includes/Parser.php

   1 <?php
   2 /**
   3  * File for Parser and related classes
   4  *
   5  * @package MediaWiki
   6  * @subpackage Parser
   7  */
   8
   9 /** */
  10 require_once( 'Sanitizer.php' );
  11
  12 /**
  13  * Update this version number when the ParserOutput format
  14  * changes in an incompatible way, so the parser cache
  15  * can automatically discard old data.
  16  */
  17 define( 'MW_PARSER_VERSION', '1.5.0' );
  18
  19 /**
  20  * Variable substitution O(N^2) attack
  21  *
  22  * Without countermeasures, it would be possible to attack the parser by saving
  23  * a page filled with a large number of inclusions of large pages. The size of
  24  * the generated page would be proportional to the square of the input size.
  25  * Hence, we limit the number of inclusions of any given page, thus bringing any
  26  * attack back to O(N).
  27  */
  28
  29 define( 'MAX_INCLUDE_REPEAT', 100 );
  30 define( 'MAX_INCLUDE_SIZE', 1000000 ); // 1 Million
  31
  32 define( 'RLH_FOR_UPDATE', 1 );
  33
  34 # Allowed values for $mOutputType
  35 define( 'OT_HTML', 1 );
  36 define( 'OT_WIKI', 2 );
  37 define( 'OT_MSG' , 3 );
  38
  39 # string parameter for extractTags which will cause it
  40 # to strip HTML comments in addition to regular
  41 # <XML>-style tags. This should not be anything we
  42 # may want to use in wikisyntax
  43 define( 'STRIP_COMMENTS', 'HTMLCommentStrip' );
  44
  45 # prefix for escaping, used in two functions at least
  46 define( 'UNIQ_PREFIX', 'NaodW29');
  47
  48 # Constants needed for external link processing
  49 define( 'URL_PROTOCOLS', 'http|https|ftp|irc|gopher|news|mailto' );
  50 define( 'HTTP_PROTOCOLS', 'http|https' );
  51 # Everything except bracket, space, or control characters
  52 define( 'EXT_LINK_URL_CLASS', '[^]<>"\\x00-\\x20\\x7F]' );
  53 # Including space
  54 define( 'EXT_LINK_TEXT_CLASS', '[^\]\\x00-\\x1F\\x7F]' );
  55 define( 'EXT_IMAGE_FNAME_CLASS', '[A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]' );
  56 define( 'EXT_IMAGE_EXTENSIONS', 'gif|png|jpg|jpeg' );
  57 define( 'EXT_LINK_BRACKETED',  '/\[(\b('.URL_PROTOCOLS.'):'.EXT_LINK_URL_CLASS.'+) *('.EXT_LINK_TEXT_CLASS.'*?)\]/S' );
  58 define( 'EXT_IMAGE_REGEX',
  59         '/^('.HTTP_PROTOCOLS.':)'.  # Protocol
  60         '('.EXT_LINK_URL_CLASS.'+)\\/'.  # Hostname and path
  61         '('.EXT_IMAGE_FNAME_CLASS.'+)\\.((?i)'.EXT_IMAGE_EXTENSIONS.')$/S' # Filename
  62 );
  63
  64 /**
  65  * PHP Parser
  66  *
  67  * Processes wiki markup
  68  *
  69  * <pre>
  70  * There are three main entry points into the Parser class:
  71  * parse()
  72  *   produces HTML output
  73  * preSaveTransform().
  74  *   produces altered wiki markup.
  75  * transformMsg()
  76  *   performs brace substitution on MediaWiki messages
  77  *
  78  * Globals used:
  79  *    objects:   $wgLang, $wgLinkCache
  80  *
  81  * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
  82  *
  83  * settings:
  84  *  $wgUseTex*, $wgUseDynamicDates*, $wgInterwikiMagic*,
  85  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  86  *  $wgLocaltimezone, $wgAllowSpecialInclusion*
  87  *
  88  *  * only within ParserOptions
  89  * </pre>
  90  *
  91  * @package MediaWiki
  92  */
  93 class Parser
  94 {
  95         /**#@+
  96          * @access private
  97          */
  98         # Persistent:
  99         var $mTagHooks;
 100
 101         # Cleared with clearState():
 102         var $mOutput, $mAutonumber, $mDTopen, $mStripState = array();
 103         var $mVariables, $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 104         var $mInterwikiLinkHolders, $mLinkHolders;
 105
 106         # Temporary:
 107         var $mOptions, $mTitle, $mOutputType,
 108             $mTemplates,        // cache of already loaded templates, avoids
 109                                 // multiple SQL queries for the same string
 110             $mTemplatePath;     // stores an unsorted hash of all the templates already loaded
 111                                 // in this path. Used for loop detection.
 112
 113         /**#@-*/
 114
 115         /**
 116          * Constructor
 117          *
 118          * @access public
 119          */
 120         function Parser() {
 121                 global $wgContLang;
 122                 $this->mTemplates = array();
 123                 $this->mTemplatePath = array();
 124                 $this->mTagHooks = array();
 125                 $this->clearState();
 126         }
 127
 128         /**
 129          * Clear Parser state
 130          *
 131          * @access private
 132          */
 133         function clearState() {
 134                 $this->mOutput = new ParserOutput;
 135                 $this->mAutonumber = 0;
 136                 $this->mLastSection = '';
 137                 $this->mDTopen = false;
 138                 $this->mVariables = false;
 139                 $this->mIncludeCount = array();
 140                 $this->mStripState = array();
 141                 $this->mArgStack = array();
 142                 $this->mInPre = false;
 143                 $this->mInterwikiLinkHolders = array(
 144                         'texts' => array(),
 145                         'titles' => array()
 146                 );
 147                 $this->mLinkHolders = array(
 148                         'namespaces' => array(),
 149                         'dbkeys' => array(),
 150                         'queries' => array(),
 151                         'texts' => array(),
 152                         'titles' => array()
 153                 );
 154         }
 155
 156         /**
 157          * First pass--just handle <nowiki> sections, pass the rest off
 158          * to internalParse() which does all the real work.
 159          *
 160          * @access private
 161          * @param string $text Text we want to parse
 162          * @param Title &$title A title object
 163          * @param array $options
 164          * @param boolean $linestart
 165          * @param boolean $clearState
 166          * @return ParserOutput a ParserOutput
 167          */
 168         function parse( $text, &$title, $options, $linestart = true, $clearState = true ) {
 169                 global $wgUseTidy, $wgContLang;
 170                 $fname = 'Parser::parse';
 171                 wfProfileIn( $fname );
 172
 173                 if ( $clearState ) {
 174                         $this->clearState();
 175                 }
 176
 177                 $this->mOptions = $options;
 178                 $this->mTitle =& $title;
 179                 $this->mOutputType = OT_HTML;
 180
 181                 $this->mStripState = NULL;
 182
 183                 //$text = $this->strip( $text, $this->mStripState );
 184                 // VOODOO MAGIC FIX! Sometimes the above segfaults in PHP5.
