includes/Parser.php

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