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', '1.6.0' );
  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
  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         var $mTemplates,        // cache of already loaded templates, avoids
 103                                 // multiple SQL queries for the same string
 104             $mTemplatePath;     // stores an unsorted hash of all the templates already loaded
 105                                 // in this path. Used for loop detection.
 106
 107         # Temporary
 108         # These are variables reset at least once per parse regardless of $clearState
 109         var $mOptions,      // ParserOptions object
 110                 $mTitle,        // Title context, used for self-link rendering and similar things
 111                 $mOutputType,   // Output type, one of the OT_xxx constants
 112                 $mRevisionId;   // ID to display in {{REVISIONID}} tags
 113
 114         /**#@-*/
 115
 116         /**
 117          * Constructor
 118          *
 119          * @access public
 120          */
 121         function Parser() {
 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                 # Clear these on every parse, bug 4549
 156                 $this->mTemplates = array();
 157                 $this->mTemplatePath = array();
 158
 159                 wfRunHooks( 'ParserClearState', array( &$this ) );
 160         }
 161
 162         /**
 163          * Accessor for mUniqPrefix.
 164          *
 165          * @access public
 166          */
 167         function UniqPrefix() {
 168                 return $this->mUniqPrefix;
 169         }
 170
 171         /**
 172          * Convert wikitext to HTML
 173          * Do not call this function recursively.
 174          *
 175          * @access private
 176          * @param string $text Text we want to parse
 177          * @param Title &$title A title object
 178          * @param array $options
 179          * @param boolean $linestart
 180          * @param boolean $clearState
 181          * @param int $revid number to pass in {{REVISIONID}}
 182          * @return ParserOutput a ParserOutput
 183          */
 184         function parse( $text, &$title, $options, $linestart = true, $clearState = true, $revid = null ) {
 185                 /**
 186                  * First pass--just handle <nowiki> sections, pass the rest off
 187                  * to internalParse() which does all the real work.
 188                  */
 189
 190                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
 191                 $fname = 'Parser::parse';
 192                 wfProfileIn( $fname );
 193
 194                 if ( $clearState ) {
 195                         $this->clearState();
 196                 }
 197
 198                 $this->mOptions = $options;
 199                 $this->mTitle =& $title;
 200                 $this->mRevisionId = $revid;
 201                 $this->mOutputType = OT_HTML;
 202
 203                 $this->mStripState = NULL;
 204
 205                 //$text = $this->strip( $text, $this->mStripState );
 206                 // VOODOO MAGIC FIX! Sometimes the above segfaults in PHP5.
 207                 $x =& $this->mStripState;
 208
 209                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$x ) );
 210                 $text = $this->strip( $text, $x );
 211                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$x ) );
 212
 213                 # Hook to suspend the parser in this state
 214                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$x ) ) ) {
 215                         wfProfileOut( $fname );
 216                         return $text ;
 217                 }
 218
 219                 $text = $this->internalParse( $text );
 220
 221                 $text = $this->unstrip( $text, $this->mStripState );
 222
 223                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 224                 $fixtags = array(
 225                         # french spaces, last one Guillemet-left
 226                         # only if there is something before the space
 227                         '/(.) (?=\\?|:|;|!|\\302\\273)/' => '\\1&nbsp;\\2',
 228                         # french spaces, Guillemet-right
 229                         '/(\\302\\253) /' => '\\1&nbsp;',
 230                         '/<center *>(.*)<\\/center *>/i' => '<div class="center">\\1</div>',
 231                 );
 232                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 233
 234                 # only once and last
 235                 $text = $this->doBlockLevels( $text, $linestart );
 236
 237                 $this->replaceLinkHolders( $text );
 238
 239                 # the position of the convert() call should not be changed. it
 240                 # assumes that the links are all replaces and the only thing left
 241                 # is the <nowiki> mark.
 242                 $text = $wgContLang->convert($text);
 243                 $this->mOutput->setTitleText($wgContLang->getParsedTitle());
 244
 245                 $text = $this->unstripNoWiki( $text, $this->mStripState );
 246
 247                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 248
 249                 $text = Sanitizer::normalizeCharReferences( $text );
 250
 251                 if (($wgUseTidy and $this->mOptions->mTidy) or $wgAlwaysUseTidy) {
 252                         $text = Parser::tidy($text);
 253                 }
 254
 255                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 256
 257                 $this->mOutput->setText( $text );
 258                 wfProfileOut( $fname );
 259
 260                 return $this->mOutput;
 261         }
 262
 263         /**
 264          * Get a random string
 265          *
 266          * @access private
 267          * @static
 268          */
 269         function getRandomString() {
 270                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 271         }
 272
 273         /**
 274          * Replaces all occurrences of <$tag>content</$tag> in the text
 275          * with a random marker and returns the new text. the output parameter
 276          * $content will be an associative array filled with data on the form
 277          * $unique_marker => content.
 278          *
 279          * If $content is already set, the additional entries will be appended
 280          * If $tag is set to STRIP_COMMENTS, the function will extract
 281          * <!-- HTML comments -->
 282          *
 283          * @access private
 284          * @static
 285          */
 286         function extractTagsAndParams($tag, $text, &$content, &$tags, &$params, $uniq_prefix = ''){
 287                 $rnd = $uniq_prefix . '-' . $tag . Parser::getRandomString();
 288                 if ( !$content ) {
 289                         $content = array( );
 290                 }
 291                 $n = 1;
 292                 $stripped = '';
 293
 294                 if ( !$tags ) {
 295                         $tags = array( );
 296                 }
 297
 298                 if ( !$params ) {
 299                         $params = array( );
 300                 }
 301
 302                 if( $tag == STRIP_COMMENTS ) {
 303                         $start = '/<!--()()/';
 304                         $end   = '/-->/';
 305                 } else {
 306                         $start = "/<$tag(\\s+[^\\/>]*|\\s*)(\\/?)>/i";
 307                         $end   = "/<\\/$tag\\s*>/i";
 308                 }
 309
 310                 while ( '' != $text ) {
 311                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 312                         $stripped .= $p[0];
 313                         if( count( $p ) < 4 ) {
 314                                 break;
 315                         }
 316                         $attributes = $p[1];
 317                         $empty      = $p[2];
 318                         $inside     = $p[3];
 319
 320                         $marker = $rnd . sprintf('%08X', $n++);
 321                         $stripped .= $marker;
 322
 323                         $tags[$marker] = "<$tag$attributes$empty>";
 324                         $params[$marker] = Sanitizer::decodeTagAttributes( $attributes );
 325
 326                         if ( $empty === '/' ) {
 327                                 // Empty element tag, <tag />
 328                                 $content[$marker] = null;
 329                                 $text = $inside;
 330                         } else {
 331                                 $q = preg_split( $end, $inside, 2 );
 332                                 $content[$marker] = $q[0];
 333                                 if( count( $q ) < 2 ) {
 334                                         # No end tag -- let it run out to the end of the text.
 335                                         break;
 336                                 } else {
 337                                         $text = $q[1];
 338                                 }
 339                         }
 340                 }
 341                 return $stripped;
 342         }
 343
 344         /**
 345          * Wrapper function for extractTagsAndParams
 346          * for cases where $tags and $params isn't needed
 347          * i.e. where tags will never have params, like <nowiki>
 348          *
 349          * @access private
 350          * @static
 351          */
 352         function extractTags( $tag, $text, &$content, $uniq_prefix = '' ) {
 353                 $dummy_tags = array();
 354                 $dummy_params = array();
 355
 356                 return Parser::extractTagsAndParams( $tag, $text, $content,
 357                         $dummy_tags, $dummy_params, $uniq_prefix );
 358         }
 359
 360         /**
 361          * Strips and renders nowiki, pre, math, hiero
 362          * If $render is set, performs necessary rendering operations on plugins
 363          * Returns the text, and fills an array with data needed in unstrip()
 364          * If the $state is already a valid strip state, it adds to the state
 365          *
 366          * @param bool $stripcomments when set, HTML comments <!-- like this -->
 367          *  will be stripped in addition to other tags. This is important
 368          *  for section editing, where these comments cause confusion when
 369          *  counting the sections in the wikisource
 370          *
 371          * @access private
 372          */
 373         function strip( $text, &$state, $stripcomments = false ) {
 374                 $render = ($this->mOutputType == OT_HTML);
 375                 $html_content = array();
 376                 $nowiki_content = array();
 377                 $math_content = array();
 378                 $pre_content = array();
 379                 $comment_content = array();
 380                 $ext_content = array();
 381                 $ext_tags = array();
 382                 $ext_params = array();
 383                 $gallery_content = array();
 384
 385                 # Replace any instances of the placeholders
 386                 $uniq_prefix = $this->mUniqPrefix;
 387                 #$text = str_replace( $uniq_prefix, wfHtmlEscapeFirst( $uniq_prefix ), $text );
 388
 389                 # html
 390                 global $wgRawHtml;
 391                 if( $wgRawHtml ) {
 392                         $text = Parser::extractTags('html', $text, $html_content, $uniq_prefix);
 393                         foreach( $html_content as $marker => $content ) {
 394                                 if ($render ) {
 395                                         # Raw and unchecked for validity.
 396                                         $html_content[$marker] = $content;
 397                                 } else {
 398                                         $html_content[$marker] = '<html>'.$content.'</html>';
 399                                 }
 400                         }
 401                 }
 402
 403                 # nowiki
 404                 $text = Parser::extractTags('nowiki', $text, $nowiki_content, $uniq_prefix);
 405                 foreach( $nowiki_content as $marker => $content ) {
 406                         if( $render ){
 407                                 $nowiki_content[$marker] = wfEscapeHTMLTagsOnly( $content );
 408                         } else {
 409                                 $nowiki_content[$marker] = '<nowiki>'.$content.'</nowiki>';
 410                         }
 411                 }
 412
 413                 # math
 414                 if( $this->mOptions->getUseTeX() ) {
 415                         $text = Parser::extractTags('math', $text, $math_content, $uniq_prefix);
 416                         foreach( $math_content as $marker => $content ){
 417                                 if( $render ) {
 418                                         $math_content[$marker] = renderMath( $content );
 419                                 } else {
 420                                         $math_content[$marker] = '<math>'.$content.'</math>';
 421                                 }
 422                         }
 423                 }
 424
 425                 # pre
 426                 $text = Parser::extractTags('pre', $text, $pre_content, $uniq_prefix);
 427                 foreach( $pre_content as $marker => $content ){
 428                         if( $render ){
 429                                 $pre_content[$marker] = '<pre>' . wfEscapeHTMLTagsOnly( $content ) . '</pre>';
 430                         } else {
 431                                 $pre_content[$marker] = '<pre>'.$content.'</pre>';
 432                         }
 433                 }
 434
 435                 # gallery
 436                 $text = Parser::extractTags('gallery', $text, $gallery_content, $uniq_prefix);
 437                 foreach( $gallery_content as $marker => $content ) {
 438                         require_once( 'ImageGallery.php' );
 439                         if ( $render ) {
 440                                 $gallery_content[$marker] = $this->renderImageGallery( $content );
 441                         } else {
 442                                 $gallery_content[$marker] = '<gallery>'.$content.'</gallery>';
 443                         }
 444                 }
 445
 446                 # Comments
 447                 if($stripcomments) {
 448                         $text = Parser::extractTags(STRIP_COMMENTS, $text, $comment_content, $uniq_prefix);
 449                         foreach( $comment_content as $marker => $content ){
 450                                 $comment_content[$marker] = '<!--'.$content.'-->';
 451                         }
 452                 }
 453
 454                 # Extensions
 455                 foreach ( $this->mTagHooks as $tag => $callback ) {
 456                         $ext_content[$tag] = array();
 457                         $text = Parser::extractTagsAndParams( $tag, $text, $ext_content[$tag],
 458                                 $ext_tags[$tag], $ext_params[$tag], $uniq_prefix );
 459                         foreach( $ext_content[$tag] as $marker => $content ) {
 460                                 $full_tag = $ext_tags[$tag][$marker];
 461                                 $params = $ext_params[$tag][$marker];
 462                                 if ( $render )
 463                                         $ext_content[$tag][$marker] = call_user_func_array( $callback, array( $content, $params, &$this ) );
 464                                 else {
 465                                         if ( is_null( $content ) ) {
 466                                                 // Empty element tag
 467                                                 $ext_content[$tag][$marker] = $full_tag;
 468                                         } else {
 469                                                 $ext_content[$tag][$marker] = "$full_tag$content</$tag>";
 470                                         }
 471                                 }
 472                         }
 473                 }
 474
 475                 # Merge state with the pre-existing state, if there is one
 476                 if ( $state ) {
 477                         $state['html'] = $state['html'] + $html_content;
 478                         $state['nowiki'] = $state['nowiki'] + $nowiki_content;
 479                         $state['math'] = $state['math'] + $math_content;
 480                         $state['pre'] = $state['pre'] + $pre_content;
 481                         $state['gallery'] = $state['gallery'] + $gallery_content;
 482                         $state['comment'] = $state['comment'] + $comment_content;
 483
 484                         foreach( $ext_content as $tag => $array ) {
 485                                 if ( array_key_exists( $tag, $state ) ) {
 486                                         $state[$tag] = $state[$tag] + $array;
 487                                 }
 488                         }
 489                 } else {
 490                         $state = array(
 491                           'html' => $html_content,
 492                           'nowiki' => $nowiki_content,
 493                           'math' => $math_content,
 494                           'pre' => $pre_content,
 495                           'gallery' => $gallery_content,
 496                           'comment' => $comment_content,
 497                         ) + $ext_content;
 498                 }
 499                 return $text;
 500         }
 501
 502         /**
 503          * restores pre, math, and hiero removed by strip()
 504          *
 505          * always call unstripNoWiki() after this one
 506          * @access private
 507          */
 508         function unstrip( $text, &$state ) {
 509                 if ( !is_array( $state ) ) {
 510                         return $text;
 511                 }
 512
 513                 # Must expand in reverse order, otherwise nested tags will be corrupted
 514                 foreach( array_reverse( $state, true ) as $tag => $contentDict ) {
 515                         if( $tag != 'nowiki' && $tag != 'html' ) {
 516                                 foreach( array_reverse( $contentDict, true ) as $uniq => $content ) {
 517                                         $text = str_replace( $uniq, $content, $text );
 518                                 }
 519                         }
 520                 }
 521
 522                 return $text;
 523         }
 524
 525         /**
 526          * always call this after unstrip() to preserve the order
 527          *
 528          * @access private
 529          */
 530         function unstripNoWiki( $text, &$state ) {
 531                 if ( !is_array( $state ) ) {
 532                         return $text;
 533                 }
 534
 535                 # Must expand in reverse order, otherwise nested tags will be corrupted
 536                 for ( $content = end($state['nowiki']); $content !== false; $content = prev( $state['nowiki'] ) ) {
 537                         $text = str_replace( key( $state['nowiki'] ), $content, $text );
 538                 }
 539
 540                 global $wgRawHtml;
 541                 if ($wgRawHtml) {
 542                         for ( $content = end($state['html']); $content !== false; $content = prev( $state['html'] ) ) {
 543                                 $text = str_replace( key( $state['html'] ), $content, $text );
 544                         }
 545                 }
 546
 547                 return $text;
 548         }
 549
 550         /**
 551          * Add an item to the strip state
 552          * Returns the unique tag which must be inserted into the stripped text
 553          * The tag will be replaced with the original text in unstrip()
 554          *
 555          * @access private
 556          */
 557         function insertStripItem( $text, &$state ) {
 558                 $rnd = $this->mUniqPrefix . '-item' . Parser::getRandomString();
 559                 if ( !$state ) {
 560                         $state = array(
 561                           'html' => array(),
 562                           'nowiki' => array(),
 563                           'math' => array(),
 564                           'pre' => array(),
 565                           'comment' => array(),
 566                           'gallery' => array(),
 567                         );
 568                 }
 569                 $state['item'][$rnd] = $text;
 570                 return $rnd;
 571         }
 572
 573         /**
 574          * Interface with html tidy, used if $wgUseTidy = true.
 575          * If tidy isn't able to correct the markup, the original will be
 576          * returned in all its glory with a warning comment appended.
 577          *
 578          * Either the external tidy program or the in-process tidy extension
 579          * will be used depending on availability. Override the default
 580          * $wgTidyInternal setting to disable the internal if it's not working.
 581          *
 582          * @param string $text Hideous HTML input
 583          * @return string Corrected HTML output
 584          * @access public
 585          * @static
 586          */
 587         function tidy( $text ) {
 588                 global $wgTidyInternal;
 589                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 590 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 591 '<head><title>test</title></head><body>'.$text.'</body></html>';
 592                 if( $wgTidyInternal ) {
 593                         $correctedtext = Parser::internalTidy( $wrappedtext );
 594                 } else {
 595                         $correctedtext = Parser::externalTidy( $wrappedtext );
 596                 }
 597                 if( is_null( $correctedtext ) ) {
 598                         wfDebug( "Tidy error detected!\n" );
 599                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 600                 }
 601                 return $correctedtext;
 602         }
 603
 604         /**
 605          * Spawn an external HTML tidy process and get corrected markup back from it.
 606          *
 607          * @access private
 608          * @static
 609          */
 610         function externalTidy( $text ) {
 611                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 612                 $fname = 'Parser::externalTidy';
 613                 wfProfileIn( $fname );
 614
 615                 $cleansource = '';
 616                 $opts = ' -utf8';
 617
 618                 $descriptorspec = array(
 619                         0 => array('pipe', 'r'),
 620                         1 => array('pipe', 'w'),
 621                         2 => array('file', '/dev/null', 'a')
 622                 );
 623                 $pipes = array();
 624                 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 625                 if (is_resource($process)) {
 626                         // Theoretically, this style of communication could cause a deadlock
 627                         // here. If the stdout buffer fills up, then writes to stdin could
 628                         // block. This doesn't appear to happen with tidy, because tidy only
 629                         // writes to stdout after it's finished reading from stdin. Search
 630                         // for tidyParseStdin and tidySaveStdout in console/tidy.c
 631                         fwrite($pipes[0], $text);
 632                         fclose($pipes[0]);
 633                         while (!feof($pipes[1])) {
 634                                 $cleansource .= fgets($pipes[1], 1024);
 635                         }
 636                         fclose($pipes[1]);
 637                         proc_close($process);
 638                 }
 639
 640                 wfProfileOut( $fname );
 641
 642                 if( $cleansource == '' && $text != '') {
 643                         // Some kind of error happened, so we couldn't get the corrected text.
 644                         // Just give up; we'll use the source text and append a warning.
 645                         return null;
 646                 } else {
 647                         return $cleansource;
 648                 }
 649         }
 650
 651         /**
 652          * Use the HTML tidy PECL extension to use the tidy library in-process,
 653          * saving the overhead of spawning a new process. Currently written to
 654          * the PHP 4.3.x version of the extension, may not work on PHP 5.
 655          *
 656          * 'pear install tidy' should be able to compile the extension module.