 185                 $x =& $this->mStripState;
 186                 $text = $this->strip( $text, $x );
 187
 188                 $text = $this->internalParse( $text );
 189
 190
 191                 $text = $this->unstrip( $text, $this->mStripState );
 192
 193                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 194                 $fixtags = array(
 195                         # french spaces, last one Guillemet-left
 196                         # only if there is something before the space
 197                         '/(.) (?=\\?|:|;|!|\\302\\273)/' => '\\1&nbsp;\\2',
 198                         # french spaces, Guillemet-right
 199                         '/(\\302\\253) /' => '\\1&nbsp;',
 200                         '/<center *>/i' => '<div class="center">',
 201                         '/<\\/center *>/i' => '</div>',
 202                 );
 203                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 204
 205                 # only once and last
 206                 $text = $this->doBlockLevels( $text, $linestart );
 207
 208                 $this->replaceLinkHolders( $text );
 209
 210                 $dashReplace = array(
 211                         '/ - /' => "&nbsp;&ndash; ", # N dash
 212                         '/(?<=[\d])-(?=[\d])/' => "&ndash;", # N dash between numbers
 213                         '/ -- /' => "&nbsp;&mdash; " # M dash
 214                 );
 215                 $text = preg_replace( array_keys($dashReplace), array_values($dashReplace), $text );
 216
 217                 # the position of the convert() call should not be changed. it
 218                 # assumes that the links are all replaces and the only thing left
 219                 # is the <nowiki> mark.
 220                 $text = $wgContLang->convert($text);
 221                 $this->mOutput->setTitleText($wgContLang->getParsedTitle());
 222
 223                 $text = $this->unstripNoWiki( $text, $this->mStripState );
 224
 225                 $text = Sanitizer::normalizeCharReferences( $text );
 226                 global $wgUseTidy;
 227                 if ($wgUseTidy) {
 228                         $text = Parser::tidy($text);
 229                 }
 230
 231                 $this->mOutput->setText( $text );
 232                 wfProfileOut( $fname );
 233                 return $this->mOutput;
 234         }
 235
 236         /**
 237          * Get a random string
 238          *
 239          * @access private
 240          * @static
 241          */
 242         function getRandomString() {
 243                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 244         }
 245
 246         /**
 247          * Replaces all occurrences of <$tag>content</$tag> in the text
 248          * with a random marker and returns the new text. the output parameter
 249          * $content will be an associative array filled with data on the form
 250          * $unique_marker => content.
 251          *
 252          * If $content is already set, the additional entries will be appended
 253          * If $tag is set to STRIP_COMMENTS, the function will extract
 254          * <!-- HTML comments -->
 255          *
 256          * @access private
 257          * @static
 258          */
 259         function extractTagsAndParams($tag, $text, &$content, &$tags, &$params, $uniq_prefix = ''){
 260                 $rnd = $uniq_prefix . '-' . $tag . Parser::getRandomString();
 261                 if ( !$content ) {
 262                         $content = array( );
 263                 }
 264                 $n = 1;
 265                 $stripped = '';
 266
 267                 if ( !$tags ) {
 268                         $tags = array( );
 269                 }
 270
 271                 if ( !$params ) {
 272                         $params = array( );
 273                 }
 274
 275                 if( $tag == STRIP_COMMENTS ) {
 276                         $start = '/<!--()/';
 277                         $end   = '/-->/';
 278                 } else {
 279                         $start = "/<$tag(\\s+[^>]*|\\s*)>/i";
 280                         $end   = "/<\\/$tag\\s*>/i";
 281                 }
 282
 283                 while ( '' != $text ) {
 284                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 285                         $stripped .= $p[0];
 286                         if( count( $p ) < 3 ) {
 287                                 break;
 288                         }
 289                         $attributes = $p[1];
 290                         $inside     = $p[2];
 291
 292                         $marker = $rnd . sprintf('%08X', $n++);
 293                         $stripped .= $marker;
 294
 295                         $tags[$marker] = "<$tag$attributes>";
 296                         $params[$marker] = Sanitizer::decodeTagAttributes( $attributes );
 297
 298                         $q = preg_split( $end, $inside, 2 );
 299                         $content[$marker] = $q[0];
 300                         if( count( $q ) < 1 ) {
 301                                 # No end tag -- let it run out to the end of the text.
 302                                 break;
 303                         } else {
 304                                 $text = $q[1];
 305                         }
 306                 }
 307                 return $stripped;
 308         }
 309
 310         /**
 311          * Wrapper function for extractTagsAndParams
 312          * for cases where $tags and $params isn't needed
 313          * i.e. where tags will never have params, like <nowiki>
 314          *
 315          * @access private
 316          * @static
 317          */
 318         function extractTags( $tag, $text, &$content, $uniq_prefix = '' ) {
 319                 $dummy_tags = array();
 320                 $dummy_params = array();
 321
 322                 return Parser::extractTagsAndParams( $tag, $text, $content,
 323                         $dummy_tags, $dummy_params, $uniq_prefix );
 324         }
 325
 326         /**
 327          * Strips and renders nowiki, pre, math, hiero
 328          * If $render is set, performs necessary rendering operations on plugins
 329          * Returns the text, and fills an array with data needed in unstrip()
 330          * If the $state is already a valid strip state, it adds to the state
 331          *
 332          * @param bool $stripcomments when set, HTML comments <!-- like this -->
 333          *  will be stripped in addition to other tags. This is important
 334          *  for section editing, where these comments cause confusion when
 335          *  counting the sections in the wikisource
 336          *
 337          * @access private
 338          */
 339         function strip( $text, &$state, $stripcomments = false ) {
 340                 $render = ($this->mOutputType == OT_HTML);
 341                 $html_content = array();
 342                 $nowiki_content = array();
 343                 $math_content = array();
 344                 $pre_content = array();
 345                 $comment_content = array();
 346                 $ext_content = array();
 347                 $ext_tags = array();
 348                 $ext_params = array();
 349                 $gallery_content = array();
 350
 351                 # Replace any instances of the placeholders
 352                 $uniq_prefix = UNIQ_PREFIX;
 353                 #$text = str_replace( $uniq_prefix, wfHtmlEscapeFirst( $uniq_prefix ), $text );
 354
 355                 # html
 356                 global $wgRawHtml, $wgWhitelistEdit;
 357                 if( $wgRawHtml && $wgWhitelistEdit ) {
 358                         $text = Parser::extractTags('html', $text, $html_content, $uniq_prefix);
 359                         foreach( $html_content as $marker => $content ) {
 360                                 if ($render ) {
 361                                         # Raw and unchecked for validity.
