk] = 1;
3407                                                 } else {
3408                                                         $colours[$pdbk] = 2;
3409                                                 }
3410                                         } else {
3411                                                 $colours[$pdbk] = 1;
3412                                         }
3413                                 }
3414                         }
3415                         wfProfileOut( $fname.'-check' );
3416
3417                         # Construct search and replace arrays
3418                         wfProfileIn( $fname.'-construct' );
3419                         $wgOutputReplace = array();
3420                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3421                                 $pdbk = $pdbks[$key];
3422                                 $searchkey = "<!--LINK $key-->";
3423                                 $title = $this->mLinkHolders['titles'][$key];
3424                                 if ( empty( $colours[$pdbk] ) ) {
3425                                         $wgLinkCache->addBadLinkObj( $title );
3426                                         $colours[$pdbk] = 0;
3427                                         $wgOutputReplace[$searchkey] = $sk->makeBrokenLinkObj( $title,
3428                                                                         $this->mLinkHolders['texts'][$key],
3429                                                                         $this->mLinkHolders['queries'][$key] );
3430                                 } elseif ( $colours[$pdbk] == 1 ) {
3431                                         $wgOutputReplace[$searchkey] = $sk->makeKnownLinkObj( $title,
3432                                                                         $this->mLinkHolders['texts'][$key],
3433                                                                         $this->mLinkHolders['queries'][$key] );
3434                                 } elseif ( $colours[$pdbk] == 2 ) {
3435                                         $wgOutputReplace[$searchkey] = $sk->makeStubLinkObj( $title,
3436                                                                         $this->mLinkHolders['texts'][$key],
3437                                                                         $this->mLinkHolders['queries'][$key] );
3438                                 }
3439                         }
3440                         wfProfileOut( $fname.'-construct' );
3441
3442                         # Do the thing
3443                         wfProfileIn( $fname.'-replace' );
3444
3445                         $text = preg_replace_callback(
3446                                 '/(<!--LINK .*?-->)/',
3447                                 "wfOutputReplaceMatches",
3448                                 $text);
3449
3450                         wfProfileOut( $fname.'-replace' );
3451                 }
3452
3453                 # Now process interwiki link holders
3454                 # This is quite a bit simpler than internal links
3455                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
3456                         wfProfileIn( $fname.'-interwiki' );
3457                         # Make interwiki link HTML
3458                         $wgOutputReplace = array();
3459                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
3460                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
3461                                 $wgOutputReplace[$key] = $sk->makeLinkObj( $title, $link );
3462                         }
3463
3464                         $text = preg_replace_callback(
3465                                 '/<!--IWLINK (.*?)-->/',
3466                                 "wfOutputReplaceMatches",
3467                                 $text );
3468                         wfProfileOut( $fname.'-interwiki' );
3469                 }
3470
3471                 wfProfileOut( $fname );
3472                 return $colours;
3473         }
3474
3475         /**
3476          * Replace <!--LINK--> link placeholders with plain text of links
3477          * (not HTML-formatted).
3478          * @param string $text
3479          * @return string
3480          */
3481         function replaceLinkHoldersText( $text ) {
3482                 global $wgUser, $wgLinkCache;
3483                 global $wgOutputReplace;
3484
3485                 $fname = 'Parser::replaceLinkHoldersText';
3486                 wfProfileIn( $fname );
3487
3488                 $text = preg_replace_callback(
3489                         '/<!--(LINK|IWLINK) (.*?)-->/',
3490                         array( &$this, 'replaceLinkHoldersTextCallback' ),
3491                         $text );
3492
3493                 wfProfileOut( $fname );
3494                 return $text;
3495         }
3496
3497         /**
3498          * @param array $matches
3499          * @return string
3500          * @access private
3501          */
3502         function replaceLinkHoldersTextCallback( $matches ) {
3503                 $type = $matches[1];
3504                 $key  = $matches[2];
3505                 if( $type == 'LINK' ) {
3506                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
3507                                 return $this->mLinkHolders['texts'][$key];
3508                         }
3509                 } elseif( $type == 'IWLINK' ) {
3510                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
3511                                 return $this->mInterwikiLinkHolders['texts'][$key];
3512                         }
3513                 }
3514                 return $matches[0];
3515         }
3516
3517         /**
3518          * Renders an image gallery from a text with one line per image.