 657          *
 658          * @access private
 659          * @static
 660          */
 661         function internalTidy( $text ) {
 662                 global $wgTidyConf;
 663                 $fname = 'Parser::internalTidy';
 664                 wfProfileIn( $fname );
 665
 666                 tidy_load_config( $wgTidyConf );
 667                 tidy_set_encoding( 'utf8' );
 668                 tidy_parse_string( $text );
 669                 tidy_clean_repair();
 670                 if( tidy_get_status() == 2 ) {
 671                         // 2 is magic number for fatal error
 672                         // http://www.php.net/manual/en/function.tidy-get-status.php
 673                         $cleansource = null;
 674                 } else {
 675                         $cleansource = tidy_get_output();
 676                 }
 677                 wfProfileOut( $fname );
 678                 return $cleansource;
 679         }
 680
 681         /**
 682          * parse the wiki syntax used to render tables
 683          *
 684          * @access private
 685          */
 686         function doTableStuff ( $t ) {
 687                 $fname = 'Parser::doTableStuff';
 688                 wfProfileIn( $fname );
 689
 690                 $t = explode ( "\n" , $t ) ;
 691                 $td = array () ; # Is currently a td tag open?
 692                 $ltd = array () ; # Was it TD or TH?
 693                 $tr = array () ; # Is currently a tr tag open?
 694                 $ltr = array () ; # tr attributes
 695                 $has_opened_tr = array(); # Did this table open a <tr> element?
 696                 $indent_level = 0; # indent level of the table
 697                 foreach ( $t AS $k => $x )
 698                 {
 699                         $x = trim ( $x ) ;
 700                         $fc = substr ( $x , 0 , 1 ) ;
 701                         if ( preg_match( '/^(:*)\{\|(.*)$/', $x, $matches ) ) {
 702                                 $indent_level = strlen( $matches[1] );
 703
 704                                 $attributes = $this->unstripForHTML( $matches[2] );
 705
 706                                 $t[$k] = str_repeat( '<dl><dd>', $indent_level ) .
 707                                         '<table' . Sanitizer::fixTagAttributes ( $attributes, 'table' ) . '>' ;
 708                                 array_push ( $td , false ) ;
 709                                 array_push ( $ltd , '' ) ;
 710                                 array_push ( $tr , false ) ;
 711                                 array_push ( $ltr , '' ) ;
 712                                 array_push ( $has_opened_tr, false );
 713                         }
 714                         else if ( count ( $td ) == 0 ) { } # Don't do any of the following
 715                         else if ( '|}' == substr ( $x , 0 , 2 ) ) {
 716                                 $z = "</table>" . substr ( $x , 2);
 717                                 $l = array_pop ( $ltd ) ;
 718                                 if ( !array_pop ( $has_opened_tr ) ) $z = "<tr><td></td></tr>" . $z ;
 719                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 720                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 721                                 array_pop ( $ltr ) ;
 722                                 $t[$k] = $z . str_repeat( '</dd></dl>', $indent_level );
 723                         }
 724                         else if ( '|-' == substr ( $x , 0 , 2 ) ) { # Allows for |---------------
 725                                 $x = substr ( $x , 1 ) ;
 726                                 while ( $x != '' && substr ( $x , 0 , 1 ) == '-' ) $x = substr ( $x , 1 ) ;
 727                                 $z = '' ;
 728                                 $l = array_pop ( $ltd ) ;
 729                                 array_pop ( $has_opened_tr );
 730                                 array_push ( $has_opened_tr , true ) ;
 731                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 732                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 733                                 array_pop ( $ltr ) ;
 734                                 $t[$k] = $z ;
 735                                 array_push ( $tr , false ) ;
 736                                 array_push ( $td , false ) ;
 737                                 array_push ( $ltd , '' ) ;
 738                                 $attributes = $this->unstripForHTML( $x );
 739                                 array_push ( $ltr , Sanitizer::fixTagAttributes ( $attributes, 'tr' ) ) ;
 740                         }
 741                         else if ( '|' == $fc || '!' == $fc || '|+' == substr ( $x , 0 , 2 ) ) { # Caption
 742                                 # $x is a table row
 743                                 if ( '|+' == substr ( $x , 0 , 2 ) ) {
 744                                         $fc = '+' ;
 745                                         $x = substr ( $x , 1 ) ;
 746                                 }
 747                                 $after = substr ( $x , 1 ) ;
 748                                 if ( $fc == '!' ) $after = str_replace ( '!!' , '||' , $after ) ;
 749                                 $after = explode ( '||' , $after ) ;
 750                                 $t[$k] = '' ;
 751
 752                                 # Loop through each table cell
 753                                 foreach ( $after AS $theline )
 754                                 {
 755                                         $z = '' ;
 756                                         if ( $fc != '+' )
 757                                         {
 758                                                 $tra = array_pop ( $ltr ) ;
 759                                                 if ( !array_pop ( $tr ) ) $z = '<tr'.$tra.">\n" ;
 760                                                 array_push ( $tr , true ) ;
 761                                                 array_push ( $ltr , '' ) ;
 762                                                 array_pop ( $has_opened_tr );
 763                                                 array_push ( $has_opened_tr , true ) ;
 764                                         }
 765
 766                                         $l = array_pop ( $ltd ) ;
 767                                         if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 768                                         if ( $fc == '|' ) $l = 'td' ;
 769                                         else if ( $fc == '!' ) $l = 'th' ;
 770                                         else if ( $fc == '+' ) $l = 'caption' ;
 771                                         else $l = '' ;
 772                                         array_push ( $ltd , $l ) ;
 773
 774                                         # Cell parameters
 775                                         $y = explode ( '|' , $theline , 2 ) ;
 776                                         # Note that a '|' inside an invalid link should not
 777                                         # be mistaken as delimiting cell parameters
 778                                         if ( strpos( $y[0], '[[' ) !== false ) {
 779                                                 $y = array ($theline);
 780                                         }
 781                                         if ( count ( $y ) == 1 )
 782                                                 $y = "{$z}<{$l}>{$y[0]}" ;
 783                                         else {
 784                                                 $attributes = $this->unstripForHTML( $y[0] );
 785                                                 $y = "{$z}<{$l}".Sanitizer::fixTagAttributes($attributes, $l).">{$y[1]}" ;
 786                                         }
 787                                         $t[$k] .= $y ;
 788                                         array_push ( $td , true ) ;
 789                                 }
 790                         }
 791                 }
 792
 793                 # Closing open td, tr && table
 794                 while ( count ( $td ) > 0 )
 795                 {
 796                         $l = array_pop ( $ltd ) ;
 797                         if ( array_pop ( $td ) ) $t[] = '</td>' ;
 798                         if ( array_pop ( $tr ) ) $t[] = '</tr>' ;
 799                         if ( !array_pop ( $has_opened_tr ) ) $t[] = "<tr><td></td></tr>" ;
 800                         $t[] = '</table>' ;
 801                 }
 802
 803                 $t = implode ( "\n" , $t ) ;
 804                 wfProfileOut( $fname );
 805                 return $t ;
 806         }
 807
 808         /**
 809          * Helper function for parse() that transforms wiki markup into
 810          * HTML. Only called for $mOutputType == OT_HTML.
 811          *
 812          * @access private
 813          */
 814         function internalParse( $text ) {
 815                 global $wgContLang;
 816                 $args = array();
 817                 $isMain = true;
 818                 $fname = 'Parser::internalParse';
 819                 wfProfileIn( $fname );
 820
 821                 # Remove <noinclude> tags and <includeonly> sections
 822                 $text = strtr( $text, array( '<onlyinclude>' => '' , '</onlyinclude>' => '' ) );
 823                 $text = strtr( $text, array( '<noinclude>' => '', '</noinclude>' => '') );
 824                 $text = preg_replace( '/<includeonly>.*?<\/includeonly>/s', '', $text );
 825
 826                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ) );
 827                 $text = $this->replaceVariables( $text, $args );
 828
 829                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 830
 831                 $text = $this->doHeadings( $text );
 832                 if($this->mOptions->getUseDynamicDates()) {
 833                         $df =& DateFormatter::getInstance();
 834                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 835                 }
 836                 $text = $this->doAllQuotes( $text );
 837                 $text = $this->replaceInternalLinks( $text );
 838                 $text = $this->replaceExternalLinks( $text );
 839
 840                 # replaceInternalLinks may sometimes leave behind
 841                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 842                 $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
 843
 844                 $text = $this->doMagicLinks( $text );
 845                 $text = $this->doTableStuff( $text );
 846                 $text = $this->formatHeadings( $text, $isMain );
 847
 848                 wfProfileOut( $fname );
 849                 return $text;
 850         }
 851
 852         /**
 853          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 854          * magic external links.
 855          *
 856          * @access private
 857          */
 858         function &doMagicLinks( &$text ) {
 859                 $text = $this->magicISBN( $text );
 860                 $text = $this->magicRFC( $text, 'RFC ', 'rfcurl' );
 861                 $text = $this->magicRFC( $text, 'PMID ', 'pubmedurl' );
 862                 return $text;
 863         }
 864
 865         /**
 866          * Parse headers and return html
 867          *
 868          * @access private
 869          */
 870         function doHeadings( $text ) {
 871                 $fname = 'Parser::doHeadings';
 872                 wfProfileIn( $fname );
 873                 for ( $i = 6; $i >= 1; --$i ) {
 874                         $h = str_repeat( '=', $i );
 875                         $text = preg_replace( "/^{$h}(.+){$h}(\\s|$)/m",
 876                           "<h{$i}>\\1</h{$i}>\\2", $text );
 877                 }
 878                 wfProfileOut( $fname );
 879                 return $text;
 880         }
 881
 882         /**
 883          * Replace single quotes with HTML markup
 884          * @access private
 885          * @return string the altered text
 886          */
 887         function doAllQuotes( $text ) {
 888                 $fname = 'Parser::doAllQuotes';
 889                 wfProfileIn( $fname );
 890                 $outtext = '';
 891                 $lines = explode( "\n", $text );
 892                 foreach ( $lines as $line ) {
 893                         $outtext .= $this->doQuotes ( $line ) . "\n";
 894                 }
 895                 $outtext = substr($outtext, 0,-1);
 896                 wfProfileOut( $fname );
 897                 return $outtext;
 898         }
 899
 900         /**
 901          * Helper function for doAllQuotes()
 902          * @access private
 903          */
 904         function doQuotes( $text ) {
 905                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
 906                 if ( count( $arr ) == 1 )
 907                         return $text;
 908                 else
 909                 {
 910                         # First, do some preliminary work. This may shift some apostrophes from
 911                         # being mark-up to being text. It also counts the number of occurrences
 912                         # of bold and italics mark-ups.
 913                         $i = 0;
 914                         $numbold = 0;
 915                         $numitalics = 0;
 916                         foreach ( $arr as $r )
 917                         {
 918                                 if ( ( $i % 2 ) == 1 )
 919                                 {
 920                                         # If there are ever four apostrophes, assume the first is supposed to
 921                                         # be text, and the remaining three constitute mark-up for bold text.
 922                                         if ( strlen( $arr[$i] ) == 4 )
 923                                         {
 924                                                 $arr[$i-1] .= "'";
 925                                                 $arr[$i] = "'''";
 926                                         }
 927                                         # If there are more than 5 apostrophes in a row, assume they're all
 928                                         # text except for the last 5.
 929                                         else if ( strlen( $arr[$i] ) > 5 )
 930                                         {
 931                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
 932                                                 $arr[$i] = "'''''";
 933                                         }
 934                                         # Count the number of occurrences of bold and italics mark-ups.
 935                                         # We are not counting sequences of five apostrophes.
 936                                         if ( strlen( $arr[$i] ) == 2 ) $numitalics++;  else
 937                                         if ( strlen( $arr[$i] ) == 3 ) $numbold++;     else
 938                                         if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
 939                                 }
 940                                 $i++;
 941                         }
 942
 943                         # If there is an odd number of both bold and italics, it is likely
 944                         # that one of the bold ones was meant to be an apostrophe followed
 945                         # by italics. Which one we cannot know for certain, but it is more
 946                         # likely to be one that has a single-letter word before it.
 947                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
 948                         {
 949                                 $i = 0;
 950                                 $firstsingleletterword = -1;
 951                                 $firstmultiletterword = -1;
 952                                 $firstspace = -1;
 953                                 foreach ( $arr as $r )
 954                                 {
 955                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
 956                                         {
 957                                                 $x1 = substr ($arr[$i-1], -1);
 958                                                 $x2 = substr ($arr[$i-1], -2, 1);
 959                                                 if ($x1 == ' ') {
 960                                                         if ($firstspace == -1) $firstspace = $i;
 961                                                 } else if ($x2 == ' ') {
 962                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
 963                                                 } else {
 964                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
 965                                                 }
 966                                         }
 967                                         $i++;
 968                                 }
 969
 970                                 # If there is a single-letter word, use it!
 971                                 if ($firstsingleletterword > -1)
 972                                 {
 973                                         $arr [ $firstsingleletterword ] = "''";
 974                                         $arr [ $firstsingleletterword-1 ] .= "'";
 975                                 }
 976                                 # If not, but there's a multi-letter word, use that one.
 977                                 else if ($firstmultiletterword > -1)
 978                                 {
 979                                         $arr [ $firstmultiletterword ] = "''";
 980                                         $arr [ $firstmultiletterword-1 ] .= "'";
 981                                 }
 982                                 # ... otherwise use the first one that has neither.
 983                                 # (notice that it is possible for all three to be -1 if, for example,
 984                                 # there is only one pentuple-apostrophe in the line)
 985                                 else if ($firstspace > -1)
 986                                 {
 987                                         $arr [ $firstspace ] = "''";
 988                                         $arr [ $firstspace-1 ] .= "'";
 989                                 }
 990                         }
 991
 992                         # Now let's actually convert our apostrophic mush to HTML!
 993                         $output = '';
 994                         $buffer = '';
 995                         $state = '';
 996                         $i = 0;
 997                         foreach ($arr as $r)
 998                         {
 999                                 if (($i % 2) == 0)
1000                                 {
1001                                         if ($state == 'both')
1002                                                 $buffer .= $r;
1003                                         else
1004                                                 $output .= $r;
1005                                 }
1006                                 else
1007                                 {
1008                                         if (strlen ($r) == 2)
1009                                         {
1010                                                 if ($state == 'i')
1011                                                 { $output .= '</i>'; $state = ''; }
1012                                                 else if ($state == 'bi')
1013                                                 { $output .= '</i>'; $state = 'b'; }
1014                                                 else if ($state == 'ib')
1015                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1016                                                 else if ($state == 'both')
1017                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1018                                                 else # $state can be 'b' or ''
1019                                                 { $output .= '<i>'; $state .= 'i'; }
1020                                         }
1021                                         else if (strlen ($r) == 3)
1022                                         {
1023                                                 if ($state == 'b')
1024                                                 { $output .= '</b>'; $state = ''; }
1025                                                 else if ($state == 'bi')
1026                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1027                                                 else if ($state == 'ib')
1028                                                 { $output .= '</b>'; $state = 'i'; }
1029                                                 else if ($state == 'both')
1030                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1031                                                 else # $state can be 'i' or ''
1032                                                 { $output .= '<b>'; $state .= 'b'; }
1033                                         }
1034                                         else if (strlen ($r) == 5)
1035                                         {
1036                                                 if ($state == 'b')
1037                                                 { $output .= '</b><i>'; $state = 'i'; }
1038                                                 else if ($state == 'i')
1039                                                 { $output .= '</i><b>'; $state = 'b'; }
1040                                                 else if ($state == 'bi')
1041                                                 { $output .= '</i></b>'; $state = ''; }
1042                                                 else if ($state == 'ib')
1043                                                 { $output .= '</b></i>'; $state = ''; }
1044                                                 else if ($state == 'both')
1045                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1046                                                 else # ($state == '')
1047                                                 { $buffer = ''; $state = 'both'; }
1048                                         }
1049                                 }
1050                                 $i++;
1051                         }
1052                         # Now close all remaining tags.  Notice that the order is important.
1053                         if ($state == 'b' || $state == 'ib')
1054                                 $output .= '</b>';
1055                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
1056                                 $output .= '</i>';
1057                         if ($state == 'bi')
1058                                 $output .= '</b>';
1059                         if ($state == 'both')
1060                                 $output .= '<b><i>'.$buffer.'</i></b>';
1061                         return $output;
1062                 }
1063         }
1064
1065         /**
1066          * Replace external links
1067          *
1068          * Note: this is all very hackish and the order of execution matters a lot.
1069          * Make sure to run maintenance/parserTests.php if you change this code.
1070          *
1071          * @access private
1072          */
1073         function replaceExternalLinks( $text ) {
1074                 global $wgContLang;
1075                 $fname = 'Parser::replaceExternalLinks';
1076                 wfProfileIn( $fname );
1077
1078                 $sk =& $this->mOptions->getSkin();
1079
1080                 $bits = preg_split( EXT_LINK_BRACKETED, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1081
1082                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
1083
1084                 $i = 0;
1085                 while ( $i<count( $bits ) ) {
1086                         $url = $bits[$i++];
1087                         $protocol = $bits[$i++];
1088                         $spaces = $bits[$i++];
1089                         $text = $bits[$i++];
1090                         $trail = $bits[$i++];
1091
1092                         # The characters '<' and '>' (which were escaped by
1093                         # removeHTMLtags()) should not be included in
1094                         # URLs, per RFC 2396.
1095                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1096                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1097                                 $url = substr($url, 0, $m2[0][1]);
1098                         }
1099
1100                         # If the link text is an image URL, replace it with an <img> tag
1101                         # This happened by accident in the original parser, but some people used it extensively
1102                         $img = $this->maybeMakeExternalImage( $text );
1103                         if ( $img !== false ) {
1104                                 $text = $img;
1105                         }
1106
1107                         $dtrail = '';
1108
1109                         # Set linktype for CSS - if URL==text, link is essentially free
1110                         $linktype = ($text == $url) ? 'free' : 'text';
1111
1112                         # No link text, e.g. [http://domain.tld/some.link]
1113                         if ( $text == '' ) {
1114                                 # Autonumber if allowed
1115                                 if ( strpos( HTTP_PROTOCOLS, str_replace('/','\/', $protocol) ) !== false ) {
1116                                         $text = '[' . ++$this->mAutonumber . ']';
1117                                         $linktype = 'autonumber';
1118                                 } else {
1119                                         # Otherwise just use the URL
1120                                         $text = htmlspecialchars( $url );
1121                                         $linktype = 'free';
1122                                 }
1123                         } else {
1124                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1125                                 # Check for trail
1126                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1127                         }
1128
1129                         $text = $wgContLang->markNoConversion($text);
1130
1131                         # Replace &amp; from obsolete syntax with &.
