includes/Parser.php

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