3519          * text labels may be given by using |-style alternative text. E.g.
3520          *   Image:one.jpg|The number "1"
3521          *   Image:tree.jpg|A tree
3522          * given as text will return the HTML of a gallery with two images,
3523          * labeled 'The number "1"' and
3524          * 'A tree'.
3525          *
3526          * @static
3527          */
3528         function renderImageGallery( $text ) {
3529                 # Setup the parser
3530                 global $wgUser, $wgTitle;
3531                 $parserOptions = ParserOptions::newFromUser( $wgUser );
3532                 $localParser = new Parser();
3533
3534                 global $wgLinkCache;
3535                 $ig = new ImageGallery();
3536                 $ig->setShowBytes( false );
3537                 $ig->setShowFilename( false );
3538                 $lines = explode( "\n", $text );
3539
3540                 foreach ( $lines as $line ) {
3541                         # match lines like these:
3542                         # Image:someimage.jpg|This is some image
3543                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
3544                         # Skip empty lines
3545                         if ( count( $matches ) == 0 ) {
3546                                 continue;
3547                         }
3548                         $nt = Title::newFromURL( $matches[1] );
3549                         if( is_null( $nt ) ) {
3550                                 # Bogus title. Ignore these so we don't bomb out later.
3551                                 continue;
3552                         }
3553                         if ( isset( $matches[3] ) ) {
3554                                 $label = $matches[3];
3555                         } else {
3556                                 $label = '';
3557                         }
3558
3559                         $html = $localParser->parse( $label , $wgTitle, $parserOptions );
3560                         $html = $html->mText;
3561
3562                         $ig->add( new Image( $nt ), $html );
3563                         $wgLinkCache->addImageLinkObj( $nt );
3564                 }
3565                 return $ig->toHTML();
3566         }
3567
3568         /**
3569          * Parse image options text and use it to make an image
3570          */
3571         function makeImage( &$nt, $options ) {
3572                 global $wgContLang, $wgUseImageResize, $wgUser;
3573
3574                 $align = '';
3575
3576                 # Check if the options text is of the form "options|alt text"
3577                 # Options are:
3578                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
3579                 #  * left               no resizing, just left align. label is used for alt= only
3580                 #  * right              same, but right aligned
3581                 #  * none               same, but not aligned
3582                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
3583                 #  * center             center the image
3584                 #  * framed             Keep original image size, no magnify-button.
3585
3586                 $part = explode( '|', $options);
3587
3588                 $mwThumb  =& MagicWord::get( MAG_IMG_THUMBNAIL );
3589                 $mwManualThumb =& MagicWord::get( MAG_IMG_MANUALTHUMB );
3590                 $mwLeft   =& MagicWord::get( MAG_IMG_LEFT );
3591                 $mwRight  =& MagicWord::get( MAG_IMG_RIGHT );
3592                 $mwNone   =& MagicWord::get( MAG_IMG_NONE );
3593                 $mwWidth  =& MagicWord::get( MAG_IMG_WIDTH );
3594                 $mwCenter =& MagicWord::get( MAG_IMG_CENTER );
3595                 $mwFramed =& MagicWord::get( MAG_IMG_FRAMED );
3596                 $caption = '';
3597
3598                 $width = $height = $framed = $thumb = false;
3599                 $manual_thumb = '' ;
3600
3601                 foreach( $part as $key => $val ) {
3602                         if ( $wgUseImageResize && ! is_null( $mwThumb->matchVariableStartToEnd($val) ) ) {
3603                                 $thumb=true;
3604                         } elseif ( ! is_null( $match = $mwManualThumb->matchVariableStartToEnd($val) ) ) {
3605                                 # use manually specified thumbnail
3606                                 $thumb=true;
3607                                 $manual_thumb = $match;
3608                         } elseif ( ! is_null( $mwRight->matchVariableStartToEnd($val) ) ) {
3609                                 # remember to set an alignment, don't render immediately
3610                                 $align = 'right';
3611                         } elseif ( ! is_null( $mwLeft->matchVariableStartToEnd($val) ) ) {
3612                                 # remember to set an alignment, don't render immediately
3613                                 $align = 'left';
3614                         } elseif ( ! is_null( $mwCenter->matchVariableStartToEnd($val) ) ) {
3615                                 # remember to set an alignment, don't render immediately
3616                                 $align = 'center';
3617                         } elseif ( ! is_null( $mwNone->matchVariableStartToEnd($val) ) ) {
3618                                 # remember to set an alignment, don't render immediately
3619                                 $align = 'none';