1132                         # All HTML entities will be escaped by makeExternalLink()
1133                         $url = str_replace( '&amp;', '&', $url );
1134                         # Replace unnecessary URL escape codes with the referenced character
1135                         # This prevents spammers from hiding links from the filters
1136                         $url = Parser::replaceUnusualEscapes( $url );
1137
1138                         # Process the trail (i.e. everything after this link up until start of the next link),
1139                         # replacing any non-bracketed links
1140                         $trail = $this->replaceFreeExternalLinks( $trail );
1141
1142                         # Use the encoded URL
1143                         # This means that users can paste URLs directly into the text
1144                         # Funny characters like &ouml; aren't valid in URLs anyway
1145                         # This was changed in August 2004
1146                         if ( strlen ( $spaces ) < 2 ) {
1147                                 # Normal case
1148                                 $s .= $sk->makeExternalLink( $url, $text, false, $linktype ) . $dtrail . $trail;
1149                         } else {
1150                                 # Fix for [url  text] (notice the two blanks)
1151                                 $s .= '[' . $sk->makeExternalLink( $url, $url, false, "free" ) . $spaces . $text . ']' . $dtrail . $trail;
1152                         }
1153
1154                         # Register link in the output object
1155                         $this->mOutput->addExternalLink( $url );
1156                 }
1157
1158                 wfProfileOut( $fname );
1159                 return $s;
1160         }
1161
1162         /**
1163          * Replace anything that looks like a URL with a link
1164          * @access private
1165          */
1166         function replaceFreeExternalLinks( $text ) {
1167                 global $wgContLang;
1168                 $fname = 'Parser::replaceFreeExternalLinks';
1169                 wfProfileIn( $fname );
1170
1171                 $bits = preg_split( '/(\b(?:' . wfUrlProtocols() . '))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1172                 $s = array_shift( $bits );
1173                 $i = 0;
1174
1175                 $sk =& $this->mOptions->getSkin();
1176
1177                 while ( $i < count( $bits ) ){
1178                         $protocol = $bits[$i++];
1179                         $remainder = $bits[$i++];
1180
1181                         if ( preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1182                                 # Found some characters after the protocol that look promising
1183                                 $url = $protocol . $m[1];
1184                                 $trail = $m[2];
1185
1186                                 # The characters '<' and '>' (which were escaped by
1187                                 # removeHTMLtags()) should not be included in
1188                                 # URLs, per RFC 2396.
1189                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1190                                         $trail = substr($url, $m2[0][1]) . $trail;
1191                                         $url = substr($url, 0, $m2[0][1]);
1192                                 }
1193
1194                                 # Move trailing punctuation to $trail
1195                                 $sep = ',;\.:!?';
1196                                 # If there is no left bracket, then consider right brackets fair game too
1197                                 if ( strpos( $url, '(' ) === false ) {
1198                                         $sep .= ')';
1199                                 }
1200
1201                                 $numSepChars = strspn( strrev( $url ), $sep );
1202                                 if ( $numSepChars ) {
1203                                         $trail = substr( $url, -$numSepChars ) . $trail;
1204                                         $url = substr( $url, 0, -$numSepChars );
1205                                 }
1206
1207                                 # Replace &amp; from obsolete syntax with &.
1208                                 # All HTML entities will be escaped by makeExternalLink()
1209                                 # or maybeMakeExternalImage()
1210                                 $url = str_replace( '&amp;', '&', $url );
1211                                 # Replace unnecessary URL escape codes with their equivalent characters
1212                                 $url = Parser::replaceUnusualEscapes( $url );
1213
1214                                 # Is this an external image?
1215                                 $text = $this->maybeMakeExternalImage( $url );
1216                                 if ( $text === false ) {
1217                                         # Not an image, make a link
1218                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free' );
1219                                         # Register it in the output object
1220                                         $this->mOutput->addExternalLink( $url );
1221                                 }
1222                                 $s .= $text . $trail;
1223                         } else {
1224                                 $s .= $protocol . $remainder;
1225                         }
1226                 }
1227                 wfProfileOut( $fname );
1228                 return $s;
1229         }
1230
1231         /**
1232          * Replace unusual URL escape codes with their equivalent characters
1233          * @param string
1234          * @return string
1235          * @static
1236          */
1237         function replaceUnusualEscapes( $url ) {
1238                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1239                         array( 'Parser', 'replaceUnusualEscapesCallback' ), $url );
1240         }
1241
1242         /**
1243          * Callback function used in replaceUnusualEscapes().
1244          * Replaces unusual URL escape codes with their equivalent character
1245          * @static
1246          * @access private
1247          */
1248         function replaceUnusualEscapesCallback( $matches ) {
1249                 $char = urldecode( $matches[0] );
1250                 $ord = ord( $char );
1251                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1252                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1253                         // No, shouldn't be escaped
1254                         return $char;
1255                 } else {
1256                         // Yes, leave it escaped
1257                         return $matches[0];
1258                 }
1259         }
1260
1261         /**
1262          * make an image if it's allowed, either through the global
1263          * option or through the exception
1264          * @access private
1265          */
1266         function maybeMakeExternalImage( $url ) {
1267                 $sk =& $this->mOptions->getSkin();
1268                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1269                 $imagesexception = !empty($imagesfrom);
1270                 $text = false;
1271                 if ( $this->mOptions->getAllowExternalImages()
1272                      || ( $imagesexception && strpos( $url, $imagesfrom ) === 0 ) ) {
1273                         if ( preg_match( EXT_IMAGE_REGEX, $url ) ) {
1274                                 # Image found
1275                                 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
1276                         }
1277                 }
1278                 return $text;
1279         }
1280
1281         /**
1282          * Process [[ ]] wikilinks
1283          *
1284          * @access private
1285          */
1286         function replaceInternalLinks( $s ) {
1287                 global $wgContLang;
1288                 static $fname = 'Parser::replaceInternalLinks' ;
1289
1290                 wfProfileIn( $fname );
1291
1292                 wfProfileIn( $fname.'-setup' );
1293                 static $tc = FALSE;
1294                 # the % is needed to support urlencoded titles as well
1295                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1296
1297                 $sk =& $this->mOptions->getSkin();
1298
1299                 #split the entire text string on occurences of [[
1300                 $a = explode( '[[', ' ' . $s );
1301                 #get the first element (all text up to first [[), and remove the space we added
1302                 $s = array_shift( $a );
1303                 $s = substr( $s, 1 );
1304
1305                 # Match a link having the form [[namespace:link|alternate]]trail
1306                 static $e1 = FALSE;
1307                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1308                 # Match cases where there is no "]]", which might still be images
1309                 static $e1_img = FALSE;
1310                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1311                 # Match the end of a line for a word that's not followed by whitespace,
1312                 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1313                 $e2 = wfMsgForContent( 'linkprefix' );
1314
1315                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1316
1317                 if( is_null( $this->mTitle ) ) {
1318                         wfDebugDieBacktrace( 'nooo' );
1319                 }
1320                 $nottalk = !$this->mTitle->isTalkPage();
1321
1322                 if ( $useLinkPrefixExtension ) {
1323                         if ( preg_match( $e2, $s, $m ) ) {
1324                                 $first_prefix = $m[2];
1325                         } else {
1326                                 $first_prefix = false;
1327                         }
1328                 } else {
1329                         $prefix = '';
1330                 }
1331
1332                 $selflink = $this->mTitle->getPrefixedText();
1333                 wfProfileOut( $fname.'-setup' );
1334
1335                 $checkVariantLink = sizeof($wgContLang->getVariants())>1;
1336                 $useSubpages = $this->areSubpagesAllowed();
1337
1338                 # Loop for each link
1339                 for ($k = 0; isset( $a[$k] ); $k++) {
1340                         $line = $a[$k];
1341                         if ( $useLinkPrefixExtension ) {
1342                                 wfProfileIn( $fname.'-prefixhandling' );
1343                                 if ( preg_match( $e2, $s, $m ) ) {
1344                                         $prefix = $m[2];
1345                                         $s = $m[1];
1346                                 } else {
1347                                         $prefix='';
1348                                 }
1349                                 # first link
1350                                 if($first_prefix) {
1351                                         $prefix = $first_prefix;
1352                                         $first_prefix = false;
1353                                 }
1354                                 wfProfileOut( $fname.'-prefixhandling' );
1355                         }
1356
1357                         $might_be_img = false;
1358
1359                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1360                                 $text = $m[2];
1361                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1362                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1363                                 # the real problem is with the $e1 regex
1364                                 # See bug 1300.
1365                                 #
1366                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1367                                 # and no image is in sight. See bug 2095.
1368                                 #
1369                                 if( $text !== '' && preg_match( "/^\](.*)/s", $m[3], $n ) ) {
1370                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1371                                         $m[3] = $n[1];
1372                                 }
1373                                 # fix up urlencoded title texts
1374                                 if(preg_match('/%/', $m[1] )) $m[1] = urldecode($m[1]);
1375                                 $trail = $m[3];
1376                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1377                                 $might_be_img = true;
1378                                 $text = $m[2];
1379                                 if(preg_match('/%/', $m[1] )) $m[1] = urldecode($m[1]);
1380                                 $trail = "";
1381                         } else { # Invalid form; output directly
1382                                 $s .= $prefix . '[[' . $line ;
1383                                 continue;
1384                         }
1385
1386                         # Don't allow internal links to pages containing
1387                         # PROTO: where PROTO is a valid URL protocol; these
1388                         # should be external links.
1389                         if (preg_match('/^(\b(?:' . wfUrlProtocols() . '))/', $m[1])) {
1390                                 $s .= $prefix . '[[' . $line ;
1391                                 continue;
1392                         }
1393
1394                         # Make subpage if necessary
1395                         if( $useSubpages ) {
1396                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1397                         } else {
1398                                 $link = $m[1];
1399                         }
1400
1401                         $noforce = (substr($m[1], 0, 1) != ':');
1402                         if (!$noforce) {
1403                                 # Strip off leading ':'
1404                                 $link = substr($link, 1);
1405                         }
1406
1407                         $nt = Title::newFromText( $this->unstripNoWiki($link, $this->mStripState) );
1408                         if( !$nt ) {
1409                                 $s .= $prefix . '[[' . $line;
1410                                 continue;
1411                         }
1412
1413                         #check other language variants of the link
1414                         #if the article does not exist
1415                         if( $checkVariantLink
1416                             && $nt->getArticleID() == 0 ) {
1417                                 $wgContLang->findVariantLink($link, $nt);
1418                         }
1419
1420                         $ns = $nt->getNamespace();
1421                         $iw = $nt->getInterWiki();
1422
1423                         if ($might_be_img) { # if this is actually an invalid link
1424                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1425                                         $found = false;
1426                                         while (isset ($a[$k+1]) ) {
1427                                                 #look at the next 'line' to see if we can close it there
1428                                                 $spliced = array_splice( $a, $k + 1, 1 );
1429                                                 $next_line = array_shift( $spliced );
1430                                                 if( preg_match("/^(.*?]].*?)]](.*)$/sD", $next_line, $m) ) {
1431                                                 # the first ]] closes the inner link, the second the image
1432                                                         $found = true;
1433                                                         $text .= '[[' . $m[1];
1434                                                         $trail = $m[2];
1435                                                         break;
1436                                                 } elseif( preg_match("/^.*?]].*$/sD", $next_line, $m) ) {
1437                                                         #if there's exactly one ]] that's fine, we'll keep looking
1438                                                         $text .= '[[' . $m[0];
1439                                                 } else {
1440                                                         #if $next_line is invalid too, we need look no further
1441                                                         $text .= '[[' . $next_line;
1442                                                         break;
1443                                                 }
1444                                         }
1445                                         if ( !$found ) {
1446                                                 # we couldn't find the end of this imageLink, so output it raw
1447                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1448                                                 $text = $this->replaceInternalLinks($text);
1449                                                 $s .= $prefix . '[[' . $link . '|' . $text;
1450                                                 # note: no $trail, because without an end, there *is* no trail
1451                                                 continue;
1452                                         }
1453                                 } else { #it's not an image, so output it raw
1454                                         $s .= $prefix . '[[' . $link . '|' . $text;
1455                                         # note: no $trail, because without an end, there *is* no trail
1456                                         continue;
1457                                 }
1458                         }
1459
1460                         $wasblank = ( '' == $text );
1461                         if( $wasblank ) $text = $link;
1462
1463
1464                         # Link not escaped by : , create the various objects
1465                         if( $noforce ) {
1466
1467                                 # Interwikis
1468                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1469                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1470                                         $s = rtrim($s . "\n");
1471                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1472                                         continue;
1473                                 }
1474
1475                                 if ( $ns == NS_IMAGE ) {
1476                                         wfProfileIn( "$fname-image" );
1477                                         if ( !wfIsBadImage( $nt->getDBkey() ) ) {
1478                                                 # recursively parse links inside the image caption
1479                                                 # actually, this will parse them in any other parameters, too,
1480                                                 # but it might be hard to fix that, and it doesn't matter ATM
1481                                                 $text = $this->replaceExternalLinks($text);
1482                                                 $text = $this->replaceInternalLinks($text);
1483
1484                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1485                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text ) ) . $trail;
1486                                                 $this->mOutput->addImage( $nt->getDBkey() );
1487
1488                                                 wfProfileOut( "$fname-image" );
1489                                                 continue;
1490                                         }
1491                                         wfProfileOut( "$fname-image" );
1492
1493                                 }
1494
1495                                 if ( $ns == NS_CATEGORY ) {
1496                                         wfProfileIn( "$fname-category" );
1497                                         $s = rtrim($s . "\n"); # bug 87
1498
1499                                         if ( $wasblank ) {
1500                                                 if ( $this->mTitle->getNamespace() == NS_CATEGORY ) {
1501                                                         $sortkey = $this->mTitle->getText();
1502                                                 } else {
1503                                                         $sortkey = $this->mTitle->getPrefixedText();
1504                                                 }
1505                                         } else {
1506                                                 $sortkey = $text;
1507                                         }
1508                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1509                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1510                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1511
1512                                         /**
1513                                          * Strip the whitespace Category links produce, see bug 87
1514                                          * @todo We might want to use trim($tmp, "\n") here.
1515                                          */
1516                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1517
1518                                         wfProfileOut( "$fname-category" );
1519                                         continue;
1520                                 }
1521                         }
1522
1523                         if( ( $nt->getPrefixedText() === $selflink ) &&
1524                             ( $nt->getFragment() === '' ) ) {
1525                                 # Self-links are handled specially; generally de-link and change to bold.
1526                                 $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1527                                 continue;
1528                         }
1529
1530                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1531                         if( $ns == NS_MEDIA ) {
1532                                 $link = $sk->makeMediaLinkObj( $nt, $text );
1533                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1534                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1535                                 $this->mOutput->addImage( $nt->getDBkey() );
1536                                 continue;
1537                         } elseif( $ns == NS_SPECIAL ) {
1538                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1539                                 continue;
1540                         } elseif( $ns == NS_IMAGE ) {
1541                                 $img = Image::newFromTitle( $nt );
1542                                 if( $img->exists() ) {
1543                                         // Force a blue link if the file exists; may be a remote
1544                                         // upload on the shared repository, and we want to see its
1545                                         // auto-generated page.
1546                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1547                                         continue;
1548                                 }
1549                         }
1550                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1551                 }
1552                 wfProfileOut( $fname );
1553                 return $s;
1554         }
1555
1556         /**
1557          * Make a link placeholder. The text returned can be later resolved to a real link with
1558          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1559          * parsing of interwiki links, and secondly to allow all extistence checks and
1560          * article length checks (for stub links) to be bundled into a single query.
1561          *
1562          */
1563         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1564                 if ( ! is_object($nt) ) {
1565                         # Fail gracefully
1566                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1567                 } else {
1568                         # Separate the link trail from the rest of the link
1569                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1570
1571                         if ( $nt->isExternal() ) {
1572                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1573                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1574                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1575                         } else {
1576                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1577                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1578                                 $this->mLinkHolders['queries'][] = $query;
1579                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1580                                 $this->mLinkHolders['titles'][] = $nt;
1581
1582                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1583                         }
1584                 }
1585                 return $retVal;
1586         }
1587
1588         /**
1589          * Render a forced-blue link inline; protect against double expansion of
1590          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1591          * Since this little disaster has to split off the trail text to avoid
1592          * breaking URLs in the following text without breaking trails on the
1593          * wiki links, it's been made into a horrible function.
1594          *
1595          * @param Title $nt
1596          * @param string $text
1597          * @param string $query
1598          * @param string $trail
1599          * @param string $prefix
1600          * @return string HTML-wikitext mix oh yuck
1601          */
1602         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1603                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1604                 $sk =& $this->mOptions->getSkin();
1605                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1606                 return $this->armorLinks( $link ) . $trail;
1607         }
1608
1609         /**
1610          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1611          * going to go through further parsing steps before inline URL expansion.
1612          *
1613          * In particular this is important when using action=render, which causes
1614          * full URLs to be included.
1615          *
1616          * Oh man I hate our multi-layer parser!
1617          *
1618          * @param string more-or-less HTML
1619          * @return string less-or-more HTML with NOPARSE bits
1620          */
1621         function armorLinks( $text ) {
1622                 return preg_replace( "/\b(" . wfUrlProtocols() . ')/',
1623                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1624         }
1625
1626         /**
1627          * Return true if subpage links should be expanded on this page.
1628          * @return bool
1629          */
1630         function areSubpagesAllowed() {
1631                 # Some namespaces don't allow subpages
1632                 global $wgNamespacesWithSubpages;
1633                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1634         }
1635
1636         /**
1637          * Handle link to subpage if necessary
1638          * @param string $target the source of the link
1639          * @param string &$text the link text, modified as necessary
1640          * @return string the full name of the link
1641          * @access private
1642          */
1643         function maybeDoSubpageLink($target, &$text) {
1644                 # Valid link forms:
1645                 # Foobar -- normal
1646                 # :Foobar -- override special treatment of prefix (images, language links)
1647                 # /Foobar -- convert to CurrentPage/Foobar
1648                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1649                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1650                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1651
1652                 $fname = 'Parser::maybeDoSubpageLink';
1653                 wfProfileIn( $fname );
1654                 $ret = $target; # default return value is no change
1655
1656                 # Some namespaces don't allow subpages,
1657                 # so only perform processing if subpages are allowed
1658                 if( $this->areSubpagesAllowed() ) {
1659                         # Look at the first character
1660                         if( $target != '' && $target{0} == '/' ) {
1661                                 # / at end means we don't want the slash to be shown
1662                                 if( substr( $target, -1, 1 ) == '/' ) {
1663                                         $target = substr( $target, 1, -1 );
1664                                         $noslash = $target;
1665                                 } else {
1666                                         $noslash = substr( $target, 1 );
1667                                 }
1668
1669                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash);
1670                                 if( '' === $text ) {
1671                                         $text = $target;
1672                                 } # this might be changed for ugliness reasons
1673                         } else {
1674                                 # check for .. subpage backlinks
1675                                 $dotdotcount = 0;
1676                                 $nodotdot = $target;
1677                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1678                                         ++$dotdotcount;
1679                                         $nodotdot = substr( $nodotdot, 3 );
1680                                 }
1681                                 if($dotdotcount > 0) {
1682                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1683                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1684                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1685                                                 # / at the end means don't show full path
1686                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1687                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1688                                                         if( '' === $text ) {
1689                                                                 $text = $nodotdot;
1690                                                         }
1691                                                 }
1692                                                 $nodotdot = trim( $nodotdot );
1693                                                 if( $nodotdot != '' ) {
1694                                                         $ret .= '/' . $nodotdot;
1695                                                 }
1696                                         }
1697                                 }
1698                         }
1699                 }
1700
1701                 wfProfileOut( $fname );
1702                 return $ret;
1703         }
1704
1705         /**#@+
1706          * Used by doBlockLevels()
1707          * @access private
1708          */
1709         /* private */ function closeParagraph() {
1710                 $result = '';
1711                 if ( '' != $this->mLastSection ) {
1712                         $result = '</' . $this->mLastSection  . ">\n";
1713                 }
1714                 $this->mInPre = false;
1715                 $this->mLastSection = '';
1716                 return $result;
1717         }
1718         # getCommon() returns the length of the longest common substring
1719         # of both arguments, starting at the beginning of both.