 362                                         $html_content[$marker] = $content;
 363                                 } else {
 364                                         $html_content[$marker] = '<html>'.$content.'</html>';
 365                                 }
 366                         }
 367                 }
 368
 369                 # nowiki
 370                 $text = Parser::extractTags('nowiki', $text, $nowiki_content, $uniq_prefix);
 371                 foreach( $nowiki_content as $marker => $content ) {
 372                         if( $render ){
 373                                 $nowiki_content[$marker] = wfEscapeHTMLTagsOnly( $content );
 374                         } else {
 375                                 $nowiki_content[$marker] = '<nowiki>'.$content.'</nowiki>';
 376                         }
 377                 }
 378
 379                 # math
 380                 $text = Parser::extractTags('math', $text, $math_content, $uniq_prefix);
 381                 foreach( $math_content as $marker => $content ){
 382                         if( $render ) {
 383                                 if( $this->mOptions->getUseTeX() ) {
 384                                         $math_content[$marker] = renderMath( $content );
 385                                 } else {
 386                                         $math_content[$marker] = '&lt;math&gt;'.$content.'&lt;math&gt;';
 387                                 }
 388                         } else {
 389                                 $math_content[$marker] = '<math>'.$content.'</math>';
 390                         }
 391                 }
 392
 393                 # pre
 394                 $text = Parser::extractTags('pre', $text, $pre_content, $uniq_prefix);
 395                 foreach( $pre_content as $marker => $content ){
 396                         if( $render ){
 397                                 $pre_content[$marker] = '<pre>' . wfEscapeHTMLTagsOnly( $content ) . '</pre>';
 398                         } else {
 399                                 $pre_content[$marker] = '<pre>'.$content.'</pre>';
 400                         }
 401                 }
 402
 403                 # gallery
 404                 $text = Parser::extractTags('gallery', $text, $gallery_content, $uniq_prefix);
 405                 foreach( $gallery_content as $marker => $content ) {
 406                         require_once( 'ImageGallery.php' );
 407                         if ( $render ) {
 408                                 $gallery_content[$marker] = Parser::renderImageGallery( $content );
 409                         } else {
 410                                 $gallery_content[$marker] = '<gallery>'.$content.'</gallery>';
 411                         }
 412                 }
 413
 414                 # Comments
 415                 if($stripcomments) {
 416                         $text = Parser::extractTags(STRIP_COMMENTS, $text, $comment_content, $uniq_prefix);
 417                         foreach( $comment_content as $marker => $content ){
 418                                 $comment_content[$marker] = '<!--'.$content.'-->';
 419                         }
 420                 }
 421
 422                 # Extensions
 423                 foreach ( $this->mTagHooks as $tag => $callback ) {
 424                         $ext_content[$tag] = array();
 425                         $text = Parser::extractTagsAndParams( $tag, $text, $ext_content[$tag],
 426                                 $ext_tags[$tag], $ext_params[$tag], $uniq_prefix );
 427                         foreach( $ext_content[$tag] as $marker => $content ) {
 428                                 $full_tag = $ext_tags[$tag][$marker];
 429                                 $params = $ext_params[$tag][$marker];
 430                                 if ( $render ) {
 431                                         $ext_content[$tag][$marker] = $callback( $content, $params );
 432                                 } else {
 433                                         $ext_content[$tag][$marker] = "$full_tag$content</$tag>";
 434                                 }
 435                         }
 436                 }
 437
 438                 # Merge state with the pre-existing state, if there is one
 439                 if ( $state ) {
 440                         $state['html'] = $state['html'] + $html_content;
 441                         $state['nowiki'] = $state['nowiki'] + $nowiki_content;
 442                         $state['math'] = $state['math'] + $math_content;
 443                         $state['pre'] = $state['pre'] + $pre_content;
 444                         $state['comment'] = $state['comment'] + $comment_content;
 445                         $state['gallery'] = $state['gallery'] + $gallery_content;
 446
 447                         foreach( $ext_content as $tag => $array ) {
 448                                 if ( array_key_exists( $tag, $state ) ) {
 449                                         $state[$tag] = $state[$tag] + $array;
 450                                 }
 451                         }
 452                 } else {
 453                         $state = array(
 454                           'html' => $html_content,
 455                           'nowiki' => $nowiki_content,
 456                           'math' => $math_content,
 457                           'pre' => $pre_content,
 458                           'comment' => $comment_content,
 459                           'gallery' => $gallery_content,
 460                         ) + $ext_content;
 461                 }
 462                 return $text;
 463         }
 464
 465         /**
 466          * restores pre, math, and hiero removed by strip()
 467          *
 468          * always call unstripNoWiki() after this one
 469          * @access private
 470          */
 471         function unstrip( $text, &$state ) {
 472                 # Must expand in reverse order, otherwise nested tags will be corrupted
 473                 $contentDict = end( $state );
 474                 for ( $contentDict = end( $state ); $contentDict !== false; $contentDict = prev( $state ) ) {
 475                         if( key($state) != 'nowiki' && key($state) != 'html') {
 476                                 for ( $content = end( $contentDict ); $content !== false; $content = prev( $contentDict ) ) {
 477                                         $text = str_replace( key( $contentDict ), $content, $text );
 478                                 }
 479                         }
 480                 }
 481
 482                 return $text;
 483         }
 484
 485         /**
 486          * always call this after unstrip() to preserve the order
 487          *
 488          * @access private
 489          */
 490         function unstripNoWiki( $text, &$state ) {
 491                 # Must expand in reverse order, otherwise nested tags will be corrupted
 492                 for ( $content = end($state['nowiki']); $content !== false; $content = prev( $state['nowiki'] ) ) {
 493                         $text = str_replace( key( $state['nowiki'] ), $content, $text );
 494                 }
 495
 496                 global $wgRawHtml;
 497                 if ($wgRawHtml) {
 498                         for ( $content = end($state['html']); $content !== false; $content = prev( $state['html'] ) ) {
 499                                 $text = str_replace( key( $state['html'] ), $content, $text );
 500                         }
 501                 }
 502
 503                 return $text;
 504         }
 505
 506         /**
 507          * Add an item to the strip state
 508          * Returns the unique tag which must be inserted into the stripped text
 509          * The tag will be replaced with the original text in unstrip()
 510          *
 511          * @access private
 512          */
 513         function insertStripItem( $text, &$state ) {
 514                 $rnd = UNIQ_PREFIX . '-item' . Parser::getRandomString();
 515                 if ( !$state ) {
 516                         $state = array(
 517                           'html' => array(),
 518                           'nowiki' => array(),
 519                           'math' => array(),
 520                           'pre' => array()
 521                         );
 522                 }
 523                 $state['item'][$rnd] = $text;
 524                 return $rnd;
 525         }
 526
 527         /**
 528          * Interface with html tidy, used if $wgUseTidy = true.
 529          * If tidy isn't able to correct the markup, the original will be
 530          * returned in all its glory with a warning comment appended.
 531          *
 532          * Either the external tidy program or the in-process tidy extension
 533          * will be used depending on availability. Override the default
 534          * $wgTidyInternal setting to disable the internal if it's not working.