3620                         } elseif ( $wgUseImageResize && ! is_null( $match = $mwWidth->matchVariableStartToEnd($val) ) ) {
3621                                 wfDebug( "MAG_IMG_WIDTH match: $match\n" );
3622                                 # $match is the image width in pixels
3623                                 if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $match, $m ) ) {
3624                                         $width = intval( $m[1] );
3625                                         $height = intval( $m[2] );
3626                                 } else {
3627                                         $width = intval($match);
3628                                 }
3629                         } elseif ( ! is_null( $mwFramed->matchVariableStartToEnd($val) ) ) {
3630                                 $framed=true;
3631                         } else {
3632                                 $caption = $val;
3633                         }
3634                 }
3635                 # Strip bad stuff out of the alt text
3636                 $alt = $this->replaceLinkHoldersText( $caption );
3637                 $alt = Sanitizer::stripAllTags( $alt );
3638
3639                 # Linker does the rest
3640                 $sk =& $this->mOptions->getSkin();
3641                 return $sk->makeImageLinkObj( $nt, $caption, $alt, $align, $width, $height, $framed, $thumb, $manual_thumb );
3642         }
3643
3644         /**
3645          * Set a flag in the output object indicating that the content is dynamic and
3646          * shouldn't be cached.
3647          */
3648         function disableCache() {
3649                 $this->mOutput->mCacheTime = -1;
3650         }
3651
3652         /**
3653          * Callback from the Sanitizer for expanding items found in HTML attribute
3654          * values, so they can be safely tested and escaped.
3655          * @param string $text
3656          * @param array $args
3657          * @return string
3658          * @access private
3659          */
3660         function attributeStripCallback( &$text, $args ) {
3661                 $text = $this->replaceVariables( $text, $args );
3662                 $text = $this->unstripForHTML( $text );
3663                 return $text;
3664         }
3665
3666         function unstripForHTML( $text ) {
3667                 $text = $this->unstrip( $text, $this->mStripState );
3668                 $text = $this->unstripNoWiki( $text, $this->mStripState );
3669                 return $text;
3670         }
3671
3672         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
3673         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
3674         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
3675 }
3676
3677 /**
3678  * @todo document
3679  * @package MediaWiki
3680  */
3681 class ParserOutput
3682 {
3683         var $mText, $mLanguageLinks, $mCategoryLinks, $mContainsOldMagic;
3684         var $mCacheTime; # Timestamp on this article, or -1 for uncacheable. Used in ParserCache.
3685         var $mVersion;   # Compatibility check
3686         var $mTitleText; # title text of the chosen language variant
3687
3688         function ParserOutput( $text = '', $languageLinks = array(), $categoryLinks = array(),
3689                 $containsOldMagic = false, $titletext = '' )
3690         {
3691                 $this->mText = $text;
3692                 $this->mLanguageLinks = $languageLinks;
3693                 $this->mCategoryLinks = $categoryLinks;
3694                 $this->mContainsOldMagic = $containsOldMagic;
3695                 $this->mCacheTime = '';
3696                 $this->mVersion = MW_PARSER_VERSION;
3697                 $this->mTitleText = $titletext;
3698         }
3699
3700         function getText()                   { return $this->mText; }
3701         function getLanguageLinks()          { return $this->mLanguageLinks; }
3702         function getCategoryLinks()          { return array_keys( $this->mCategoryLinks ); }
3703         function getCacheTime()              { return $this->mCacheTime; }
3704         function getTitleText()              { return $this->mTitleText; }
3705         function containsOldMagic()          { return $this->mContainsOldMagic; }
3706         function setText( $text )            { return wfSetVar( $this->mText, $text ); }
3707         function setLanguageLinks( $ll )     { return wfSetVar( $this->mLanguageLinks, $ll ); }
3708         function setCategoryLinks( $cl )     { return wfSetVar( $this->mCategoryLinks, $cl ); }
3709         function setContainsOldMagic( $com ) { return wfSetVar( $this->mContainsOldMagic, $com ); }
3710         function setCacheTime( $t )          { return wfSetVar( $this->mCacheTime, $t ); }
3711         function setTitleText( $t )          { return wfSetVar ($this->mTitleText, $t); }
3712
3713         function addCategoryLink( $c )       { $this->mCategoryLinks[$c] = 1; }
3714
3715         function merge( $other ) {
3716                 $this->mLanguageLinks = array_merge( $this->mLanguageLinks, $other->mLanguageLinks );
3717                 $this->mCategoryLinks = array_merge( $this->mCategoryLinks, $this->mLanguageLinks );
3718                 $this->mContainsOldMagic = $this->mContainsOldMagic || $other->mContainsOldMagic;
3719         }
3720
3721         /**
3722          * Return true if this cached output object predates the global or
3723          * per-article cache invalidation timestamps, or if it comes from
3724          * an incompatible older version.