1720         #
1721         /* private */ function getCommon( $st1, $st2 ) {
1722                 $fl = strlen( $st1 );
1723                 $shorter = strlen( $st2 );
1724                 if ( $fl < $shorter ) { $shorter = $fl; }
1725
1726                 for ( $i = 0; $i < $shorter; ++$i ) {
1727                         if ( $st1{$i} != $st2{$i} ) { break; }
1728                 }
1729                 return $i;
1730         }
1731         # These next three functions open, continue, and close the list
1732         # element appropriate to the prefix character passed into them.
1733         #
1734         /* private */ function openList( $char ) {
1735                 $result = $this->closeParagraph();
1736
1737                 if ( '*' == $char ) { $result .= '<ul><li>'; }
1738                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
1739                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
1740                 else if ( ';' == $char ) {
1741                         $result .= '<dl><dt>';
1742                         $this->mDTopen = true;
1743                 }
1744                 else { $result = '<!-- ERR 1 -->'; }
1745
1746                 return $result;
1747         }
1748
1749         /* private */ function nextItem( $char ) {
1750                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
1751                 else if ( ':' == $char || ';' == $char ) {
1752                         $close = '</dd>';
1753                         if ( $this->mDTopen ) { $close = '</dt>'; }
1754                         if ( ';' == $char ) {
1755                                 $this->mDTopen = true;
1756                                 return $close . '<dt>';
1757                         } else {
1758                                 $this->mDTopen = false;
1759                                 return $close . '<dd>';
1760                         }
1761                 }
1762                 return '<!-- ERR 2 -->';
1763         }
1764
1765         /* private */ function closeList( $char ) {
1766                 if ( '*' == $char ) { $text = '</li></ul>'; }
1767                 else if ( '#' == $char ) { $text = '</li></ol>'; }
1768                 else if ( ':' == $char ) {
1769                         if ( $this->mDTopen ) {
1770                                 $this->mDTopen = false;
1771                                 $text = '</dt></dl>';
1772                         } else {
1773                                 $text = '</dd></dl>';
1774                         }
1775                 }
1776                 else {  return '<!-- ERR 3 -->'; }
1777                 return $text."\n";
1778         }
1779         /**#@-*/
1780
1781         /**
1782          * Make lists from lines starting with ':', '*', '#', etc.
1783          *
1784          * @access private
1785          * @return string the lists rendered as HTML
1786          */
1787         function doBlockLevels( $text, $linestart ) {
1788                 $fname = 'Parser::doBlockLevels';
1789                 wfProfileIn( $fname );
1790
1791                 # Parsing through the text line by line.  The main thing
1792                 # happening here is handling of block-level elements p, pre,
1793                 # and making lists from lines starting with * # : etc.
1794                 #
1795                 $textLines = explode( "\n", $text );
1796
1797                 $lastPrefix = $output = '';
1798                 $this->mDTopen = $inBlockElem = false;
1799                 $prefixLength = 0;
1800                 $paragraphStack = false;
1801
1802                 if ( !$linestart ) {
1803                         $output .= array_shift( $textLines );
1804                 }
1805                 foreach ( $textLines as $oLine ) {
1806                         $lastPrefixLength = strlen( $lastPrefix );
1807                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
1808                         $preOpenMatch = preg_match('/<pre/i', $oLine );
1809                         if ( !$this->mInPre ) {
1810                                 # Multiple prefixes may abut each other for nested lists.
1811                                 $prefixLength = strspn( $oLine, '*#:;' );
1812                                 $pref = substr( $oLine, 0, $prefixLength );
1813
1814                                 # eh?
1815                                 $pref2 = str_replace( ';', ':', $pref );
1816                                 $t = substr( $oLine, $prefixLength );
1817                                 $this->mInPre = !empty($preOpenMatch);
1818                         } else {
1819                                 # Don't interpret any other prefixes in preformatted text
1820                                 $prefixLength = 0;
1821                                 $pref = $pref2 = '';
1822                                 $t = $oLine;
1823                         }
1824
1825                         # List generation
1826                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
1827                                 # Same as the last item, so no need to deal with nesting or opening stuff
1828                                 $output .= $this->nextItem( substr( $pref, -1 ) );
1829                                 $paragraphStack = false;
1830
1831                                 if ( substr( $pref, -1 ) == ';') {
1832                                         # The one nasty exception: definition lists work like this:
1833                                         # ; title : definition text
1834                                         # So we check for : in the remainder text to split up the
1835                                         # title and definition, without b0rking links.
1836                                         $term = $t2 = '';
1837                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
1838                                                 $t = $t2;
1839                                                 $output .= $term . $this->nextItem( ':' );
1840                                         }
1841                                 }
1842                         } elseif( $prefixLength || $lastPrefixLength ) {
1843                                 # Either open or close a level...
1844                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
1845                                 $paragraphStack = false;
1846
1847                                 while( $commonPrefixLength < $lastPrefixLength ) {
1848                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
1849                                         --$lastPrefixLength;
1850                                 }
1851                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
1852                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
1853                                 }
1854                                 while ( $prefixLength > $commonPrefixLength ) {
1855                                         $char = substr( $pref, $commonPrefixLength, 1 );
1856                                         $output .= $this->openList( $char );
1857
1858                                         if ( ';' == $char ) {
1859                                                 # FIXME: This is dupe of code above
1860                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
1861                                                         $t = $t2;
1862                                                         $output .= $term . $this->nextItem( ':' );
1863                                                 }
1864                                         }
1865                                         ++$commonPrefixLength;
1866                                 }
1867                                 $lastPrefix = $pref2;
1868                         }
1869                         if( 0 == $prefixLength ) {
1870                                 wfProfileIn( "$fname-paragraph" );
1871                                 # No prefix (not in list)--go to paragraph mode
1872                                 // XXX: use a stack for nestable elements like span, table and div
1873                                 $openmatch = preg_match('/(<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
1874                                 $closematch = preg_match(
1875                                         '/(<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
1876                                         '<td|<th|<div|<\\/div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul)/iS', $t );
1877                                 if ( $openmatch or $closematch ) {
1878                                         $paragraphStack = false;
1879                                         $output .= $this->closeParagraph();
1880                                         if ( $preOpenMatch and !$preCloseMatch ) {
1881                                                 $this->mInPre = true;
1882                                         }
1883                                         if ( $closematch ) {
1884                                                 $inBlockElem = false;
1885                                         } else {
1886                                                 $inBlockElem = true;
1887                                         }
1888                                 } else if ( !$inBlockElem && !$this->mInPre ) {
1889                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
1890                                                 // pre
1891                                                 if ($this->mLastSection != 'pre') {
1892                                                         $paragraphStack = false;
1893                                                         $output .= $this->closeParagraph().'<pre>';
1894                                                         $this->mLastSection = 'pre';
1895                                                 }
1896                                                 $t = substr( $t, 1 );
1897                                         } else {
1898                                                 // paragraph
1899                                                 if ( '' == trim($t) ) {
1900                                                         if ( $paragraphStack ) {
1901                                                                 $output .= $paragraphStack.'<br />';
1902                                                                 $paragraphStack = false;
1903                                                                 $this->mLastSection = 'p';
1904                                                         } else {
1905                                                                 if ($this->mLastSection != 'p' ) {
1906                                                                         $output .= $this->closeParagraph();
1907                                                                         $this->mLastSection = '';
1908                                                                         $paragraphStack = '<p>';
1909                                                                 } else {
1910                                                                         $paragraphStack = '</p><p>';
1911                                                                 }
1912                                                         }
1913                                                 } else {
1914                                                         if ( $paragraphStack ) {
1915                                                                 $output .= $paragraphStack;
1916                                                                 $paragraphStack = false;
1917                                                                 $this->mLastSection = 'p';
1918                                                         } else if ($this->mLastSection != 'p') {
1919                                                                 $output .= $this->closeParagraph().'<p>';
1920                                                                 $this->mLastSection = 'p';
1921                                                         }
1922                                                 }
1923                                         }
1924                                 }
1925                                 wfProfileOut( "$fname-paragraph" );
1926                         }
1927                         // somewhere above we forget to get out of pre block (bug 785)
1928                         if($preCloseMatch && $this->mInPre) {
1929                                 $this->mInPre = false;
1930                         }
1931                         if ($paragraphStack === false) {
1932                                 $output .= $t."\n";
1933                         }
1934                 }
1935                 while ( $prefixLength ) {
1936                         $output .= $this->closeList( $pref2{$prefixLength-1} );
1937                         --$prefixLength;
1938                 }
1939                 if ( '' != $this->mLastSection ) {
1940                         $output .= '</' . $this->mLastSection . '>';
1941                         $this->mLastSection = '';
1942                 }
1943
1944                 wfProfileOut( $fname );
1945                 return $output;
1946         }
1947
1948         /**
1949          * Split up a string on ':', ignoring any occurences inside
1950          * <a>..</a> or <span>...</span>
1951          * @param string $str the string to split
1952          * @param string &$before set to everything before the ':'
1953          * @param string &$after set to everything after the ':'
1954          * return string the position of the ':', or false if none found
1955          */
1956         function findColonNoLinks($str, &$before, &$after) {
1957                 # I wonder if we should make this count all tags, not just <a>
1958                 # and <span>. That would prevent us from matching a ':' that
1959                 # comes in the middle of italics other such formatting....
1960                 # -- Wil
1961                 $fname = 'Parser::findColonNoLinks';
1962                 wfProfileIn( $fname );
1963                 $pos = 0;
1964                 do {
1965                         $colon = strpos($str, ':', $pos);
1966
1967                         if ($colon !== false) {
1968                                 $before = substr($str, 0, $colon);
1969                                 $after = substr($str, $colon + 1);
1970
1971                                 # Skip any ':' within <a> or <span> pairs
1972                                 $a = substr_count($before, '<a');
1973                                 $s = substr_count($before, '<span');
1974                                 $ca = substr_count($before, '</a>');
1975                                 $cs = substr_count($before, '</span>');
1976
1977                                 if ($a <= $ca and $s <= $cs) {
1978                                         # Tags are balanced before ':'; ok
1979                                         break;
1980                                 }
1981                                 $pos = $colon + 1;
1982                         }
1983                 } while ($colon !== false);
1984                 wfProfileOut( $fname );
1985                 return $colon;
1986         }
1987
1988         /**
1989          * Return value of a magic variable (like PAGENAME)
1990          *
1991          * @access private
1992          */
1993         function getVariableValue( $index ) {
1994                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
1995
1996                 /**
1997                  * Some of these require message or data lookups and can be
1998                  * expensive to check many times.
1999                  */
2000                 static $varCache = array();
2001                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$varCache ) ) )
2002                         if ( isset( $varCache[$index] ) )
2003                                 return $varCache[$index];
2004
2005                 $ts = time();
2006                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2007
2008                 switch ( $index ) {
2009                         case MAG_CURRENTMONTH:
2010                                 return $varCache[$index] = $wgContLang->formatNum( date( 'm', $ts ) );
2011                         case MAG_CURRENTMONTHNAME:
2012                                 return $varCache[$index] = $wgContLang->getMonthName( date( 'n', $ts ) );
2013                         case MAG_CURRENTMONTHNAMEGEN:
2014                                 return $varCache[$index] = $wgContLang->getMonthNameGen( date( 'n', $ts ) );
2015                         case MAG_CURRENTMONTHABBREV:
2016                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( date( 'n', $ts ) );
2017                         case MAG_CURRENTDAY:
2018                                 return $varCache[$index] = $wgContLang->formatNum( date( 'j', $ts ) );
2019                         case MAG_CURRENTDAY2:
2020                                 return $varCache[$index] = $wgContLang->formatNum( date( 'd', $ts ) );
2021                         case MAG_PAGENAME:
2022                                 return $this->mTitle->getText();
2023                         case MAG_PAGENAMEE:
2024                                 return $this->mTitle->getPartialURL();
2025                         case MAG_FULLPAGENAME:
2026                                 return $this->mTitle->getPrefixedText();
2027                         case MAG_FULLPAGENAMEE:
2028                                 return $this->mTitle->getPrefixedURL();
2029                         case MAG_REVISIONID:
2030                                 return $this->mRevisionId;
2031                         case MAG_NAMESPACE:
2032                                 return $wgContLang->getNsText( $this->mTitle->getNamespace() );
2033                         case MAG_NAMESPACEE:
2034                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2035                         case MAG_CURRENTDAYNAME:
2036                                 return $varCache[$index] = $wgContLang->getWeekdayName( date( 'w', $ts ) + 1 );
2037                         case MAG_CURRENTYEAR:
2038                                 return $varCache[$index] = $wgContLang->formatNum( date( 'Y', $ts ), true );
2039                         case MAG_CURRENTTIME:
2040                                 return $varCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2041                         case MAG_CURRENTWEEK:
2042                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2043                                 // int to remove the padding
2044                                 return $varCache[$index] = $wgContLang->formatNum( (int)date( 'W', $ts ) );
2045                         case MAG_CURRENTDOW:
2046                                 return $varCache[$index] = $wgContLang->formatNum( date( 'w', $ts ) );
2047                         case MAG_NUMBEROFARTICLES:
2048                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfArticles() );
2049                         case MAG_NUMBEROFFILES:
2050                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfFiles() );
2051                         case MAG_SITENAME:
2052                                 return $wgSitename;
2053                         case MAG_SERVER:
2054                                 return $wgServer;
2055                         case MAG_SERVERNAME:
2056                                 return $wgServerName;
2057                         case MAG_SCRIPTPATH:
2058                                 return $wgScriptPath;
2059                         default:
2060                                 $ret = null;
2061                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$varCache, &$index, &$ret ) ) )
2062                                         return $ret;
2063                                 else
2064                                         return null;
2065                 }
2066         }
2067
2068         /**
2069          * initialise the magic variables (like CURRENTMONTHNAME)
2070          *
2071          * @access private
2072          */
2073         function initialiseVariables() {
2074                 $fname = 'Parser::initialiseVariables';
2075                 wfProfileIn( $fname );
2076                 global $wgVariableIDs;
2077                 $this->mVariables = array();
2078                 foreach ( $wgVariableIDs as $id ) {
2079                         $mw =& MagicWord::get( $id );
2080                         $mw->addToArray( $this->mVariables, $id );
2081                 }
2082                 wfProfileOut( $fname );
2083         }
2084
2085         /**
2086          * parse any parentheses in format ((title|part|part))
2087          * and call callbacks to get a replacement text for any found piece
2088          *
2089          * @param string $text The text to parse
2090          * @param array $callbacks rules in form:
2091          *     '{' => array(                            # opening parentheses
2092          *                                      'end' => '}',   # closing parentheses
2093          *                                      'cb' => array(2 => callback,    # replacement callback to call if {{..}} is found
2094          *                                                                4 => callback         # replacement callback to call if {{{{..}}}} is found
2095          *                                                                )
2096          *                                      )
2097          * @access private
2098          */
2099         function replace_callback ($text, $callbacks) {
2100                 $openingBraceStack = array();   # this array will hold a stack of parentheses which are not closed yet
2101                 $lastOpeningBrace = -1;                 # last not closed parentheses
2102
2103                 for ($i = 0; $i < strlen($text); $i++) {
2104                         # check for any opening brace
2105                         $rule = null;
2106                         $nextPos = -1;
2107                         foreach ($callbacks as $key => $value) {
2108                                 $pos = strpos ($text, $key, $i);
2109                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)) {
2110                                         $rule = $value;
2111                                         $nextPos = $pos;
2112                                 }
2113                         }
2114
2115                         if ($lastOpeningBrace >= 0) {
2116                                 $pos = strpos ($text, $openingBraceStack[$lastOpeningBrace]['braceEnd'], $i);
2117
2118                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)){
2119                                         $rule = null;
2120                                         $nextPos = $pos;
2121                                 }
2122
2123                                 $pos = strpos ($text, '|', $i);
2124
2125                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)){
2126                                         $rule = null;
2127                                         $nextPos = $pos;
2128                                 }
2129                         }
2130
2131                         if ($nextPos == -1)
2132                                 break;
2133
2134                         $i = $nextPos;
2135
2136                         # found openning brace, lets add it to parentheses stack
2137                         if (null != $rule) {
2138                                 $piece = array('brace' => $text[$i],
2139                                                            'braceEnd' => $rule['end'],
2140                                                            'count' => 1,
2141                                                            'title' => '',
2142                                                            'parts' => null);
2143
2144                                 # count openning brace characters
2145                                 while ($i+1 < strlen($text) && $text[$i+1] == $piece['brace']) {
2146                                         $piece['count']++;
2147                                         $i++;
2148                                 }
2149
2150                                 $piece['startAt'] = $i+1;
2151                                 $piece['partStart'] = $i+1;
2152
2153                                 # we need to add to stack only if openning brace count is enough for any given rule
2154                                 foreach ($rule['cb'] as $cnt => $fn) {
2155                                         if ($piece['count'] >= $cnt) {
2156                                                 $lastOpeningBrace ++;
2157                                                 $openingBraceStack[$lastOpeningBrace] = $piece;
2158                                                 break;
2159                                         }
2160                                 }
2161
2162                                 continue;
2163                         }
2164                         else if ($lastOpeningBrace >= 0) {
2165                                 # first check if it is a closing brace
2166                                 if ($openingBraceStack[$lastOpeningBrace]['braceEnd'] == $text[$i]) {
2167                                         # lets check if it is enough characters for closing brace
2168                                         $count = 1;
2169                                         while ($i+$count < strlen($text) && $text[$i+$count] == $text[$i])
2170                                                 $count++;
2171
2172                                         # if there are more closing parentheses than opening ones, we parse less
2173                                         if ($openingBraceStack[$lastOpeningBrace]['count'] < $count)
2174                                                 $count = $openingBraceStack[$lastOpeningBrace]['count'];
2175
2176                                         # check for maximum matching characters (if there are 5 closing characters, we will probably need only 3 - depending on the rules)
2177                                         $matchingCount = 0;
2178                                         $matchingCallback = null;
2179                                         foreach ($callbacks[$openingBraceStack[$lastOpeningBrace]['brace']]['cb'] as $cnt => $fn) {
2180                                                 if ($count >= $cnt && $matchingCount < $cnt) {
2181                                                         $matchingCount = $cnt;
2182                                                         $matchingCallback = $fn;
2183                                                 }
2184                                         }
2185
2186                                         if ($matchingCount == 0) {
2187                                                 $i += $count - 1;
2188                                                 continue;
2189                                         }
2190
2191                                         # lets set a title or last part (if '|' was found)
2192                                         if (null === $openingBraceStack[$lastOpeningBrace]['parts'])
2193                                                 $openingBraceStack[$lastOpeningBrace]['title'] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2194                                         else
2195                                                 $openingBraceStack[$lastOpeningBrace]['parts'][] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2196
2197                                         $pieceStart = $openingBraceStack[$lastOpeningBrace]['startAt'] - $matchingCount;
2198                                         $pieceEnd = $i + $matchingCount;
2199
2200                                         if( is_callable( $matchingCallback ) ) {
2201                                                 $cbArgs = array (
2202                                                                                  'text' => substr($text, $pieceStart, $pieceEnd - $pieceStart),
2203                                                                                  'title' => trim($openingBraceStack[$lastOpeningBrace]['title']),
2204                                                                                  'parts' => $openingBraceStack[$lastOpeningBrace]['parts'],
2205                                                                                  'lineStart' => (($pieceStart > 0) && ($text[$pieceStart-1] == '\n')),
2206                                                                                  );
2207                                                 # finally we can call a user callback and replace piece of text
2208                                                 $replaceWith = call_user_func( $matchingCallback, $cbArgs );
2209                                                 $text = substr($text, 0, $pieceStart) . $replaceWith . substr($text, $pieceEnd);
2210                                                 $i = $pieceStart + strlen($replaceWith) - 1;
2211                                         }
2212                                         else {
2213                                                 # null value for callback means that parentheses should be parsed, but not replaced
2214                                                 $i += $matchingCount - 1;
2215                                         }
2216
2217                                         # reset last openning parentheses, but keep it in case there are unused characters
2218                                         $piece = array('brace' => $openingBraceStack[$lastOpeningBrace]['brace'],
2219                                                                    'braceEnd' => $openingBraceStack[$lastOpeningBrace]['braceEnd'],
2220                                                                    'count' => $openingBraceStack[$lastOpeningBrace]['count'],
2221                                                                    'title' => '',
2222                                                                    'parts' => null,
2223                                                                    'startAt' => $openingBraceStack[$lastOpeningBrace]['startAt']);
2224                                         $openingBraceStack[$lastOpeningBrace--] = null;
2225
2226                                         if ($matchingCount < $piece['count']) {
2227                                                 $piece['count'] -= $matchingCount;
2228                                                 $piece['startAt'] -= $matchingCount;
2229                                                 $piece['partStart'] = $piece['startAt'];
2230                                                 # do we still qualify for any callback with remaining count?