 535          *
 536          * @param string $text Hideous HTML input
 537          * @return string Corrected HTML output
 538          * @access public
 539          * @static
 540          */
 541         function tidy( $text ) {
 542                 global $wgTidyInternal;
 543                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 544 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 545 '<head><title>test</title></head><body>'.$text.'</body></html>';
 546                 if( $wgTidyInternal ) {
 547                         $correctedtext = Parser::internalTidy( $wrappedtext );
 548                 } else {
 549                         $correctedtext = Parser::externalTidy( $wrappedtext );
 550                 }
 551                 if( is_null( $correctedtext ) ) {
 552                         wfDebug( "Tidy error detected!\n" );
 553                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 554                 }
 555                 return $correctedtext;
 556         }
 557
 558         /**
 559          * Spawn an external HTML tidy process and get corrected markup back from it.
 560          *
 561          * @access private
 562          * @static
 563          */
 564         function externalTidy( $text ) {
 565                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 566                 $fname = 'Parser::externalTidy';
 567                 wfProfileIn( $fname );
 568
 569                 $cleansource = '';
 570                 $opts = ' -utf8';
 571
 572                 $descriptorspec = array(
 573                         0 => array('pipe', 'r'),
 574                         1 => array('pipe', 'w'),
 575                         2 => array('file', '/dev/null', 'a')
 576                 );
 577                 $pipes = array();
 578                 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 579                 if (is_resource($process)) {
 580                         fwrite($pipes[0], $text);
 581                         fclose($pipes[0]);
 582                         while (!feof($pipes[1])) {
 583                                 $cleansource .= fgets($pipes[1], 1024);
 584                         }
 585                         fclose($pipes[1]);
 586                         proc_close($process);
 587                 }
 588
 589                 wfProfileOut( $fname );
 590
 591                 if( $cleansource == '' && $text != '') {
 592                         // Some kind of error happened, so we couldn't get the corrected text.
 593                         // Just give up; we'll use the source text and append a warning.
 594                         return null;
 595                 } else {
 596                         return $cleansource;
 597                 }
 598         }
 599
 600         /**
 601          * Use the HTML tidy PECL extension to use the tidy library in-process,
 602          * saving the overhead of spawning a new process. Currently written to
 603          * the PHP 4.3.x version of the extension, may not work on PHP 5.
 604          *
 605          * 'pear install tidy' should be able to compile the extension module.
 606          *
 607          * @access private
 608          * @static
 609          */
 610         function internalTidy( $text ) {
 611                 global $wgTidyConf;
 612                 $fname = 'Parser::internalTidy';
 613                 wfProfileIn( $fname );
 614
 615                 tidy_load_config( $wgTidyConf );
 616                 tidy_set_encoding( 'utf8' );
 617                 tidy_parse_string( $text );
 618                 tidy_clean_repair();
 619                 if( tidy_get_status() == 2 ) {
 620                         // 2 is magic number for fatal error
 621                         // http://www.php.net/manual/en/function.tidy-get-status.php
 622                         $cleansource = null;
 623                 } else {
 624                         $cleansource = tidy_get_output();
 625                 }
 626                 wfProfileOut( $fname );
 627                 return $cleansource;
 628         }
 629
 630         /**
 631          * parse the wiki syntax used to render tables
 632          *
 633          * @access private
 634          */
 635         function doTableStuff ( $t ) {
 636                 $fname = 'Parser::doTableStuff';
 637                 wfProfileIn( $fname );
 638
 639                 $t = explode ( "\n" , $t ) ;
 640                 $td = array () ; # Is currently a td tag open?
 641                 $ltd = array () ; # Was it TD or TH?
 642                 $tr = array () ; # Is currently a tr tag open?
 643                 $ltr = array () ; # tr attributes
 644                 $indent_level = 0; # indent level of the table
 645                 foreach ( $t AS $k => $x )
 646                 {
 647                         $x = trim ( $x ) ;
 648                         $fc = substr ( $x , 0 , 1 ) ;
 649                         if ( preg_match( '/^(:*)\{\|(.*)$/', $x, $matches ) ) {
 650                                 $indent_level = strlen( $matches[1] );
 651                                 $t[$k] = str_repeat( '<dl><dd>', $indent_level ) .
 652                                         '<table' . Sanitizer::fixTagAttributes ( $matches[2], 'table' ) . '>' ;
 653                                 array_push ( $td , false ) ;
 654                                 array_push ( $ltd , '' ) ;
 655                                 array_push ( $tr , false ) ;
 656                                 array_push ( $ltr , '' ) ;
 657                         }
 658                         else if ( count ( $td ) == 0 ) { } # Don't do any of the following
 659                         else if ( '|}' == substr ( $x , 0 , 2 ) ) {
 660                                 $z = "</table>" . substr ( $x , 2);
 661                                 $l = array_pop ( $ltd ) ;
 662                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 663                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 664                                 array_pop ( $ltr ) ;
 665                                 $t[$k] = $z . str_repeat( '</dd></dl>', $indent_level );
 666                         }
 667                         else if ( '|-' == substr ( $x , 0 , 2 ) ) { # Allows for |---------------
 668                                 $x = substr ( $x , 1 ) ;
 669                                 while ( $x != '' && substr ( $x , 0 , 1 ) == '-' ) $x = substr ( $x , 1 ) ;
 670                                 $z = '' ;
 671                                 $l = array_pop ( $ltd ) ;
 672                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 673                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 674                                 array_pop ( $ltr ) ;
 675                                 $t[$k] = $z ;
 676                                 array_push ( $tr , false ) ;
 677                                 array_push ( $td , false ) ;
 678                                 array_push ( $ltd , '' ) ;
 679                                 array_push ( $ltr , Sanitizer::fixTagAttributes ( $x, 'tr' ) ) ;
 680                         }
 681                         else if ( '|' == $fc || '!' == $fc || '|+' == substr ( $x , 0 , 2 ) ) { # Caption
 682                                 # $x is a table row
 683                                 if ( '|+' == substr ( $x , 0 , 2 ) ) {
 684                                         $fc = '+' ;
 685                                         $x = substr ( $x , 1 ) ;
 686                                 }
 687                                 $after = substr ( $x , 1 ) ;
 688                                 if ( $fc == '!' ) $after = str_replace ( '!!' , '||' , $after ) ;
 689                                 $after = explode ( '||' , $after ) ;
 690                                 $t[$k] = '' ;
 691
 692                                 # Loop through each table cell
 693                                 foreach ( $after AS $theline )
 694                                 {
 695                                         $z = '' ;
 696                                         if ( $fc != '+' )
 697                                         {
 698                                                 $tra = array_pop ( $ltr ) ;
 699                                                 if ( !array_pop ( $tr ) ) $z = '<tr'.$tra.">\n" ;
 700                                                 array_push ( $tr , true ) ;
 701                                                 array_push ( $ltr , '' ) ;
 702                                         }
 703
 704                                         $l = array_pop ( $ltd ) ;
 705                                         if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 706                                         if ( $fc == '|' ) $l = 'td' ;
 707                                         else if ( $fc == '!' ) $l = 'th' ;
 708                                         else if ( $fc == '+' ) $l = 'caption' ;
 709                                         else $l = '' ;
 710                                         array_push ( $ltd , $l ) ;
 711
 712                                         # Cell parameters
 713                                         $y = explode ( '|' , $theline , 2 ) ;
 714                                         # Note that a '|' inside an invalid link should not
 715                                         # be mistaken as delimiting cell parameters
 716                                         if ( strpos( $y[0], '[[' ) !== false ) {
 717                                                 $y = array ($theline);
 718                                         }
 719                                         if ( count ( $y ) == 1 )
 720                                                 $y = "{$z}<{$l}>{$y[0]}" ;
 721                                         else $y = $y = "{$z}<{$l}".Sanitizer::fixTagAttributes($y[0], $l).">{$y[1]}" ;
 722                                         $t[$k] .= $y ;
 723                                         array_push ( $td , true ) ;
 724                                 }
 725                         }
 726                 }
 727
 728                 # Closing open td, tr && table
 729                 while ( count ( $td ) > 0 )
 730                 {
 731                         if ( array_pop ( $td ) ) $t[] = '</td>' ;
 732                         if ( array_pop ( $tr ) ) $t[] = '</tr>' ;
 733                         $t[] = '</table>' ;
 734                 }
 735
 736                 $t = implode ( "\n" , $t ) ;
 737                 wfProfileOut( $fname );
 738                 return $t ;
 739         }
 740
 741         /**
 742          * Helper function for parse() that transforms wiki markup into
 743          * HTML. Only called for $mOutputType == OT_HTML.