3725          *
3726          * @param string $touched the affected article's last touched timestamp
3727          * @return bool
3728          * @access public
3729          */
3730         function expired( $touched ) {
3731                 global $wgCacheEpoch;
3732                 return $this->getCacheTime() == -1 || // parser says it's uncacheable
3733                        $this->getCacheTime() < $touched ||
3734                        $this->getCacheTime() <= $wgCacheEpoch ||
3735                        !isset( $this->mVersion ) ||
3736                        version_compare( $this->mVersion, MW_PARSER_VERSION, "lt" );
3737         }
3738 }
3739
3740 /**
3741  * Set options of the Parser
3742  * @todo document
3743  * @package MediaWiki
3744  */
3745 class ParserOptions
3746 {
3747         # All variables are private
3748         var $mUseTeX;                    # Use texvc to expand <math> tags
3749         var $mUseDynamicDates;           # Use DateFormatter to format dates
3750         var $mInterwikiMagic;            # Interlanguage links are removed and returned in an array
3751         var $mAllowExternalImages;       # Allow external images inline
3752         var $mAllowExternalImagesFrom;   # If not, any exception?
3753         var $mSkin;                      # Reference to the preferred skin
3754         var $mDateFormat;                # Date format index
3755         var $mEditSection;               # Create "edit section" links
3756         var $mNumberHeadings;            # Automatically number headings
3757         var $mAllowSpecialInclusion;     # Allow inclusion of special pages
3758
3759         function getUseTeX()                        { return $this->mUseTeX; }
3760         function getUseDynamicDates()               { return $this->mUseDynamicDates; }
3761         function getInterwikiMagic()                { return $this->mInterwikiMagic; }
3762         function getAllowExternalImages()           { return $this->mAllowExternalImages; }
3763         function getAllowExternalImagesFrom()       { return $this->mAllowExternalImagesFrom; }
3764         function &getSkin()                         { return $this->mSkin; }
3765         function getDateFormat()                    { return $this->mDateFormat; }
3766         function getEditSection()                   { return $this->mEditSection; }
3767         function getNumberHeadings()                { return $this->mNumberHeadings; }
3768         function getAllowSpecialInclusion()         { return $this->mAllowSpecialInclusion; }
3769
3770
3771         function setUseTeX( $x )                    { return wfSetVar( $this->mUseTeX, $x ); }
3772         function setUseDynamicDates( $x )           { return wfSetVar( $this->mUseDynamicDates, $x ); }
3773         function setInterwikiMagic( $x )            { return wfSetVar( $this->mInterwikiMagic, $x ); }
3774         function setAllowExternalImages( $x )       { return wfSetVar( $this->mAllowExternalImages, $x ); }
3775         function setAllowExternalImagesFrom( $x )   { return wfSetVar( $this->mAllowExternalImagesFrom, $x ); }
3776         function setDateFormat( $x )                { return wfSetVar( $this->mDateFormat, $x ); }
3777         function setEditSection( $x )               { return wfSetVar( $this->mEditSection, $x ); }
3778         function setNumberHeadings( $x )            { return wfSetVar( $this->mNumberHeadings, $x ); }
3779         function setAllowSpecialInclusion( $x )     { return wfSetVar( $this->mAllowSpecialInclusion, $x ); }
3780
3781         function setSkin( &$x ) { $this->mSkin =& $x; }
3782
3783         function ParserOptions() {
3784                 global $wgUser;
3785                 $this->initialiseFromUser( $wgUser );
3786         }
3787
3788         /**
3789          * Get parser options
3790          * @static
3791          */