2231                                                 foreach ($callbacks[$piece['brace']]['cb'] as $cnt => $fn) {
2232                                                         if ($piece['count'] >= $cnt) {
2233                                                                 $lastOpeningBrace ++;
2234                                                                 $openingBraceStack[$lastOpeningBrace] = $piece;
2235                                                                 break;
2236                                                         }
2237                                                 }
2238                                         }
2239                                         continue;
2240                                 }
2241
2242                                 # lets set a title if it is a first separator, or next part otherwise
2243                                 if ($text[$i] == '|') {
2244                                         if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2245                                                 $openingBraceStack[$lastOpeningBrace]['title'] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2246                                                 $openingBraceStack[$lastOpeningBrace]['parts'] = array();
2247                                         }
2248                                         else
2249                                                 $openingBraceStack[$lastOpeningBrace]['parts'][] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2250
2251                                         $openingBraceStack[$lastOpeningBrace]['partStart'] = $i + 1;
2252                                 }
2253                         }
2254                 }
2255
2256                 return $text;
2257         }
2258
2259         /**
2260          * Replace magic variables, templates, and template arguments
2261          * with the appropriate text. Templates are substituted recursively,
2262          * taking care to avoid infinite loops.
2263          *
2264          * Note that the substitution depends on value of $mOutputType:
2265          *  OT_WIKI: only {{subst:}} templates
2266          *  OT_MSG: only magic variables
2267          *  OT_HTML: all templates and magic variables
2268          *
2269          * @param string $tex The text to transform
2270          * @param array $args Key-value pairs representing template parameters to substitute
2271          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2272          * @access private
2273          */
2274         function replaceVariables( $text, $args = array(), $argsOnly = false ) {
2275                 # Prevent too big inclusions
2276                 if( strlen( $text ) > MAX_INCLUDE_SIZE ) {
2277                         return $text;
2278                 }
2279
2280                 $fname = 'Parser::replaceVariables';
2281                 wfProfileIn( $fname );
2282
2283                 # This function is called recursively. To keep track of arguments we need a stack:
2284                 array_push( $this->mArgStack, $args );
2285
2286                 $braceCallbacks = array();
2287                 if ( !$argsOnly ) {
2288                         $braceCallbacks[2] = array( &$this, 'braceSubstitution' );
2289                 }
2290                 if ( $this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI ) {
2291                         $braceCallbacks[3] = array( &$this, 'argSubstitution' );
2292                 }
2293                 $callbacks = array();
2294                 $callbacks['{'] = array('end' => '}', 'cb' => $braceCallbacks);
2295                 $callbacks['['] = array('end' => ']', 'cb' => array(2=>null));
2296                 $text = $this->replace_callback ($text, $callbacks);
2297
2298                 array_pop( $this->mArgStack );
2299
2300                 wfProfileOut( $fname );
2301                 return $text;
2302         }
2303
2304         /**
2305          * Replace magic variables
2306          * @access private
2307          */
2308         function variableSubstitution( $matches ) {
2309                 $fname = 'Parser::variableSubstitution';
2310                 $varname = $matches[1];
2311                 wfProfileIn( $fname );
2312                 if ( !$this->mVariables ) {
2313                         $this->initialiseVariables();
2314                 }
2315                 $skip = false;
2316                 if ( $this->mOutputType == OT_WIKI ) {
2317                         # Do only magic variables prefixed by SUBST
2318                         $mwSubst =& MagicWord::get( MAG_SUBST );
2319                         if (!$mwSubst->matchStartAndRemove( $varname ))
2320                                 $skip = true;
2321                         # Note that if we don't substitute the variable below,
2322                         # we don't remove the {{subst:}} magic word, in case
2323                         # it is a template rather than a magic variable.
2324                 }
2325                 if ( !$skip && array_key_exists( $varname, $this->mVariables ) ) {
2326                         $id = $this->mVariables[$varname];
2327                         $text = $this->getVariableValue( $id );
2328                         $this->mOutput->mContainsOldMagic = true;
2329                 } else {
2330                         $text = $matches[0];
2331                 }
2332                 wfProfileOut( $fname );
2333                 return $text;
2334         }
2335
2336         # Split template arguments
2337         function getTemplateArgs( $argsString ) {
2338                 if ( $argsString === '' ) {
2339                         return array();
2340                 }
2341
2342                 $args = explode( '|', substr( $argsString, 1 ) );
2343
2344                 # If any of the arguments contains a '[[' but no ']]', it needs to be
2345                 # merged with the next arg because the '|' character between belongs
2346                 # to the link syntax and not the template parameter syntax.
2347                 $argc = count($args);
2348
2349                 for ( $i = 0; $i < $argc-1; $i++ ) {
2350                         if ( substr_count ( $args[$i], '[[' ) != substr_count ( $args[$i], ']]' ) ) {
2351                                 $args[$i] .= '|'.$args[$i+1];
2352                                 array_splice($args, $i+1, 1);
2353                                 $i--;
2354                                 $argc--;
2355                         }
2356                 }
2357
2358                 return $args;
2359         }
2360
2361         /**
2362          * Return the text of a template, after recursively
2363          * replacing any variables or templates within the template.
2364          *
2365          * @param array $piece The parts of the template
2366          *  $piece['text']: matched text
2367          *  $piece['title']: the title, i.e. the part before the |
2368          *  $piece['parts']: the parameter array
2369          * @return string the text of the template
2370          * @access private
2371          */
2372         function braceSubstitution( $piece ) {
2373                 global $wgContLang;
2374                 $fname = 'Parser::braceSubstitution';
2375                 wfProfileIn( $fname );
2376
2377                 # Flags
2378                 $found = false;             # $text has been filled
2379                 $nowiki = false;            # wiki markup in $text should be escaped
2380                 $noparse = false;           # Unsafe HTML tags should not be stripped, etc.
2381                 $noargs = false;            # Don't replace triple-brace arguments in $text
2382                 $replaceHeadings = false;   # Make the edit section links go to the template not the article
2383                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2384                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2385
2386                 # Title object, where $text came from
2387                 $title = NULL;
2388
2389                 $linestart = '';
2390
2391                 # $part1 is the bit before the first |, and must contain only title characters
2392                 # $args is a list of arguments, starting from index 0, not including $part1
2393
2394                 $part1 = $piece['title'];
2395                 # If the third subpattern matched anything, it will start with |
2396
2397                 if (null == $piece['parts']) {
2398                         $replaceWith = $this->variableSubstitution (array ($piece['text'], $piece['title']));
2399                         if ($replaceWith != $piece['text']) {
2400                                 $text = $replaceWith;
2401                                 $found = true;
2402                                 $noparse = true;
2403                                 $noargs = true;
2404                         }
2405                 }
2406
2407                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2408                 $argc = count( $args );
2409
2410                 # SUBST
2411                 if ( !$found ) {
2412                         $mwSubst =& MagicWord::get( MAG_SUBST );
2413                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor ($this->mOutputType == OT_WIKI) ) {
2414                                 # One of two possibilities is true:
2415                                 # 1) Found SUBST but not in the PST phase
2416                                 # 2) Didn't find SUBST and in the PST phase
2417                                 # In either case, return without further processing
2418                                 $text = $piece['text'];
2419                                 $found = true;
2420                                 $noparse = true;
2421                                 $noargs = true;
2422                         }
2423                 }
2424
2425                 # MSG, MSGNW, INT and RAW
2426                 if ( !$found ) {
2427                         # Check for MSGNW:
2428                         $mwMsgnw =& MagicWord::get( MAG_MSGNW );
2429                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2430                                 $nowiki = true;
2431                         } else {
2432                                 # Remove obsolete MSG:
2433                                 $mwMsg =& MagicWord::get( MAG_MSG );
2434                                 $mwMsg->matchStartAndRemove( $part1 );
2435                         }
2436
2437                         # Check for RAW:
2438                         $mwRaw =& MagicWord::get( MAG_RAW );
2439                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2440                                 $forceRawInterwiki = true;
2441                         }
2442
2443                         # Check if it is an internal message
2444                         $mwInt =& MagicWord::get( MAG_INT );
2445                         if ( $mwInt->matchStartAndRemove( $part1 ) ) {
2446                                 if ( $this->incrementIncludeCount( 'int:'.$part1 ) ) {
2447                                         $text = $linestart . wfMsgReal( $part1, $args, true );
2448                                         $found = true;
2449                                 }
2450                         }
2451                 }
2452
2453                 # NS
2454                 if ( !$found ) {
2455                         # Check for NS: (namespace expansion)
2456                         $mwNs = MagicWord::get( MAG_NS );
2457                         if ( $mwNs->matchStartAndRemove( $part1 ) ) {
2458                                 if ( intval( $part1 ) || $part1 == "0" ) {
2459                                         $text = $linestart . $wgContLang->getNsText( intval( $part1 ) );
2460                                         $found = true;
2461                                 } else {
2462                                         $index = Namespace::getCanonicalIndex( strtolower( $part1 ) );
2463                                         if ( !is_null( $index ) ) {
2464                                                 $text = $linestart . $wgContLang->getNsText( $index );
2465                                                 $found = true;
2466                                         }
2467                                 }
2468                         }
2469                 }
2470
2471                 # LCFIRST, UCFIRST, LC and UC
2472                 if ( !$found ) {
2473                         $lcfirst =& MagicWord::get( MAG_LCFIRST );
2474                         $ucfirst =& MagicWord::get( MAG_UCFIRST );
2475                         $lc =& MagicWord::get( MAG_LC );
2476                         $uc =& MagicWord::get( MAG_UC );
2477                         if ( $lcfirst->matchStartAndRemove( $part1 ) ) {
2478                                 $text = $linestart . $wgContLang->lcfirst( $part1 );
2479                                 $found = true;
2480                         } else if ( $ucfirst->matchStartAndRemove( $part1 ) ) {
2481                                 $text = $linestart . $wgContLang->ucfirst( $part1 );
2482                                 $found = true;
2483                         } else if ( $lc->matchStartAndRemove( $part1 ) ) {
2484                                 $text = $linestart . $wgContLang->lc( $part1 );
2485                                 $found = true;
2486                         } else if ( $uc->matchStartAndRemove( $part1 ) ) {
2487                                  $text = $linestart . $wgContLang->uc( $part1 );
2488                                  $found = true;
2489                         }
2490                 }
2491
2492                 # LOCALURL and FULLURL
2493                 if ( !$found ) {
2494                         $mwLocal =& MagicWord::get( MAG_LOCALURL );
2495                         $mwLocalE =& MagicWord::get( MAG_LOCALURLE );
2496                         $mwFull =& MagicWord::get( MAG_FULLURL );
2497                         $mwFullE =& MagicWord::get( MAG_FULLURLE );
2498
2499
2500                         if ( $mwLocal->matchStartAndRemove( $part1 ) ) {
2501                                 $func = 'getLocalURL';
2502                         } elseif ( $mwLocalE->matchStartAndRemove( $part1 ) ) {
2503                                 $func = 'escapeLocalURL';
2504                         } elseif ( $mwFull->matchStartAndRemove( $part1 ) ) {
2505                                 $func = 'getFullURL';
2506                         } elseif ( $mwFullE->matchStartAndRemove( $part1 ) ) {
2507                                 $func = 'escapeFullURL';
2508                         } else {
2509                                 $func = false;
2510                         }
2511
2512                         if ( $func !== false ) {
2513                                 $title = Title::newFromText( $part1 );
2514                                 if ( !is_null( $title ) ) {
2515                                         if ( $argc > 0 ) {
2516                                                 $text = $linestart . $title->$func( $args[0] );
2517                                         } else {
2518                                                 $text = $linestart . $title->$func();
2519                                         }
2520                                         $found = true;
2521                                 }
2522                         }
2523                 }
2524
2525                 # GRAMMAR
2526                 if ( !$found && $argc == 1 ) {
2527                         $mwGrammar =& MagicWord::get( MAG_GRAMMAR );
2528                         if ( $mwGrammar->matchStartAndRemove( $part1 ) ) {
2529                                 $text = $linestart . $wgContLang->convertGrammar( $args[0], $part1 );
2530                                 $found = true;
2531                         }
2532                 }
2533
2534                 # PLURAL
2535                 if ( !$found && $argc >= 2 ) {
2536                         $mwPluralForm =& MagicWord::get( MAG_PLURAL );
2537                         if ( $mwPluralForm->matchStartAndRemove( $part1 ) ) {
2538                                 if ($argc==2) {$args[2]=$args[1];}
2539                                 $text = $linestart . $wgContLang->convertPlural( $part1, $args[0], $args[1], $args[2]);
2540                                 $found = true;
2541                         }
2542                 }
2543
2544                 # Template table test
2545
2546                 # Did we encounter this template already? If yes, it is in the cache
2547                 # and we need to check for loops.
2548                 if ( !$found && isset( $this->mTemplates[$piece['title']] ) ) {
2549                         $found = true;
2550
2551                         # Infinite loop test
2552                         if ( isset( $this->mTemplatePath[$part1] ) ) {
2553                                 $noparse = true;
2554                                 $noargs = true;
2555                                 $found = true;
2556                                 $text = $linestart .
2557                                         '{{' . $part1 . '}}' .
2558                                         '<!-- WARNING: template loop detected -->';
2559                                 wfDebug( "$fname: template loop broken at '$part1'\n" );
2560                         } else {
2561                                 # set $text to cached message.
2562                                 $text = $linestart . $this->mTemplates[$piece['title']];
2563                         }
2564                 }
2565
2566                 # Load from database
2567                 $lastPathLevel = $this->mTemplatePath;
2568                 if ( !$found ) {
2569                         $ns = NS_TEMPLATE;
2570                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage='' );
2571                         if ($subpage !== '') {
2572                                 $ns = $this->mTitle->getNamespace();
2573                         }
2574                         $title = Title::newFromText( $part1, $ns );
2575
2576                         if ( !is_null( $title ) ) {
2577                                 if ( !$title->isExternal() ) {
2578                                         # Check for excessive inclusion
2579                                         $dbk = $title->getPrefixedDBkey();
2580                                         if ( $this->incrementIncludeCount( $dbk ) ) {
2581                                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() ) {
2582                                                         # Capture special page output
2583                                                         $text = SpecialPage::capturePath( $title );
2584                                                         if ( is_string( $text ) ) {
2585                                                                 $found = true;
2586                                                                 $noparse = true;
2587                                                                 $noargs = true;
2588                                                                 $isHTML = true;
2589                                                                 $this->disableCache();
2590                                                         }
2591                                                 } else {
2592                                                         $articleContent = $this->fetchTemplate( $title );
2593                                                         if ( $articleContent !== false ) {
2594                                                                 $found = true;
2595                                                                 $text = $articleContent;
2596                                                                 $replaceHeadings = true;
2597                                                         }
2598                                                 }
2599                                         }
2600
2601                                         # If the title is valid but undisplayable, make a link to it
2602                                         if ( $this->mOutputType == OT_HTML && !$found ) {
2603                                                 $text = '[['.$title->getPrefixedText().']]';
2604                                                 $found = true;
2605                                         }
2606                                 } elseif ( $title->isTrans() ) {
2607                                         // Interwiki transclusion
2608                                         if ( $this->mOutputType == OT_HTML && !$forceRawInterwiki ) {
2609                                                 $text = $this->interwikiTransclude( $title, 'render' );
2610                                                 $isHTML = true;
2611                                                 $noparse = true;
2612                                         } else {
2613                                                 $text = $this->interwikiTransclude( $title, 'raw' );
2614                                                 $replaceHeadings = true;
2615                                         }
2616                                         $found = true;
2617                                 }
2618
2619                                 # Template cache array insertion
2620                                 # Use the original $piece['title'] not the mangled $part1, so that
2621                                 # modifiers such as RAW: produce separate cache entries
2622                                 if( $found ) {
2623                                         $this->mTemplates[$piece['title']] = $text;
2624                                         $text = $linestart . $text;
2625                                 }
2626                         }
2627                 }
2628
2629                 # Recursive parsing, escaping and link table handling
2630                 # Only for HTML output
2631                 if ( $nowiki && $found && $this->mOutputType == OT_HTML ) {
2632                         $text = wfEscapeWikiText( $text );
2633                 } elseif ( ($this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI) && $found ) {
2634                         if ( !$noargs ) {
2635                                 # Clean up argument array
2636                                 $assocArgs = array();
2637                                 $index = 1;
2638                                 foreach( $args as $arg ) {
2639                                         $eqpos = strpos( $arg, '=' );
2640                                         if ( $eqpos === false ) {
2641                                                 $assocArgs[$index++] = $arg;
2642                                         } else {
2643                                                 $name = trim( substr( $arg, 0, $eqpos ) );
2644                                                 $value = trim( substr( $arg, $eqpos+1 ) );
2645                                                 if ( $value === false ) {
2646                                                         $value = '';
2647                                                 }
2648                                                 if ( $name !== false ) {
2649                                                         $assocArgs[$name] = $value;
2650                                                 }
2651                                         }
2652                                 }
2653
2654                                 # Add a new element to the templace recursion path
2655                                 $this->mTemplatePath[$part1] = 1;
2656                         }
2657
2658                         if ( !$noparse ) {
2659                                 # If there are any <onlyinclude> tags, only include them
2660                                 if ( in_string( '<onlyinclude>', $text ) && in_string( '</onlyinclude>', $text ) ) {
2661                                         preg_match_all( '/<onlyinclude>(.*?)\n?<\/onlyinclude>/s', $text, $m );
2662                                         $text = '';
2663                                         foreach ($m[1] as $piece)
2664                                                 $text .= $piece;
2665                                 }
2666                                 # Remove <noinclude> sections and <includeonly> tags
2667                                 $text = preg_replace( '/<noinclude>.*?<\/noinclude>/s', '', $text );
2668                                 $text = strtr( $text, array( '<includeonly>' => '' , '</includeonly>' => '' ) );
2669
2670                                 if( $this->mOutputType == OT_HTML ) {
2671                                         # Strip <nowiki>, <pre>, etc.