 744          *
 745          * @access private
 746          */
 747         function internalParse( $text ) {
 748                 global $wgContLang;
 749                 $args = array();
 750                 $isMain = true;
 751                 $fname = 'Parser::internalParse';
 752                 wfProfileIn( $fname );
 753
 754                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'replaceVariables' ) );
 755                 $text = $this->replaceVariables( $text, $args );
 756
 757                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 758
 759                 $text = $this->doHeadings( $text );
 760                 if($this->mOptions->getUseDynamicDates()) {
 761                         $df =& DateFormatter::getInstance();
 762                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 763                 }
 764                 $text = $this->doAllQuotes( $text );
 765                 $text = $this->replaceInternalLinks( $text );
 766                 $text = $this->replaceExternalLinks( $text );
 767
 768                 # replaceInternalLinks may sometimes leave behind
 769                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 770                 $text = str_replace("http-noparse://","http://",$text);
 771
 772                 $text = $this->doMagicLinks( $text );
 773                 $text = $this->doTableStuff( $text );
 774                 $text = $this->formatHeadings( $text, $isMain );
 775
 776                 wfProfileOut( $fname );
 777                 return $text;
 778         }
 779
 780         /**
 781          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 782          * magic external links.
 783          *
 784          * @access private
 785          */
 786         function &doMagicLinks( &$text ) {
 787                 $text = $this->magicISBN( $text );
 788                 $text = $this->magicRFC( $text, 'RFC ', 'rfcurl' );
 789                 $text = $this->magicRFC( $text, 'PMID ', 'pubmedurl' );
 790                 return $text;
 791         }
 792
 793         /**
 794          * Parse ^^ tokens and return html
 795          *
 796          * @access private
 797          */
 798         function doExponent( $text ) {
 799                 $fname = 'Parser::doExponent';
 800                 wfProfileIn( $fname );
 801                 $text = preg_replace('/\^\^(.*)\^\^/','<small><sup>\\1</sup></small>', $text);
 802                 wfProfileOut( $fname );
 803                 return $text;
 804         }
 805
 806         /**
 807          * Parse headers and return html
 808          *
 809          * @access private
 810          */
 811         function doHeadings( $text ) {
 812                 $fname = 'Parser::doHeadings';
 813                 wfProfileIn( $fname );
 814                 for ( $i = 6; $i >= 1; --$i ) {
 815                         $h = substr( '======', 0, $i );
 816                         $text = preg_replace( "/^{$h}(.+){$h}(\\s|$)/m",
 817                           "<h{$i}>\\1</h{$i}>\\2", $text );
 818                 }
 819                 wfProfileOut( $fname );
 820                 return $text;
 821         }
 822
 823         /**
 824          * Replace single quotes with HTML markup
 825          * @access private
 826          * @return string the altered text
 827          */
 828         function doAllQuotes( $text ) {
 829                 $fname = 'Parser::doAllQuotes';
 830                 wfProfileIn( $fname );
 831                 $outtext = '';
 832                 $lines = explode( "\n", $text );
 833                 foreach ( $lines as $line ) {
 834                         $outtext .= $this->doQuotes ( $line ) . "\n";
 835                 }
 836                 $outtext = substr($outtext, 0,-1);
 837                 wfProfileOut( $fname );
 838                 return $outtext;
 839         }
 840
 841         /**
 842          * Helper function for doAllQuotes()
 843          * @access private
 844          */
 845         function doQuotes( $text ) {
 846                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
 847                 if ( count( $arr ) == 1 )
 848                         return $text;
 849                 else
 850                 {
 851                         # First, do some preliminary work. This may shift some apostrophes from
 852                         # being mark-up to being text. It also counts the number of occurrences
 853                         # of bold and italics mark-ups.
 854                         $i = 0;
 855                         $numbold = 0;
 856                         $numitalics = 0;
 857                         foreach ( $arr as $r )
 858                         {
 859                                 if ( ( $i % 2 ) == 1 )
 860                                 {
 861                                         # If there are ever four apostrophes, assume the first is supposed to
 862                                         # be text, and the remaining three constitute mark-up for bold text.
 863                                         if ( strlen( $arr[$i] ) == 4 )
 864                                         {
 865                                                 $arr[$i-1] .= "'";
 866                                                 $arr[$i] = "'''";
 867                                         }
 868                                         # If there are more than 5 apostrophes in a row, assume they're all
 869                                         # text except for the last 5.
 870                                         else if ( strlen( $arr[$i] ) > 5 )
 871                                         {
 872                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
 873                                                 $arr[$i] = "'''''";
 874                                         }
 875                                         # Count the number of occurrences of bold and italics mark-ups.
 876                                         # We are not counting sequences of five apostrophes.
 877                                         if ( strlen( $arr[$i] ) == 2 ) $numitalics++;  else
 878                                         if ( strlen( $arr[$i] ) == 3 ) $numbold++;     else
 879                                         if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
 880                                 }
 881                                 $i++;
 882                         }
 883
 884                         # If there is an odd number of both bold and italics, it is likely
 885                         # that one of the bold ones was meant to be an apostrophe followed
 886                         # by italics. Which one we cannot know for certain, but it is more
 887                         # likely to be one that has a single-letter word before it.