3792         function newFromUser( &$user ) {
3793                 $popts = new ParserOptions;
3794                 $popts->initialiseFromUser( $user );
3795                 return $popts;
3796         }
3797
3798         /** Get user options */
3799         function initialiseFromUser( &$userInput ) {
3800                 global $wgUseTeX, $wgUseDynamicDates, $wgInterwikiMagic, $wgAllowExternalImages,
3801                        $wgAllowExternalImagesFrom, $wgAllowSpecialInclusion;
3802                 $fname = 'ParserOptions::initialiseFromUser';
3803                 wfProfileIn( $fname );
3804                 if ( !$userInput ) {
3805                         $user = new User;
3806                         $user->setLoaded( true );
3807                 } else {
3808                         $user =& $userInput;
3809                 }
3810
3811                 $this->mUseTeX = $wgUseTeX;
3812                 $this->mUseDynamicDates = $wgUseDynamicDates;
3813                 $this->mInterwikiMagic = $wgInterwikiMagic;
3814                 $this->mAllowExternalImages = $wgAllowExternalImages;
3815                 $this->mAllowExternalImagesFrom = $wgAllowExternalImagesFrom;
3816                 wfProfileIn( $fname.'-skin' );
3817                 $this->mSkin =& $user->getSkin();
3818                 wfProfileOut( $fname.'-skin' );
3819                 $this->mDateFormat = $user->getOption( 'date' );
3820                 $this->mEditSection = true;
3821                 $this->mNumberHeadings = $user->getOption( 'numberheadings' );
3822                 $this->mAllowSpecialInclusion = $wgAllowSpecialInclusion;
3823                 wfProfileOut( $fname );
3824         }
3825 }
3826
3827 /**
3828  * Callback function used by Parser::replaceLinkHolders()
3829  * to substitute link placeholders.
3830  */
3831 function &wfOutputReplaceMatches( $matches ) {
3832         global $wgOutputReplace;
3833         return $wgOutputReplace[$matches[1]];
3834 }
3835
3836 /**
3837  * Return the total number of articles
3838  */
3839 function wfNumberOfArticles() {
3840         global $wgNumberOfArticles;
3841
3842         wfLoadSiteStats();
3843         return $wgNumberOfArticles;
3844 }
3845
3846 /**
3847  * Return the number of files
3848  */
3849 function wfNumberOfFiles() {
3850         $fname = 'Parser::wfNumberOfFiles';
3851
3852         wfProfileIn( $fname );
3853         $dbr =& wfGetDB( DB_SLAVE );
3854         $res = $dbr->selectField('image', 'COUNT(*)', array(), $fname );
3855         wfProfileOut( $fname );
3856
3857         return $res;
3858 }
3859
3860 /**
3861  * Get various statistics from the database
3862  * @private
3863  */
3864 function wfLoadSiteStats() {
3865         global $wgNumberOfArticles, $wgTotalViews, $wgTotalEdits;
3866         $fname = 'wfLoadSiteStats';
3867
3868         if ( -1 != $wgNumberOfArticles ) return;
3869         $dbr =& wfGetDB( DB_SLAVE );
3870         $s = $dbr->selectRow( 'site_stats',
3871                 array( 'ss_total_views', 'ss_total_edits', 'ss_good_articles' ),
3872                 array( 'ss_row_id' => 1 ), $fname
3873         );
3874
3875         if ( $s === false ) {
3876                 return;
3877         } else {
3878                 $wgTotalViews = $s->ss_total_views;
3879                 $wgTotalEdits = $s->ss_total_edits;
3880                 $wgNumberOfArticles = $s->ss_good_articles;
3881         }
3882 }
3883
3884 /**
3885  * Escape html tags
3886  * Basicly replacing " > and < with HTML entities ( &quot;, &gt;, &lt;)
3887  *
3888  * @param string $in Text that might contain HTML tags
3889  * @return string Escaped string
3890  */
3891 function wfEscapeHTMLTagsOnly( $in ) {
3892         return str_replace(
3893                 array( '"', '>', '<' ),
3894                 array( '&quot;', '&gt;', '&lt;' ),
3895                 $in );
3896 }
3897
3898 ?>