2672                                         $text = $this->strip( $text, $this->mStripState );
2673                                         $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'replaceVariables' ), $assocArgs );
2674                                 }
2675                                 $text = $this->replaceVariables( $text, $assocArgs );
2676
2677                                 # If the template begins with a table or block-level
2678                                 # element, it should be treated as beginning a new line.
2679                                 if (!$piece['lineStart'] && preg_match('/^({\\||:|;|#|\*)/', $text)) {
2680                                         $text = "\n" . $text;
2681                                 }
2682                         } elseif ( !$noargs ) {
2683                                 # $noparse and !$noargs
2684                                 # Just replace the arguments, not any double-brace items
2685                                 # This is used for rendered interwiki transclusion
2686                                 $text = $this->replaceVariables( $text, $assocArgs, true );
2687                         }
2688                 }
2689                 # Prune lower levels off the recursion check path
2690                 $this->mTemplatePath = $lastPathLevel;
2691
2692                 if ( !$found ) {
2693                         wfProfileOut( $fname );
2694                         return $piece['text'];
2695                 } else {
2696                         if ( $isHTML ) {
2697                                 # Replace raw HTML by a placeholder
2698                                 # Add a blank line preceding, to prevent it from mucking up
2699                                 # immediately preceding headings
2700                                 $text = "\n\n" . $this->insertStripItem( $text, $this->mStripState );
2701                         } else {
2702                                 # replace ==section headers==
2703                                 # XXX this needs to go away once we have a better parser.
2704                                 if ( $this->mOutputType != OT_WIKI && $replaceHeadings ) {
2705                                         if( !is_null( $title ) )
2706                                                 $encodedname = base64_encode($title->getPrefixedDBkey());
2707                                         else
2708                                                 $encodedname = base64_encode("");
2709                                         $m = preg_split('/(^={1,6}.*?={1,6}\s*?$)/m', $text, -1,
2710                                                 PREG_SPLIT_DELIM_CAPTURE);
2711                                         $text = '';
2712                                         $nsec = 0;
2713                                         for( $i = 0; $i < count($m); $i += 2 ) {
2714                                                 $text .= $m[$i];
2715                                                 if (!isset($m[$i + 1]) || $m[$i + 1] == "") continue;
2716                                                 $hl = $m[$i + 1];
2717                                                 if( strstr($hl, "<!--MWTEMPLATESECTION") ) {
2718                                                         $text .= $hl;
2719                                                         continue;
2720                                                 }
2721                                                 preg_match('/^(={1,6})(.*?)(={1,6})\s*?$/m', $hl, $m2);
2722                                                 $text .= $m2[1] . $m2[2] . "<!--MWTEMPLATESECTION="
2723                                                         . $encodedname . "&" . base64_encode("$nsec") . "-->" . $m2[3];
2724
2725                                                 $nsec++;
2726                                         }
2727                                 }
2728                         }
2729                 }
2730
2731                 # Prune lower levels off the recursion check path
2732                 $this->mTemplatePath = $lastPathLevel;
2733
2734                 if ( !$found ) {
2735                         wfProfileOut( $fname );
2736                         return $piece['text'];
2737                 } else {
2738                         wfProfileOut( $fname );
2739                         return $text;
2740                 }
2741         }
2742
2743         /**
2744          * Fetch the unparsed text of a template and register a reference to it.
2745          */
2746         function fetchTemplate( $title ) {
2747                 $text = false;
2748                 // Loop to fetch the article, with up to 1 redirect
2749                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
2750                         $rev = Revision::newFromTitle( $title );
2751                         $this->mOutput->addTemplate( $title, $title->getArticleID() );
2752                         if ( !$rev ) {
2753                                 break;
2754                         }
2755                         $text = $rev->getText();
2756                         if ( $text === false ) {
2757                                 break;
2758                         }
2759                         // Redirect?
2760                         $title = Title::newFromRedirect( $text );
2761                 }
2762                 return $text;
2763         }
2764
2765         /**
2766          * Transclude an interwiki link.
2767          */
2768         function interwikiTransclude( $title, $action ) {
2769                 global $wgEnableScaryTranscluding, $wgCanonicalNamespaceNames;
2770
2771                 if (!$wgEnableScaryTranscluding)
2772                         return wfMsg('scarytranscludedisabled');
2773
2774                 // The namespace will actually only be 0 or 10, depending on whether there was a leading :
2775                 // But we'll handle it generally anyway
2776                 if ( $title->getNamespace() ) {
2777                         // Use the canonical namespace, which should work anywhere
2778                         $articleName = $wgCanonicalNamespaceNames[$title->getNamespace()] . ':' . $title->getDBkey();
2779                 } else {
2780                         $articleName = $title->getDBkey();
2781                 }
2782
2783                 $url = str_replace('$1', urlencode($articleName), Title::getInterwikiLink($title->getInterwiki()));
2784                 $url .= "?action=$action";
2785                 if (strlen($url) > 255)
2786                         return wfMsg('scarytranscludetoolong');
2787                 return $this->fetchScaryTemplateMaybeFromCache($url);
2788         }
2789
2790         function fetchScaryTemplateMaybeFromCache($url) {
2791                 global $wgTranscludeCacheExpiry;
2792                 $dbr =& wfGetDB(DB_SLAVE);
2793                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
2794                                 array('tc_url' => $url));
2795                 if ($obj) {
2796                         $time = $obj->tc_time;
2797                         $text = $obj->tc_contents;
2798                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
2799                                 return $text;
2800                         }
2801                 }
2802
2803                 $text = wfGetHTTP($url);
2804                 if (!$text)
2805                         return wfMsg('scarytranscludefailed', $url);
2806
2807                 $dbw =& wfGetDB(DB_MASTER);
2808                 $dbw->replace('transcache', array('tc_url'), array(
2809                         'tc_url' => $url,
2810                         'tc_time' => time(),
2811                         'tc_contents' => $text));
2812                 return $text;
2813         }
2814
2815
2816         /**
2817          * Triple brace replacement -- used for template arguments
2818          * @access private
2819          */
2820         function argSubstitution( $matches ) {
2821                 $arg = trim( $matches['title'] );
2822                 $text = $matches['text'];
2823                 $inputArgs = end( $this->mArgStack );
2824
2825                 if ( array_key_exists( $arg, $inputArgs ) ) {
2826                         $text = $inputArgs[$arg];
2827                 } else if ($this->mOutputType == OT_HTML && null != $matches['parts'] && count($matches['parts']) > 0) {
2828                         $text = $matches['parts'][0];
2829                 }
2830
2831                 return $text;
2832         }
2833
2834         /**
2835          * Returns true if the function is allowed to include this entity
2836          * @access private
2837          */
2838         function incrementIncludeCount( $dbk ) {
2839                 if ( !array_key_exists( $dbk, $this->mIncludeCount ) ) {
2840                         $this->mIncludeCount[$dbk] = 0;
2841                 }
2842                 if ( ++$this->mIncludeCount[$dbk] <= MAX_INCLUDE_REPEAT ) {
2843                         return true;
2844                 } else {
2845                         return false;
2846                 }
2847         }
2848
2849         /**
2850          * This function accomplishes several tasks:
2851          * 1) Auto-number headings if that option is enabled
2852          * 2) Add an [edit] link to sections for logged in users who have enabled the option
2853          * 3) Add a Table of contents on the top for users who have enabled the option
2854          * 4) Auto-anchor headings
2855          *
2856          * It loops through all headlines, collects the necessary data, then splits up the
2857          * string and re-inserts the newly formatted headlines.
2858          *
2859          * @param string $text
2860          * @param boolean $isMain
2861          * @access private
2862          */
2863         function formatHeadings( $text, $isMain=true ) {
2864                 global $wgMaxTocLevel, $wgContLang;
2865
2866                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
2867                 $doShowToc = true;
2868                 $forceTocHere = false;
2869                 if( !$this->mTitle->userCanEdit() ) {
2870                         $showEditLink = 0;
2871                 } else {
2872                         $showEditLink = $this->mOptions->getEditSection();
2873                 }
2874
2875                 # Inhibit editsection links if requested in the page
2876                 $esw =& MagicWord::get( MAG_NOEDITSECTION );
2877                 if( $esw->matchAndRemove( $text ) ) {
2878                         $showEditLink = 0;
2879                 }
2880                 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
2881                 # do not add TOC
2882                 $mw =& MagicWord::get( MAG_NOTOC );
2883                 if( $mw->matchAndRemove( $text ) ) {
2884                         $doShowToc = false;
2885                 }
2886
2887                 # Get all headlines for numbering them and adding funky stuff like [edit]
2888                 # links - this is for later, but we need the number of headlines right now
2889                 $numMatches = preg_match_all( '/<H([1-6])(.*?'.'>)(.*?)<\/H[1-6] *>/i', $text, $matches );
2890
2891                 # if there are fewer than 4 headlines in the article, do not show TOC
2892                 if( $numMatches < 4 ) {
2893                         $doShowToc = false;
2894                 }
2895
2896                 # if the string __TOC__ (not case-sensitive) occurs in the HTML,
2897                 # override above conditions and always show TOC at that place
2898
2899                 $mw =& MagicWord::get( MAG_TOC );
2900                 if($mw->match( $text ) ) {
2901                         $doShowToc = true;
2902                         $forceTocHere = true;
2903                 } else {
2904                         # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
2905                         # override above conditions and always show TOC above first header
2906                         $mw =& MagicWord::get( MAG_FORCETOC );
2907                         if ($mw->matchAndRemove( $text ) ) {
2908                                 $doShowToc = true;
2909                         }
2910                 }
2911
2912                 # Never ever show TOC if no headers
2913                 if( $numMatches < 1 ) {
2914                         $doShowToc = false;
2915                 }
2916
2917                 # We need this to perform operations on the HTML
2918                 $sk =& $this->mOptions->getSkin();
2919
2920                 # headline counter
2921                 $headlineCount = 0;
2922                 $sectionCount = 0; # headlineCount excluding template sections
2923
2924                 # Ugh .. the TOC should have neat indentation levels which can be
2925                 # passed to the skin functions. These are determined here
2926                 $toc = '';
2927                 $full = '';
2928                 $head = array();
2929                 $sublevelCount = array();
2930                 $levelCount = array();
2931                 $toclevel = 0;
2932                 $level = 0;
2933                 $prevlevel = 0;
2934                 $toclevel = 0;
2935                 $prevtoclevel = 0;
2936
2937                 foreach( $matches[3] as $headline ) {
2938                         $istemplate = 0;
2939                         $templatetitle = '';
2940                         $templatesection = 0;
2941                         $numbering = '';
2942
2943                         if (preg_match("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", $headline, $mat)) {
2944                                 $istemplate = 1;
2945                                 $templatetitle = base64_decode($mat[1]);
2946                                 $templatesection = 1 + (int)base64_decode($mat[2]);
2947                                 $headline = preg_replace("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", "", $headline);
2948                         }
2949
2950                         if( $toclevel ) {
2951                                 $prevlevel = $level;
2952                                 $prevtoclevel = $toclevel;
2953                         }
2954                         $level = $matches[1][$headlineCount];
2955
2956                         if( $doNumberHeadings || $doShowToc ) {
2957
2958                                 if ( $level > $prevlevel ) {
2959                                         # Increase TOC level
2960                                         $toclevel++;
2961                                         $sublevelCount[$toclevel] = 0;
2962                                         $toc .= $sk->tocIndent();
2963                                 }
2964                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
2965                                         # Decrease TOC level, find level to jump to
2966
2967                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
2968                                                 # Can only go down to level 1
2969                                                 $toclevel = 1;
2970                                         } else {
2971                                                 for ($i = $toclevel; $i > 0; $i--) {
2972                                                         if ( $levelCount[$i] == $level ) {
2973                                                                 # Found last matching level
2974                                                                 $toclevel = $i;
2975                                                                 break;
2976                                                         }
2977                                                         elseif ( $levelCount[$i] < $level ) {
2978                                                                 # Found first matching level below current level
2979                                                                 $toclevel = $i + 1;
2980                                                                 break;
2981                                                         }
2982                                                 }
2983                                         }
2984
2985                                         $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
2986                                 }
2987                                 else {
2988                                         # No change in level, end TOC line
2989                                         $toc .= $sk->tocLineEnd();
2990                                 }
2991
2992                                 $levelCount[$toclevel] = $level;
2993
2994                                 # count number of headlines for each level
2995                                 @$sublevelCount[$toclevel]++;
2996                                 $dot = 0;
2997                                 for( $i = 1; $i <= $toclevel; $i++ ) {
2998                                         if( !empty( $sublevelCount[$i] ) ) {
2999                                                 if( $dot ) {
3000                                                         $numbering .= '.';
3001                                                 }
3002                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3003                                                 $dot = 1;
3004                                         }
3005                                 }
3006                         }
3007
3008                         # The canonized header is a version of the header text safe to use for links
3009                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3010                         $canonized_headline = $this->unstrip( $headline, $this->mStripState );
3011                         $canonized_headline = $this->unstripNoWiki( $canonized_headline, $this->mStripState );
3012
3013                         # Remove link placeholders by the link text.
3014                         #     <!--LINK number-->
3015                         # turns into
3016                         #     link text with suffix
3017                         $canonized_headline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
3018                                                             "\$this->mLinkHolders['texts'][\$1]",
3019                                                             $canonized_headline );
3020                         $canonized_headline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
3021                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
3022                                                             $canonized_headline );
3023
3024                         # strip out HTML
3025                         $canonized_headline = preg_replace( '/<.*?' . '>/','',$canonized_headline );
3026                         $tocline = trim( $canonized_headline );
3027                         $canonized_headline = Sanitizer::escapeId( $tocline );
3028                         $refers[$headlineCount] = $canonized_headline;
3029
3030                         # count how many in assoc. array so we can track dupes in anchors
3031                         @$refers[$canonized_headline]++;
3032                         $refcount[$headlineCount]=$refers[$canonized_headline];
3033
3034                         # Don't number the heading if it is the only one (looks silly)
3035                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3036                                 # the two are different if the line contains a link
3037                                 $headline=$numbering . ' ' . $headline;
3038                         }
3039
3040                         # Create the anchor for linking from the TOC to the section
3041                         $anchor = $canonized_headline;
3042                         if($refcount[$headlineCount] > 1 ) {
3043                                 $anchor .= '_' . $refcount[$headlineCount];
3044                         }
3045                         if( $doShowToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3046                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3047                         }
3048                         if( $showEditLink && ( !$istemplate || $templatetitle !== "" ) ) {
3049                                 if ( empty( $head[$headlineCount] ) ) {
3050                                         $head[$headlineCount] = '';
3051                                 }
3052                                 if( $istemplate )
3053                                         $head[$headlineCount] .= $sk->editSectionLinkForOther($templatetitle, $templatesection);
3054                                 else
3055                                         $head[$headlineCount] .= $sk->editSectionLink($this->mTitle, $sectionCount+1);
3056                         }
3057
3058                         # give headline the correct <h#> tag
3059                         @$head[$headlineCount] .= "<a name=\"$anchor\"></a><h".$level.$matches[2][$headlineCount] .$headline.'</h'.$level.'>';
3060
3061                         $headlineCount++;
3062                         if( !$istemplate )
3063                                 $sectionCount++;
3064                 }
3065
3066                 if( $doShowToc ) {
3067                         $toc .= $sk->tocUnindent( $toclevel - 1 );
3068                         $toc = $sk->tocList( $toc );
3069                 }
3070
3071                 # split up and insert constructed headlines
3072
3073                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3074                 $i = 0;
3075
3076                 foreach( $blocks as $block ) {
3077                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
3078                                 # This is the [edit] link that appears for the top block of text when
3079                                 # section editing is enabled
3080
3081                                 # Disabled because it broke block formatting
3082                                 # For example, a bullet point in the top line
3083                                 # $full .= $sk->editSectionLink(0);
3084                         }
3085                         $full .= $block;
3086                         if( $doShowToc && !$i && $isMain && !$forceTocHere) {
3087                         # Top anchor now in skin
3088                                 $full = $full.$toc;
3089                         }
3090
3091                         if( !empty( $head[$i] ) ) {
3092                                 $full .= $head[$i];
3093                         }
3094                         $i++;
3095                 }
3096                 if($forceTocHere) {
3097                         $mw =& MagicWord::get( MAG_TOC );
3098                         return $mw->replace( $toc, $full );
3099                 } else {
3100                         return $full;
3101                 }
3102         }
3103
3104         /**
3105          * Return an HTML link for the "ISBN 123456" text
3106          * @access private
3107          */
3108         function magicISBN( $text ) {
3109                 $fname = 'Parser::magicISBN';
3110                 wfProfileIn( $fname );
3111
3112                 $a = split( 'ISBN ', ' '.$text );
3113                 if ( count ( $a ) < 2 ) {
3114                         wfProfileOut( $fname );
3115                         return $text;
3116                 }
3117                 $text = substr( array_shift( $a ), 1);
3118                 $valid = '0123456789-Xx';
3119
3120                 foreach ( $a as $x ) {
3121                         $isbn = $blank = '' ;
3122                         while ( ' ' == $x{0} ) {
3123                                 $blank .= ' ';
3124                                 $x = substr( $x, 1 );
3125                         }
3126                         if ( $x == '' ) { # blank isbn
3127                                 $text .= "ISBN $blank";
3128                                 continue;
3129                         }
3130                         while ( strstr( $valid, $x{0} ) != false ) {
3131                                 $isbn .= $x{0};
3132                                 $x = substr( $x, 1 );
3133                         }
3134                         $num = str_replace( '-', '', $isbn );
3135                         $num = str_replace( ' ', '', $num );
3136                         $num = str_replace( 'x', 'X', $num );
3137
3138                         if ( '' == $num ) {
3139                                 $text .= "ISBN $blank$x";
3140                         } else {
3141                                 $titleObj = Title::makeTitle( NS_SPECIAL, 'Booksources' );
3142                                 $text .= '<a href="' .