 888                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
 889                         {
 890                                 $i = 0;
 891                                 $firstsingleletterword = -1;
 892                                 $firstmultiletterword = -1;
 893                                 $firstspace = -1;
 894                                 foreach ( $arr as $r )
 895                                 {
 896                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
 897                                         {
 898                                                 $x1 = substr ($arr[$i-1], -1);
 899                                                 $x2 = substr ($arr[$i-1], -2, 1);
 900                                                 if ($x1 == ' ') {
 901                                                         if ($firstspace == -1) $firstspace = $i;
 902                                                 } else if ($x2 == ' ') {
 903                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
 904                                                 } else {
 905                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
 906                                                 }
 907                                         }
 908                                         $i++;
 909                                 }
 910
 911                                 # If there is a single-letter word, use it!
 912                                 if ($firstsingleletterword > -1)
 913                                 {
 914                                         $arr [ $firstsingleletterword ] = "''";
 915                                         $arr [ $firstsingleletterword-1 ] .= "'";
 916                                 }
 917                                 # If not, but there's a multi-letter word, use that one.
 918                                 else if ($firstmultiletterword > -1)
 919                                 {
 920                                         $arr [ $firstmultiletterword ] = "''";
 921                                         $arr [ $firstmultiletterword-1 ] .= "'";
 922                                 }
 923                                 # ... otherwise use the first one that has neither.
 924                                 # (notice that it is possible for all three to be -1 if, for example,
 925                                 # there is only one pentuple-apostrophe in the line)
 926                                 else if ($firstspace > -1)
 927                                 {
 928                                         $arr [ $firstspace ] = "''";
 929                                         $arr [ $firstspace-1 ] .= "'";
 930                                 }
 931                         }
 932
 933                         # Now let's actually convert our apostrophic mush to HTML!
 934                         $output = '';
 935                         $buffer = '';
 936                         $state = '';
 937                         $i = 0;
 938                         foreach ($arr as $r)
 939                         {
 940                                 if (($i % 2) == 0)
 941                                 {
 942                                         if ($state == 'both')
 943                                                 $buffer .= $r;
 944                                         else
 945                                                 $output .= $r;
 946                                 }
 947                                 else
 948                                 {
 949                                         if (strlen ($r) == 2)
 950                                         {
 951                                                 if ($state == 'i')
 952                                                 { $output .= '</i>'; $state = ''; }
 953                                                 else if ($state == 'bi')
 954                                                 { $output .= '</i>'; $state = 'b'; }
 955                                                 else if ($state == 'ib')
 956                                                 { $output .= '</b></i><b>'; $state = 'b'; }
 957                                                 else if ($state == 'both')
 958                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
 959                                                 else # $state can be 'b' or ''
 960                                                 { $output .= '<i>'; $state .= 'i'; }
 961                                         }
 962                                         else if (strlen ($r) == 3)
 963                                         {
 964                                                 if ($state == 'b')
 965                                                 { $output .= '</b>'; $state = ''; }
 966                                                 else if ($state == 'bi')
 967                                                 { $output .= '</i></b><i>'; $state = 'i'; }
 968                                                 else if ($state == 'ib')
 969                                                 { $output .= '</b>'; $state = 'i'; }
 970                                                 else if ($state == 'both')
 971                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
 972                                                 else # $state can be 'i' or ''
 973                                                 { $output .= '<b>'; $state .= 'b'; }
 974                                         }
 975                                         else if (strlen ($r) == 5)
 976                                         {
 977                                                 if ($state == 'b')
 978                                                 { $output .= '</b><i>'; $state = 'i'; }
 979                                                 else if ($state == 'i')
 980                                                 { $output .= '</i><b>'; $state = 'b'; }
 981                                                 else if ($state == 'bi')
 982                                                 { $output .= '</i></b>'; $state = ''; }
 983                                                 else if ($state == 'ib')
 984                                                 { $output .= '</b></i>'; $state = ''; }
 985                                                 else if ($state == 'both')
 986                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
 987                                                 else # ($state == '')
 988                                                 { $buffer = ''; $state = 'both'; }
 989                                         }
 990                                 }
 991                                 $i++;
 992                         }
 993                         # Now close all remaining tags.  Notice that the order is important.
 994                         if ($state == 'b' || $state == 'ib')
 995                                 $output .= '</b>';
 996                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
 997                                 $output .= '</i>';
 998                         if ($state == 'bi')
 999                                 $output .= '</b>';
1000                         if ($state == 'both')
1001                                 $output .= '<b><i>'.$buffer.'</i></b>';
1002                         return $output;
1003                 }
1004         }
1005
1006         /**
1007          * Replace external links
1008          *
1009          * Note: this is all very hackish and the order of execution matters a lot.
1010          * Make sure to run maintenance/parserTests.php if you change this code.
1011          *
1012          * @access private
1013          */
1014         function replaceExternalLinks( $text ) {
1015                 global $wgContLang;
1016                 $fname = 'Parser::replaceExternalLinks';
1017                 wfProfileIn( $fname );
1018
1019                 $sk =& $this->mOptions->getSkin();
1020
1021                 $bits = preg_split( EXT_LINK_BRACKETED, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1022
1023                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
1024
1025                 $i = 0;
1026                 while ( $i<count( $bits ) ) {
1027                         $url = $bits[$i++];
1028                         $protocol = $bits[$i++];
1029                         $text = $bits[$i++];
1030                         $trail = $bits[$i++];
1031
1032                         # The characters '<' and '>' (which were escaped by
1033                         # removeHTMLtags()) should not be included in
1034                         # URLs, per RFC 2396.
1035                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1036                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1037                                 $url = substr($url, 0, $m2[0][1]);
1038                         }
1039
1040                         # If the link text is an image URL, replace it with an <img> tag
1041                         # This happened by accident in the original parser, but some people used it extensively
1042                         $img = $this->maybeMakeExternalImage( $text );
1043                         if ( $img !== false ) {
1044                                 $text = $img;
1045                         }
1046
1047                         $dtrail = '';
1048
1049                         # Set linktype for CSS - if URL==text, link is essentially free
1050                         $linktype = ($text == $url) ? 'free' : 'text';
1051
1052                         # No link text, e.g. [http://domain.tld/some.link]
1053                         if ( $text == '' ) {
1054                                 # Autonumber if allowed
1055                                 if ( strpos( HTTP_PROTOCOLS, $protocol ) !== false ) {
1056                                         $text = '[' . ++$this->mAutonumber . ']';
1057                                         $linktype = 'autonumber';
1058                                 } else {
1059                                         # Otherwise just use the URL
1060                                         $text = htmlspecialchars( $url );
1061                                         $linktype = 'free';
1062                                 }
1063                         } else {
1064                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1065                                 # Check for trail
1066                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1067                         }
1068
1069                         $text = $wgContLang->markNoConversion($text);
1070
1071                         # Replace &amp; from obsolete syntax with &.