3143                                 $titleObj->escapeLocalUrl( 'isbn='.$num ) .
3144                                         "\" class=\"internal\">ISBN $isbn</a>";
3145                                 $text .= $x;
3146                         }
3147                 }
3148                 wfProfileOut( $fname );
3149                 return $text;
3150         }
3151
3152         /**
3153          * Return an HTML link for the "RFC 1234" text
3154          *
3155          * @access private
3156          * @param string $text     Text to be processed
3157          * @param string $keyword  Magic keyword to use (default RFC)
3158          * @param string $urlmsg   Interface message to use (default rfcurl)
3159          * @return string
3160          */
3161         function magicRFC( $text, $keyword='RFC ', $urlmsg='rfcurl'  ) {
3162
3163                 $valid = '0123456789';
3164                 $internal = false;
3165
3166                 $a = split( $keyword, ' '.$text );
3167                 if ( count ( $a ) < 2 ) {
3168                         return $text;
3169                 }
3170                 $text = substr( array_shift( $a ), 1);
3171
3172                 /* Check if keyword is preceed by [[.
3173                  * This test is made here cause of the array_shift above
3174                  * that prevent the test to be done in the foreach.
3175                  */
3176                 if ( substr( $text, -2 ) == '[[' ) {
3177                         $internal = true;
3178                 }
3179
3180                 foreach ( $a as $x ) {
3181                         /* token might be empty if we have RFC RFC 1234 */
3182                         if ( $x=='' ) {
3183                                 $text.=$keyword;
3184                                 continue;
3185                                 }
3186
3187                         $id = $blank = '' ;
3188
3189                         /** remove and save whitespaces in $blank */
3190                         while ( $x{0} == ' ' ) {
3191                                 $blank .= ' ';
3192                                 $x = substr( $x, 1 );
3193                         }
3194
3195                         /** remove and save the rfc number in $id */
3196                         while ( strstr( $valid, $x{0} ) != false ) {
3197                                 $id .= $x{0};
3198                                 $x = substr( $x, 1 );
3199                         }
3200
3201                         if ( $id == '' ) {
3202                                 /* call back stripped spaces*/
3203                                 $text .= $keyword.$blank.$x;
3204                         } elseif( $internal ) {
3205                                 /* normal link */
3206                                 $text .= $keyword.$id.$x;
3207                         } else {
3208                                 /* build the external link*/
3209                                 $url = wfMsg( $urlmsg, $id);
3210                                 $sk =& $this->mOptions->getSkin();
3211                                 $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
3212                                 $text .= "<a href='{$url}'{$la}>{$keyword}{$id}</a>{$x}";
3213                         }
3214
3215                         /* Check if the next RFC keyword is preceed by [[ */
3216                         $internal = ( substr($x,-2) == '[[' );
3217                 }
3218                 return $text;
3219         }
3220
3221         /**
3222          * Transform wiki markup when saving a page by doing \r\n -> \n
3223          * conversion, substitting signatures, {{subst:}} templates, etc.
3224          *
3225          * @param string $text the text to transform
3226          * @param Title &$title the Title object for the current article
3227          * @param User &$user the User object describing the current user
3228          * @param ParserOptions $options parsing options
3229          * @param bool $clearState whether to clear the parser state first
3230          * @return string the altered wiki markup
3231          * @access public
3232          */
3233         function preSaveTransform( $text, &$title, &$user, $options, $clearState = true ) {
3234                 $this->mOptions = $options;
3235                 $this->mTitle =& $title;
3236                 $this->mOutputType = OT_WIKI;
3237
3238                 if ( $clearState ) {
3239                         $this->clearState();
3240                 }
3241
3242                 $stripState = false;
3243                 $pairs = array(
3244                         "\r\n" => "\n",
3245                 );
3246                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3247                 $text = $this->strip( $text, $stripState, true );
3248                 $text = $this->pstPass2( $text, $user );
3249                 $text = $this->unstrip( $text, $stripState );
3250                 $text = $this->unstripNoWiki( $text, $stripState );
3251                 return $text;
3252         }
3253
3254         /**
3255          * Pre-save transform helper function
3256          * @access private
3257          */
3258         function pstPass2( $text, &$user ) {
3259                 global $wgContLang, $wgLocaltimezone;
3260
3261                 /* Note: This is the timestamp saved as hardcoded wikitext to
3262                  * the database, we use $wgContLang here in order to give
3263                  * everyone the same signiture and use the default one rather
3264                  * than the one selected in each users preferences.
3265                  */
3266                 if ( isset( $wgLocaltimezone ) ) {
3267                         $oldtz = getenv( 'TZ' );
3268                         putenv( 'TZ='.$wgLocaltimezone );
3269                 }
3270                 $d = $wgContLang->timeanddate( date( 'YmdHis' ), false, false) .
3271                   ' (' . date( 'T' ) . ')';
3272                 if ( isset( $wgLocaltimezone ) ) {
3273                         putenv( 'TZ='.$oldtz );
3274                 }
3275
3276                 # Variable replacement
3277                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3278                 $text = $this->replaceVariables( $text );
3279
3280                 # Signatures
3281                 $sigText = $this->getUserSig( $user );
3282                 $text = strtr( $text, array(
3283                         '~~~~~' => $d,
3284                         '~~~~' => "$sigText $d",
3285                         '~~~' => $sigText
3286                 ) );
3287
3288                 # Context links: [[|name]] and [[name (context)|]]
3289                 #
3290                 global $wgLegalTitleChars;
3291                 $tc = "[$wgLegalTitleChars]";
3292                 $np = str_replace( array( '(', ')' ), array( '', '' ), $tc ); # No parens
3293
3294                 $namespacechar = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
3295                 $conpat = "/^({$np}+) \\(({$tc}+)\\)$/";
3296
3297                 $p1 = "/\[\[({$np}+) \\(({$np}+)\\)\\|]]/";             # [[page (context)|]]
3298                 $p2 = "/\[\[\\|({$tc}+)]]/";                                    # [[|page]]
3299                 $p3 = "/\[\[(:*$namespacechar+):({$np}+)\\|]]/";                # [[namespace:page|]] and [[:namespace:page|]]
3300                 $p4 = "/\[\[(:*$namespacechar+):({$np}+) \\(({$np}+)\\)\\|]]/"; # [[ns:page (cont)|]] and [[:ns:page (cont)|]]
3301                 $context = '';
3302                 $t = $this->mTitle->getText();
3303                 if ( preg_match( $conpat, $t, $m ) ) {
3304                         $context = $m[2];
3305                 }
3306                 $text = preg_replace( $p4, '[[\\1:\\2 (\\3)|\\2]]', $text );
3307                 $text = preg_replace( $p1, '[[\\1 (\\2)|\\1]]', $text );
3308                 $text = preg_replace( $p3, '[[\\1:\\2|\\2]]', $text );
3309
3310                 if ( '' == $context ) {
3311                         $text = preg_replace( $p2, '[[\\1]]', $text );
3312                 } else {
3313                         $text = preg_replace( $p2, "[[\\1 ({$context})|\\1]]", $text );
3314                 }
3315
3316                 # Trim trailing whitespace
3317                 # MAG_END (__END__) tag allows for trailing
3318                 # whitespace to be deliberately included
3319                 $text = rtrim( $text );
3320                 $mw =& MagicWord::get( MAG_END );
3321                 $mw->matchAndRemove( $text );
3322
3323                 return $text;
3324         }
3325
3326         /**
3327          * Fetch the user's signature text, if any, and normalize to
3328          * validated, ready-to-insert wikitext.
3329          *
3330          * @param User $user
3331          * @return string
3332          * @access private
3333          */
3334         function getUserSig( &$user ) {
3335                 global $wgContLang;
3336
3337                 $username = $user->getName();
3338                 $nickname = $user->getOption( 'nickname' );
3339                 $nickname = $nickname === '' ? $username : $nickname;
3340
3341                 if( $user->getBoolOption( 'fancysig' ) !== false ) {
3342                         # Sig. might contain markup; validate this
3343                         if( $this->validateSig( $nickname ) !== false ) {
3344                                 # Validated; clean up (if needed) and return it
3345                                 return( $this->cleanSig( $nickname ) );
3346                         } else {
3347                                 # Failed to validate; fall back to the default
3348                                 $nickname = $username;
3349                                 wfDebug( "Parser::getUserSig: $username has bad XML tags in signature.\n" );
3350                         }
3351                 }
3352
3353                 # If we're still here, make it a link to the user page
3354                 $userpage = $user->getUserPage();
3355                 return( '[[' . $userpage->getPrefixedText() . '|' . wfEscapeWikiText( $nickname ) . ']]' );
3356         }
3357
3358         /**
3359          * Check that the user's signature contains no bad XML
3360          *
3361          * @param string $text
3362          * @return mixed An expanded string, or false if invalid.
3363          */
3364         function validateSig( $text ) {
3365                 return( wfIsWellFormedXmlFragment( $text ) ? $text : false );
3366         }
3367
3368         /**
3369          * Clean up signature text
3370          *
3371          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures
3372          * 2) Substitute all transclusions
3373          *
3374          * @param string $text
3375          * @return string Signature text
3376          */
3377         function cleanSig( $text ) {
3378                 $substWord = MagicWord::get( MAG_SUBST );
3379                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3380                 $substText = '{{' . $substWord->getSynonym( 0 );
3381
3382                 $text = preg_replace( $substRegex, $substText, $text );
3383                 $text = preg_replace( '/~{3,5}/', '', $text );
3384                 $text = $this->replaceVariables( $text );
3385
3386                 return $text;
3387         }
3388
3389         /**
3390          * Set up some variables which are usually set up in parse()
3391          * so that an external function can call some class members with confidence
3392          * @access public
3393          */
3394         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
3395                 $this->mTitle =& $title;
3396                 $this->mOptions = $options;
3397                 $this->mOutputType = $outputType;
3398                 if ( $clearState ) {
3399                         $this->clearState();
3400                 }
3401         }
3402
3403         /**
3404          * Transform a MediaWiki message by replacing magic variables.
3405          *
3406          * @param string $text the text to transform
3407          * @param ParserOptions $options  options
3408          * @return string the text with variables substituted
3409          * @access public
3410          */
3411         function transformMsg( $text, $options ) {
3412                 global $wgTitle;
3413                 static $executing = false;
3414
3415                 $fname = "Parser::transformMsg";
3416
3417                 # Guard against infinite recursion
3418                 if ( $executing ) {
3419                         return $text;
3420                 }
3421                 $executing = true;
3422
3423                 wfProfileIn($fname);
3424
3425                 $this->mTitle = $wgTitle;
3426                 $this->mOptions = $options;
3427                 $this->mOutputType = OT_MSG;
3428                 $this->clearState();
3429                 $text = $this->replaceVariables( $text );
3430
3431                 $executing = false;
3432                 wfProfileOut($fname);
3433                 return $text;
3434         }
3435
3436         /**
3437          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
3438          * Callback will be called with the text within
3439          * Transform and return the text within
3440          *
3441          * @access public
3442          *
3443          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
3444          * @param mixed $callback The callback function (and object) to use for the tag
3445          *
3446          * @return The old value of the mTagHooks array associated with the hook
3447          */
3448         function setHook( $tag, $callback ) {
3449                 $oldVal = @$this->mTagHooks[$tag];
3450                 $this->mTagHooks[$tag] = $callback;
3451
3452                 return $oldVal;
3453         }
3454
3455         /**
3456          * Replace <!--LINK--> link placeholders with actual links, in the buffer
3457          * Placeholders created in Skin::makeLinkObj()
3458          * Returns an array of links found, indexed by PDBK:
3459          *  0 - broken
3460          *  1 - normal link
3461          *  2 - stub
3462          * $options is a bit field, RLH_FOR_UPDATE to select for update
3463          */
3464         function replaceLinkHolders( &$text, $options = 0 ) {
3465                 global $wgUser;
3466                 global $wgOutputReplace;
3467
3468                 $fname = 'Parser::replaceLinkHolders';
3469                 wfProfileIn( $fname );
3470
3471                 $pdbks = array();
3472                 $colours = array();
3473                 $sk =& $this->mOptions->getSkin();
3474                 $linkCache =& LinkCache::singleton();
3475
3476                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
3477                         wfProfileIn( $fname.'-check' );
3478                         $dbr =& wfGetDB( DB_SLAVE );
3479                         $page = $dbr->tableName( 'page' );
3480                         $threshold = $wgUser->getOption('stubthreshold');
3481
3482                         # Sort by namespace
3483                         asort( $this->mLinkHolders['namespaces'] );
3484
3485                         # Generate query
3486                         $query = false;
3487                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3488                                 # Make title object
3489                                 $title = $this->mLinkHolders['titles'][$key];
3490
3491                                 # Skip invalid entries.
3492                                 # Result will be ugly, but prevents crash.
3493                                 if ( is_null( $title ) ) {
3494                                         continue;
3495                                 }
3496                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
3497
3498                                 # Check if it's a static known link, e.g. interwiki
3499                                 if ( $title->isAlwaysKnown() ) {
3500                                         $colours[$pdbk] = 1;
3501                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
3502                                         $colours[$pdbk] = 1;
3503                                         $this->mOutput->addLink( $title, $id );
3504                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
3505                                         $colours[$pdbk] = 0;
3506                                 } else {
3507                                         # Not in the link cache, add it to the query
3508                                         if ( !isset( $current ) ) {
3509                                                 $current = $ns;
3510                                                 $query =  "SELECT page_id, page_namespace, page_title";
3511                                                 if ( $threshold > 0 ) {
3512                                                         $query .= ', page_len, page_is_redirect';
3513                                                 }
3514                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
3515                                         } elseif ( $current != $ns ) {
3516                                                 $current = $ns;
3517                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
3518                                         } else {
3519                                                 $query .= ', ';
3520                                         }
3521
3522                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
3523                                 }
3524                         }
3525                         if ( $query ) {
3526                                 $query .= '))';
3527                                 if ( $options & RLH_FOR_UPDATE ) {
3528                                         $query .= ' FOR UPDATE';
3529                                 }
3530
3531                                 $res = $dbr->query( $query, $fname );
3532
3533                                 # Fetch data and form into an associative array
3534                                 # non-existent = broken
3535                                 # 1 = known
3536                                 # 2 = stub
3537                                 while ( $s = $dbr->fetchObject($res) ) {
3538                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
3539                                         $pdbk = $title->getPrefixedDBkey();
3540                                         $linkCache->addGoodLinkObj( $s->page_id, $title );
3541                                         $this->mOutput->addLink( $title, $s->page_id );
3542
3543                                         if ( $threshold >  0 ) {
3544                                                 $size = $s->page_len;
3545                                                 if ( $s->page_is_redirect || $s->page_namespace != 0 || $size >= $threshold ) {
3546                                                         $colours[$pdbk] = 1;
3547                                                 } else {
3548                                                         $colours[$pdbk] = 2;
3549                                                 }
3550                                         } else {
3551                                                 $colours[$pdbk] = 1;
3552                                         }
3553                                 }
3554                         }
3555                         wfProfileOut( $fname.'-check' );
3556
3557                         # Construct search and replace arrays
3558                         wfProfileIn( $fname.'-construct' );
3559                         $wgOutputReplace = array();
3560                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3561                                 $pdbk = $pdbks[$key];
3562                                 $searchkey = "<!--LINK $key-->";
3563                                 $title = $this->mLinkHolders['titles'][$key];
3564                                 if ( empty( $colours[$pdbk] ) ) {
3565                                         $linkCache->addBadLinkObj( $title );
3566                                         $colours[$pdbk] = 0;
3567                                         $this->mOutput->addLink( $title, 0 );
3568                                         $wgOutputReplace[$searchkey] = $sk->makeBrokenLinkObj( $title,
3569                                                                         $this->mLinkHolders['texts'][$key],
3570                                                                         $this->mLinkHolders['queries'][$key] );
3571                                 } elseif ( $colours[$pdbk] == 1 ) {
3572                                         $wgOutputReplace[$searchkey] = $sk->makeKnownLinkObj( $title,
3573                                                                         $this->mLinkHolders['texts'][$key],
3574                                                                         $this->mLinkHolders['queries'][$key] );
3575                                 } elseif ( $colours[$pdbk] == 2 ) {
3576                                         $wgOutputReplace[$searchkey] = $sk->makeStubLinkObj( $title,
3577                                                                         $this->mLinkHolders['texts'][$key],
3578                                                                         $this->mLinkHolders['queries'][$key] );
3579                                 }
3580                         }
3581                         wfProfileOut( $fname.'-construct' );
3582
3583                         # Do the thing
3584                         wfProfileIn( $fname.'-replace' );
3585
3586                         $text = preg_replace_callback(
3587                                 '/(<!--LINK .*?-->)/',
3588                                 "wfOutputReplaceMatches",
3589                                 $text);
3590
3591                         wfProfileOut( $fname.'-replace' );
3592                 }
3593
3594                 # Now process interwiki link holders
3595                 # This is quite a bit simpler than internal links
3596                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
3597                         wfProfileIn( $fname.'-interwiki' );
3598                         # Make interwiki link HTML
3599                         $wgOutputReplace = array();
3600                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
3601                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
3602                                 $wgOutputReplace[$key] = $sk->makeLinkObj( $title, $link );
3603                         }
3604
3605                         $text = preg_replace_callback(
3606                                 '/<!--IWLINK (.*?)-->/',
3607                                 "wfOutputReplaceMatches",
3608                                 $text );
3609                         wfProfileOut( $fname.'-interwiki' );
3610                 }
3611
3612                 wfProfileOut( $fname );
3613                 return $colours;
3614         }
3615
3616         /**
3617          * Replace <!--LINK--> link placeholders with plain text of links
3618          * (not HTML-formatted).
3619          * @param string $text
3620          * @return string
3621          */
3622         function replaceLinkHoldersText( $text ) {
3623                 global $wgUser;
3624                 global $wgOutputReplace;
3625
3626                 $fname = 'Parser::replaceLinkHoldersText';
3627                 wfProfileIn( $fname );
3628
3629                 $text = preg_replace_callback(
3630                         '/<!--(LINK|IWLINK) (.*?)-->/',
3631                         array( &$this, 'replaceLinkHoldersTextCallback' ),
3632                         $text );
3633
3634                 wfProfileOut( $fname );
3635                 return $text;
3636         }
3637
3638         /**
3639          * @param array $matches
3640          * @return string
3641          * @access private
3642          */
3643         function replaceLinkHoldersTextCallback( $matches ) {
3644                 $type = $matches[1];
3645                 $key  = $matches[2];
3646                 if( $type == 'LINK' ) {
3647                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
3648                                 return $this->mLinkHolders['texts'][$key];
3649                         }
3650                 } elseif( $type == 'IWLINK' ) {
3651                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
3652                                 return $this->mInterwikiLinkHolders['texts'][$key];
3653                         }
3654                 }
3655                 return $matches[0];
3656         }
3657
3658         /**
3659          * Renders an image gallery from a text with one line per image.