1072                         # All HTML entities will be escaped by makeExternalLink()
1073                         # or maybeMakeExternalImage()
1074                         $url = str_replace( '&amp;', '&', $url );
1075
1076                         # Process the trail (i.e. everything after this link up until start of the next link),
1077                         # replacing any non-bracketed links
1078                         $trail = $this->replaceFreeExternalLinks( $trail );
1079
1080
1081                         # Use the encoded URL
1082                         # This means that users can paste URLs directly into the text
1083                         # Funny characters like &ouml; aren't valid in URLs anyway
1084                         # This was changed in August 2004
1085                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype ) . $dtrail . $trail;
1086                 }
1087
1088                 wfProfileOut( $fname );
1089                 return $s;
1090         }
1091
1092         /**
1093          * Replace anything that looks like a URL with a link
1094          * @access private
1095          */
1096         function replaceFreeExternalLinks( $text ) {
1097                 global $wgContLang;
1098                 $fname = 'Parser::replaceFreeExternalLinks';
1099                 wfProfileIn( $fname );
1100
1101                 $bits = preg_split( '/(\b(?:'.URL_PROTOCOLS.'):)/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1102                 $s = array_shift( $bits );
1103                 $i = 0;
1104
1105                 $sk =& $this->mOptions->getSkin();
1106
1107                 while ( $i < count( $bits ) ){
1108                         $protocol = $bits[$i++];
1109                         $remainder = $bits[$i++];
1110
1111                         if ( preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1112                                 # Found some characters after the protocol that look promising
1113                                 $url = $protocol . $m[1];
1114                                 $trail = $m[2];
1115
1116                                 # The characters '<' and '>' (which were escaped by
1117                                 # removeHTMLtags()) should not be included in
1118                                 # URLs, per RFC 2396.
1119                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1120                                         $trail = substr($url, $m2[0][1]) . $trail;
1121                                         $url = substr($url, 0, $m2[0][1]);
1122                                 }
1123
1124                                 # Move trailing punctuation to $trail
1125                                 $sep = ',;\.:!?';
1126                                 # If there is no left bracket, then consider right brackets fair game too
1127                                 if ( strpos( $url, '(' ) === false ) {
1128                                         $sep .= ')';
1129                                 }
1130
1131                                 $numSepChars = strspn( strrev( $url ), $sep );
1132                                 if ( $numSepChars ) {
1133                                         $trail = substr( $url, -$numSepChars ) . $trail;
1134                                         $url = substr( $url, 0, -$numSepChars );
1135                                 }
1136
1137                                 # Replace &amp; from obsolete syntax with &.
1138                                 # All HTML entities will be escaped by makeExternalLink()
1139                                 # or maybeMakeExternalImage()
1140                                 $url = str_replace( '&amp;', '&', $url );
1141
1142                                 # Is this an external image?
1143                                 $text = $this->maybeMakeExternalImage( $url );
1144                                 if ( $text === false ) {
1145                                         # Not an image, make a link
1146                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free' );
1147                                 }
1148                                 $s .= $text . $trail;
1149                         } else {
1150                                 $s .= $protocol . $remainder;
1151                         }
1152                 }
1153                 wfProfileOut();
1154                 return $s;
1155         }
1156
1157         /**
1158          * make an image if it's allowed
1159          * @access private
1160          */
1161         function maybeMakeExternalImage( $url ) {
1162                 $sk =& $this->mOptions->getSkin();
1163                 $text = false;
1164                 if ( $this->mOptions->getAllowExternalImages() ) {
1165                         if ( preg_match( EXT_IMAGE_REGEX, $url ) ) {
1166                                 # Image found
1167                                 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
1168                         }
1169                 }
1170                 return $text;
1171         }
1172
1173         /**
1174          * Process [[ ]] wikilinks
1175          *
1176          * @access private
1177          */
1178         function replaceInternalLinks( $s ) {
1179                 global $wgContLang, $wgLinkCache;
1180                 static $fname = 'Parser::replaceInternalLinks' ;
1181
1182                 wfProfileIn( $fname );
1183
1184                 wfProfileIn( $fname.'-setup' );
1185                 static $tc = FALSE;
1186                 # the % is needed to support urlencoded titles as well
1187                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1188
1189                 $sk =& $this->mOptions->getSkin();
1190
1191                 #split the entire text string on occurences of [[
1192                 $a = explode( '[[', ' ' . $s );
1193                 #get the first element (all text up to first [[), and remove the space we added
1194                 $s = array_shift( $a );
1195                 $s = substr( $s, 1 );
1196
1197                 # Match a link having the form [[namespace:link|alternate]]trail
1198                 static $e1 = FALSE;
1199                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1200                 # Match cases where there is no "]]", which might still be images
1201                 static $e1_img = FALSE;
1202                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1203                 # Match the end of a line for a word that's not followed by whitespace,
1204                 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1205                 static $e2 = '/^(.*?)([a-zA-Z\x80-\xff]+)$/sD';
1206
1207                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1208
1209                 if( is_null( $this->mTitle ) ) {
1210                         wfDebugDieBacktrace( 'nooo' );
1211                 }
1212                 $nottalk = !$this->mTitle->isTalkPage();
1213
1214                 if ( $useLinkPrefixExtension ) {
1215                         if ( preg_match( $e2, $s, $m ) ) {
1216                                 $first_prefix = $m[2];
1217                                 $s = $m[1];
1218                         } else {
1219                                 $first_prefix = false;
1220                         }
1221                 } else {
1222                         $prefix = '';
1223                 }
1224
1225                 $selflink = $this->mTitle->getPrefixedText();
1226                 wfProfileOut( $fname.'-setup' );
1227
1228                 $checkVariantLink = sizeof($wgContLang->getVariants())>1;
1229                 $useSubpages = $this->areSubpagesAllowed();
1230
1231                 # Loop for each link
1232                 for ($k = 0; isset( $a[$k] ); $k++) {
1233                         $line = $a[$k];
1234                         if ( $useLinkPrefixExtension ) {
1235                                 wfProfileIn( $fname.'-prefixhandling' );
1236                                 if ( preg_match( $e2, $s, $m ) ) {
1237                                         $prefix = $m[2];
1238                                         $s = $m[1];
1239                                 } else {
1240                                         $prefix='';
1241                                 }
1242                                 # first link
1243                                 if($first_prefix) {
1244                                         $prefix = $first_prefix;
1245                                         $first_prefix = false;
1246                                 }
1247                                 wfProfileOut( $fname.'-prefixhandling' );
1248                         }
1249
1250                         $might_be_img = false;
1251
1252                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1253                                 $text = $m[2];
1254                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1255                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1256                                 # the real problem is with the $e1 regex
1257                                 # See bug 1300.
1258                                 #
1259                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1260                                 # and no image is in sight. See bug 2095.