3660          * text labels may be given by using |-style alternative text. E.g.
3661          *   Image:one.jpg|The number "1"
3662          *   Image:tree.jpg|A tree
3663          * given as text will return the HTML of a gallery with two images,
3664          * labeled 'The number "1"' and
3665          * 'A tree'.
3666          */
3667         function renderImageGallery( $text ) {
3668                 # Setup the parser
3669                 $parserOptions = new ParserOptions;
3670                 $localParser = new Parser();
3671
3672                 $ig = new ImageGallery();
3673                 $ig->setShowBytes( false );
3674                 $ig->setShowFilename( false );
3675                 $lines = explode( "\n", $text );
3676
3677                 foreach ( $lines as $line ) {
3678                         # match lines like these:
3679                         # Image:someimage.jpg|This is some image
3680                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
3681                         # Skip empty lines
3682                         if ( count( $matches ) == 0 ) {
3683                                 continue;
3684                         }
3685                         $nt =& Title::newFromText( $matches[1] );
3686                         if( is_null( $nt ) ) {
3687                                 # Bogus title. Ignore these so we don't bomb out later.
3688                                 continue;
3689                         }
3690                         if ( isset( $matches[3] ) ) {
3691                                 $label = $matches[3];
3692                         } else {
3693                                 $label = '';
3694                         }
3695
3696                         $pout = $localParser->parse( $label , $this->mTitle, $parserOptions );
3697                         $html = $pout->getText();
3698
3699                         $ig->add( new Image( $nt ), $html );
3700                         $this->mOutput->addImage( $nt->getDBkey() );
3701                 }
3702                 return $ig->toHTML();
3703         }
3704
3705         /**
3706          * Parse image options text and use it to make an image
3707          */
3708         function makeImage( &$nt, $options ) {
3709                 global $wgContLang, $wgUseImageResize, $wgUser;
3710
3711                 $align = '';
3712
3713                 # Check if the options text is of the form "options|alt text"
3714                 # Options are:
3715                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
3716                 #  * left               no resizing, just left align. label is used for alt= only
3717                 #  * right              same, but right aligned
3718                 #  * none               same, but not aligned
3719                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
3720                 #  * center             center the image
3721                 #  * framed             Keep original image size, no magnify-button.
3722
3723                 $part = explode( '|', $options);
3724
3725                 $mwThumb  =& MagicWord::get( MAG_IMG_THUMBNAIL );
3726                 $mwManualThumb =& MagicWord::get( MAG_IMG_MANUALTHUMB );
3727                 $mwLeft   =& MagicWord::get( MAG_IMG_LEFT );
3728                 $mwRight  =& MagicWord::get( MAG_IMG_RIGHT );
3729                 $mwNone   =& MagicWord::get( MAG_IMG_NONE );
3730                 $mwWidth  =& MagicWord::get( MAG_IMG_WIDTH );
3731                 $mwCenter =& MagicWord::get( MAG_IMG_CENTER );
3732                 $mwFramed =& MagicWord::get( MAG_IMG_FRAMED );
3733                 $caption = '';
3734
3735                 $width = $height = $framed = $thumb = false;
3736                 $manual_thumb = '' ;
3737
3738                 foreach( $part as $key => $val ) {
3739                         if ( $wgUseImageResize && ! is_null( $mwThumb->matchVariableStartToEnd($val) ) ) {
3740                                 $thumb=true;
3741                         } elseif ( ! is_null( $match = $mwManualThumb->matchVariableStartToEnd($val) ) ) {
3742                                 # use manually specified thumbnail
3743                                 $thumb=true;
3744                                 $manual_thumb = $match;
3745                         } elseif ( ! is_null( $mwRight->matchVariableStartToEnd($val) ) ) {
3746                                 # remember to set an alignment, don't render immediately
3747                                 $align = 'right';
3748                         } elseif ( ! is_null( $mwLeft->matchVariableStartToEnd($val) ) ) {
3749                                 # remember to set an alignment, don't render immediately
3750                                 $align = 'left';
3751                         } elseif ( ! is_null( $mwCenter->matchVariableStartToEnd($val) ) ) {
3752                                 # remember to set an alignment, don't render immediately
3753                                 $align = 'center';
3754                         } elseif ( ! is_null( $mwNone->matchVariableStartToEnd($val) ) ) {
3755                                 # remember to set an alignment, don't render immediately
3756                                 $align = 'none';
3757                         } elseif ( $wgUseImageResize && ! is_null( $match = $mwWidth->matchVariableStartToEnd($val) ) ) {
3758                                 wfDebug( "MAG_IMG_WIDTH match: $match\n" );
3759                                 # $match is the image width in pixels
3760                                 if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $match, $m ) ) {
3761                                         $width = intval( $m[1] );
3762                                         $height = intval( $m[2] );
3763                                 } else {
3764                                         $width = intval($match);
3765                                 }
3766                         } elseif ( ! is_null( $mwFramed->matchVariableStartToEnd($val) ) ) {
3767                                 $framed=true;
3768                         } else {
3769                                 $caption = $val;
3770                         }
3771                 }
3772                 # Strip bad stuff out of the alt text
3773                 $alt = $this->replaceLinkHoldersText( $caption );
3774                 $alt = Sanitizer::stripAllTags( $alt );
3775
3776                 # Linker does the rest
3777                 $sk =& $this->mOptions->getSkin();
3778                 return $sk->makeImageLinkObj( $nt, $caption, $alt, $align, $width, $height, $framed, $thumb, $manual_thumb );
3779         }
3780
3781         /**
3782          * Set a flag in the output object indicating that the content is dynamic and
3783          * shouldn't be cached.
3784          */
3785         function disableCache() {
3786                 $this->mOutput->mCacheTime = -1;
3787         }
3788
3789         /**#@+
3790          * Callback from the Sanitizer for expanding items found in HTML attribute
3791          * values, so they can be safely tested and escaped.
3792          * @param string $text
3793          * @param array $args
3794          * @return string
3795          * @access private
3796          */
3797         function attributeStripCallback( &$text, $args ) {
3798                 $text = $this->replaceVariables( $text, $args );
3799                 $text = $this->unstripForHTML( $text );
3800                 return $text;
3801         }
3802
3803         function unstripForHTML( $text ) {
3804                 $text = $this->unstrip( $text, $this->mStripState );
3805                 $text = $this->unstripNoWiki( $text, $this->mStripState );
3806                 return $text;
3807         }
3808         /**#@-*/
3809
3810         /**#@+
3811          * Accessor/mutator
3812          */
3813         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
3814         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
3815         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
3816         /**#@-*/
3817
3818         /**#@+
3819          * Accessor
3820          */
3821         function getTags() { return array_keys( $this->mTagHooks ); }
3822         /**#@-*/
3823 }
3824
3825 /**
3826  * @todo document
3827  * @package MediaWiki
3828  */
3829 class ParserOutput
3830 {
3831         var $mText,             # The output text
3832                 $mLanguageLinks,    # List of the full text of language links, in the order they appear
3833                 $mCategories,       # Map of category names to sort keys
3834                 $mContainsOldMagic, # Boolean variable indicating if the input contained variables like {{CURRENTDAY}}
3835                 $mCacheTime,        # Timestamp on this article, or -1 for uncacheable. Used in ParserCache.
3836                 $mVersion,          # Compatibility check
3837                 $mTitleText,        # title text of the chosen language variant
3838                 $mLinks,            # 2-D map of NS/DBK to ID for the links in the document. ID=zero for broken.
3839                 $mTemplates,        # 2-D map of NS/DBK to ID for the template references. ID=zero for broken.
3840                 $mImages,           # DB keys of the images used, in the array key only
3841                 $mExternalLinks;    # External link URLs, in the key only
3842
3843         function ParserOutput( $text = '', $languageLinks = array(), $categoryLinks = array(),
3844                 $containsOldMagic = false, $titletext = '' )
3845         {
3846                 $this->mText = $text;
3847                 $this->mLanguageLinks = $languageLinks;
3848                 $this->mCategories = $categoryLinks;
3849                 $this->mContainsOldMagic = $containsOldMagic;
3850                 $this->mCacheTime = '';
3851                 $this->mVersion = MW_PARSER_VERSION;
3852                 $this->mTitleText = $titletext;
3853                 $this->mLinks = array();
3854                 $this->mTemplates = array();
3855                 $this->mImages = array();
3856                 $this->mExternalLinks = array();
3857         }
3858
3859         function getText()                   { return $this->mText; }
3860         function getLanguageLinks()          { return $this->mLanguageLinks; }
3861         function getCategoryLinks()          { return array_keys( $this->mCategories ); }
3862         function &getCategories()            { return $this->mCategories; }
3863         function getCacheTime()              { return $this->mCacheTime; }
3864         function getTitleText()              { return $this->mTitleText; }
3865         function &getLinks()                 { return $this->mLinks; }
3866         function &getTemplates()             { return $this->mTemplates; }
3867         function &getImages()                { return $this->mImages; }
3868         function &getExternalLinks()         { return $this->mExternalLinks; }
3869
3870         function containsOldMagic()          { return $this->mContainsOldMagic; }
3871         function setText( $text )            { return wfSetVar( $this->mText, $text ); }
3872         function setLanguageLinks( $ll )     { return wfSetVar( $this->mLanguageLinks, $ll ); }
3873         function setCategoryLinks( $cl )     { return wfSetVar( $this->mCategories, $cl ); }
3874         function setContainsOldMagic( $com ) { return wfSetVar( $this->mContainsOldMagic, $com ); }
3875         function setCacheTime( $t )          { return wfSetVar( $this->mCacheTime, $t ); }
3876         function setTitleText( $t )          { return wfSetVar ($this->mTitleText, $t); }
3877
3878         function addCategory( $c, $sort )    { $this->mCategories[$c] = $sort; }
3879         function addImage( $name )           { $this->mImages[$name] = 1; }
3880         function addLanguageLink( $t )       { $this->mLanguageLinks[] = $t; }
3881         function addExternalLink( $url )     { $this->mExternalLinks[$url] = 1; }
3882
3883         function addLink( $title, $id ) {
3884                 $ns = $title->getNamespace();
3885                 $dbk = $title->getDBkey();
3886                 if ( !isset( $this->mLinks[$ns] ) ) {
3887                         $this->mLinks[$ns] = array();
3888                 }
3889                 $this->mLinks[$ns][$dbk] = $id;
3890         }
3891
3892         function addTemplate( $title, $id ) {
3893                 $ns = $title->getNamespace();
3894                 $dbk = $title->getDBkey();
3895                 if ( !isset( $this->mTemplates[$ns] ) ) {
3896                         $this->mTemplates[$ns] = array();
3897                 }
3898                 $this->mTemplates[$ns][$dbk] = $id;
3899         }
3900
3901         /**
3902          * @deprecated
3903          */
3904         /*
3905         function merge( $other ) {
3906                 $this->mLanguageLinks = array_merge( $this->mLanguageLinks, $other->mLanguageLinks );
3907                 $this->mCategories = array_merge( $this->mCategories, $this->mLanguageLinks );
3908                 $this->mContainsOldMagic = $this->mContainsOldMagic || $other->mContainsOldMagic;
3909         }*/
3910
3911         /**
3912          * Return true if this cached output object predates the global or
3913          * per-article cache invalidation timestamps, or if it comes from
3914          * an incompatible older version.
3915          *
3916          * @param string $touched the affected article's last touched timestamp
3917          * @return bool
3918          * @access public
3919          */
3920         function expired( $touched ) {
3921                 global $wgCacheEpoch;
3922                 return $this->getCacheTime() == -1 || // parser says it's uncacheable
3923                        $this->getCacheTime() < $touched ||
3924                        $this->getCacheTime() <= $wgCacheEpoch ||
3925                        !isset( $this->mVersion ) ||
3926                        version_compare( $this->mVersion, MW_PARSER_VERSION, "lt" );
3927         }
3928 }
3929
3930 /**
3931  * Set options of the Parser
3932  * @todo document
3933  * @package MediaWiki
3934  */
3935 class ParserOptions
3936 {
3937         # All variables are private
3938         var $mUseTeX;                    # Use texvc to expand <math> tags
3939         var $mUseDynamicDates;           # Use DateFormatter to format dates
3940         var $mInterwikiMagic;            # Interlanguage links are removed and returned in an array
3941         var $mAllowExternalImages;       # Allow external images inline
3942         var $mAllowExternalImagesFrom;   # If not, any exception?
3943         var $mSkin;                      # Reference to the preferred skin
3944         var $mDateFormat;                # Date format index
3945         var $mEditSection;               # Create "edit section" links
3946         var $mNumberHeadings;            # Automatically number headings
3947         var $mAllowSpecialInclusion;     # Allow inclusion of special pages
3948         var $mTidy;                      # Ask for tidy cleanup
3949
3950         function getUseTeX()                        { return $this->mUseTeX; }
3951         function getUseDynamicDates()               { return $this->mUseDynamicDates; }
3952         function getInterwikiMagic()                { return $this->mInterwikiMagic; }
3953         function getAllowExternalImages()           { return $this->mAllowExternalImages; }
3954         function getAllowExternalImagesFrom()       { return $this->mAllowExternalImagesFrom; }
3955         function &getSkin()                         { return $this->mSkin; }
3956         function getDateFormat()                    { return $this->mDateFormat; }
3957         function getEditSection()                   { return $this->mEditSection; }
3958         function getNumberHeadings()                { return $this->mNumberHeadings; }
3959         function getAllowSpecialInclusion()         { return $this->mAllowSpecialInclusion; }
3960         function getTidy()                          { return $this->mTidy; }
3961
3962         function setUseTeX( $x )                    { return wfSetVar( $this->mUseTeX, $x ); }
3963         function setUseDynamicDates( $x )           { return wfSetVar( $this->mUseDynamicDates, $x ); }
3964         function setInterwikiMagic( $x )            { return wfSetVar( $this->mInterwikiMagic, $x ); }
3965         function setAllowExternalImages( $x )       { return wfSetVar( $this->mAllowExternalImages, $x ); }
3966         function setAllowExternalImagesFrom( $x )   { return wfSetVar( $this->mAllowExternalImagesFrom, $x ); }
3967         function setDateFormat( $x )                { return wfSetVar( $this->mDateFormat, $x ); }
3968         function setEditSection( $x )               { return wfSetVar( $this->mEditSection, $x ); }
3969         function setNumberHeadings( $x )            { return wfSetVar( $this->mNumberHeadings, $x ); }
3970         function setAllowSpecialInclusion( $x )     { return wfSetVar( $this->mAllowSpecialInclusion, $x ); }
3971         function setTidy( $x )                      { return wfSetVar( $this->mTidy, $x); }
3972         function setSkin( &$x ) { $this->mSkin =& $x; }
3973
3974         function ParserOptions() {
3975                 global $wgUser;
3976                 $this->initialiseFromUser( $wgUser );
3977         }
3978
3979         /**
3980          * Get parser options
3981          * @static
3982          */
3983         function newFromUser( &$user ) {
3984                 $popts = new ParserOptions;
3985                 $popts->initialiseFromUser( $user );
3986                 return $popts;
3987         }
3988
3989         /** Get user options */
3990         function initialiseFromUser( &$userInput ) {
3991                 global $wgUseTeX, $wgUseDynamicDates, $wgInterwikiMagic, $wgAllowExternalImages,
3992                        $wgAllowExternalImagesFrom, $wgAllowSpecialInclusion;
3993                 $fname = 'ParserOptions::initialiseFromUser';
3994                 wfProfileIn( $fname );
3995                 if ( !$userInput ) {
3996                         $user = new User;
3997                         $user->setLoaded( true );
3998                 } else {
3999                         $user =& $userInput;
4000                 }
4001
4002                 $this->mUseTeX = $wgUseTeX;
4003                 $this->mUseDynamicDates = $wgUseDynamicDates;
4004                 $this->mInterwikiMagic = $wgInterwikiMagic;
4005                 $this->mAllowExternalImages = $wgAllowExternalImages;
4006                 $this->mAllowExternalImagesFrom = $wgAllowExternalImagesFrom;
4007                 wfProfileIn( $fname.'-skin' );
4008                 $this->mSkin =& $user->getSkin();
4009                 wfProfileOut( $fname.'-skin' );
4010                 $this->mDateFormat = $user->getOption( 'date' );
4011                 $this->mEditSection = true;
4012                 $this->mNumberHeadings = $user->getOption( 'numberheadings' );
4013                 $this->mAllowSpecialInclusion = $wgAllowSpecialInclusion;
4014                 $this->mTidy = false;
4015                 wfProfileOut( $fname );
4016         }
4017 }
4018
4019 /**
4020  * Callback function used by Parser::replaceLinkHolders()
4021  * to substitute link placeholders.
4022  */
4023 function &wfOutputReplaceMatches( $matches ) {
4024         global $wgOutputReplace;
4025         return $wgOutputReplace[$matches[1]];
4026 }
4027
4028 /**
4029  * Return the total number of articles
4030  */
4031 function wfNumberOfArticles() {
4032         global $wgNumberOfArticles;
4033
4034         wfLoadSiteStats();
4035         return $wgNumberOfArticles;
4036 }
4037
4038 /**
4039  * Return the number of files
4040  */
4041 function wfNumberOfFiles() {
4042         $fname = 'Parser::wfNumberOfFiles';
4043
4044         wfProfileIn( $fname );
4045         $dbr =& wfGetDB( DB_SLAVE );
4046         $res = $dbr->selectField('image', 'COUNT(*)', array(), $fname );
4047         wfProfileOut( $fname );
4048
4049         return $res;
4050 }
4051
4052 /**
4053  * Get various statistics from the database
4054  * @access private
4055  */
4056 function wfLoadSiteStats() {
4057         global $wgNumberOfArticles, $wgTotalViews, $wgTotalEdits;
4058         $fname = 'wfLoadSiteStats';
4059
4060         if ( -1 != $wgNumberOfArticles ) return;
4061         $dbr =& wfGetDB( DB_SLAVE );
4062         $s = $dbr->selectRow( 'site_stats',
4063                 array( 'ss_total_views', 'ss_total_edits', 'ss_good_articles' ),
4064                 array( 'ss_row_id' => 1 ), $fname
4065         );
4066
4067         if ( $s === false ) {
4068                 return;
4069         } else {
4070                 $wgTotalViews = $s->ss_total_views;
4071                 $wgTotalEdits = $s->ss_total_edits;
4072                 $wgNumberOfArticles = $s->ss_good_articles;
4073         }
4074 }
4075
4076 /**
4077  * Escape html tags
4078  * Basically replacing " > and < with HTML entities ( &quot;, &gt;, &lt;)
4079  *
4080  * @param string $in Text that might contain HTML tags
4081  * @return string Escaped string
4082  */
4083 function wfEscapeHTMLTagsOnly( $in ) {
4084         return str_replace(
4085                 array( '"', '>', '<' ),
4086                 array( '&quot;', '&gt;', '&lt;' ),
4087                 $in );
4088 }
4089
4090 ?>