1261                                 #
1262                                 if( $text !== '' && preg_match( "/^\](.*)/s", $m[3], $n ) ) {
1263                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1264                                         $m[3] = $n[1];
1265                                 }
1266                                 # fix up urlencoded title texts
1267                                 if(preg_match('/%/', $m[1] )) $m[1] = urldecode($m[1]);
1268                                 $trail = $m[3];
1269                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1270                                 $might_be_img = true;
1271                                 $text = $m[2];
1272                                 if(preg_match('/%/', $m[1] )) $m[1] = urldecode($m[1]);
1273                                 $trail = "";
1274                         } else { # Invalid form; output directly
1275                                 $s .= $prefix . '[[' . $line ;
1276                                 continue;
1277                         }
1278
1279                         # Don't allow internal links to pages containing
1280                         # PROTO: where PROTO is a valid URL protocol; these
1281                         # should be external links.
1282                         if (preg_match('/^(\b(?:'.URL_PROTOCOLS.'):)/', $m[1])) {
1283                                 $s .= $prefix . '[[' . $line ;
1284                                 continue;
1285                         }
1286
1287                         # Make subpage if necessary
1288                         if( $useSubpages ) {
1289                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1290                         } else {
1291                                 $link = $m[1];
1292                         }
1293
1294                         $noforce = (substr($m[1], 0, 1) != ':');
1295                         if (!$noforce) {
1296                                 # Strip off leading ':'
1297                                 $link = substr($link, 1);
1298                         }
1299
1300                         $nt =& Title::newFromText( $this->unstripNoWiki($link, $this->mStripState) );
1301                         if( !$nt ) {
1302                                 $s .= $prefix . '[[' . $line;
1303                                 continue;
1304                         }
1305
1306                         #check other language variants of the link
1307                         #if the article does not exist
1308                         if( $checkVariantLink
1309                             && $nt->getArticleID() == 0 ) {
1310                                 $wgContLang->findVariantLink($link, $nt);
1311                         }
1312
1313                         $ns = $nt->getNamespace();
1314                         $iw = $nt->getInterWiki();
1315
1316                         if ($might_be_img) { # if this is actually an invalid link
1317                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1318                                         $found = false;
1319                                         while (isset ($a[$k+1]) ) {
1320                                                 #look at the next 'line' to see if we can close it there
1321                                                 $next_line =  array_shift(array_splice( $a, $k + 1, 1) );
1322                                                 if( preg_match("/^(.*?]].*?)]](.*)$/sD", $next_line, $m) ) {
1323                                                 # the first ]] closes the inner link, the second the image
1324                                                         $found = true;
1325                                                         $text .= '[[' . $m[1];
1326                                                         $trail = $m[2];
1327                                                         break;
1328                                                 } elseif( preg_match("/^.*?]].*$/sD", $next_line, $m) ) {
1329                                                         #if there's exactly one ]] that's fine, we'll keep looking
1330                                                         $text .= '[[' . $m[0];
1331                                                 } else {
1332                                                         #if $next_line is invalid too, we need look no further
1333                                                         $text .= '[[' . $next_line;
1334                                                         break;
1335                                                 }
1336                                         }
1337                                         if ( !$found ) {
1338                                                 # we couldn't find the end of this imageLink, so output it raw
1339                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1340                                                 $text = $this->replaceInternalLinks($text);
1341                                                 $s .= $prefix . '[[' . $link . '|' . $text;
1342                                                 # note: no $trail, because without an end, there *is* no trail
1343                                                 continue;
1344                                         }
1345                                 } else { #it's not an image, so output it raw
1346                                         $s .= $prefix . '[[' . $link . '|' . $text;
1347                                         # note: no $trail, because without an end, there *is* no trail
1348                                         continue;
1349                                 }
1350                         }
1351
1352                         $wasblank = ( '' == $text );
1353                         if( $wasblank ) $text = $link;
1354
1355
1356                         # Link not escaped by : , create the various objects
1357                         if( $noforce ) {
1358
1359                                 # Interwikis
1360                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1361                                         array_push( $this->mOutput->mLanguageLinks, $nt->getFullText() );
1362                                         $s = rtrim($s . "\n");
1363                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1364                                         continue;
1365                                 }
1366
1367                                 if ( $ns == NS_IMAGE ) {
1368                                         wfProfileIn( "$fname-image" );
1369                                         if ( !wfIsBadImage( $nt->getDBkey() ) ) {
1370                                                 # recursively parse links inside the image caption
1371                                                 # actually, this will parse them in any other parameters, too,
1372                                                 # but it might be hard to fix that, and it doesn't matter ATM
1373                                                 $text = $this->replaceExternalLinks($text);
1374                                                 $text = $this->replaceInternalLinks($text);
1375
1376                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1377                                                 $s .= $prefix . str_replace('http://', 'http-noparse://', $this->makeImage( $nt, $text ) ) . $trail;
1378                                                 $wgLinkCache->addImageLinkObj( $nt );
1379
1380                                                 wfProfileOut( "$fname-image" );
1381                                                 continue;
1382                                         }
1383                                         wfProfileOut( "$fname-image" );
1384
1385                                 }
1386
1387                                 if ( $ns == NS_CATEGORY ) {
1388                                         wfProfileIn( "$fname-category" );
1389                                         $t = $wgContLang->convert($nt->getText());
1390                                         $s = rtrim($s . "\n"); # bug 87
1391
1392                                         $wgLinkCache->suspend(); # Don't save in links/brokenlinks
1393                                         $t = $sk->makeLinkObj( $nt, $t, '', '' , $prefix );
1394                                         $wgLinkCache->resume();
1395
1396                                         if ( $wasblank ) {
1397                                                 if ( $this->mTitle->getNamespace() == NS_CATEGORY ) {
1398                                                         $sortkey = $this->mTitle->getText();
1399                                                 } else {
1400                                                         $sortkey = $this->mTitle->getPrefixedText();
1401                                                 }
1402                                         } else {
1403                                                 $sortkey = $text;
1404                                         }
1405                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1406                                         $wgLinkCache->addCategoryLinkObj( $nt, $sortkey );
1407                                         $this->mOutput->addCategoryLink( $t );
1408
1409                                         /**
1410                                          * Strip the whitespace Category links produce, see bug 87
1411                                          * @todo We might want to use trim($tmp, "\n") here.
1412                                          */
1413                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1414
1415                                         wfProfileOut( "$fname-category" );
1416                                         continue;
1417                                 }
1418                         }
1419
1420                         if( ( $nt->getPrefixedText() === $selflink ) &&
1421                             ( $nt->getFragment() === '' ) ) {
1422                                 # Self-links are handled specially; generally de-link and change to bold.
1423                                 $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1424                                 continue;
1425                         }
1426
1427                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1428                         if( $ns == NS_MEDIA ) {
1429                                 $s .= $prefix . $sk->makeMediaLinkObj( $nt, $text, true ) . $trail;
1430                                 $wgLinkCache->addImageLinkObj( $nt );
1431                                 continue;
1432                         } elseif( $ns == NS_SPECIAL ) {
1433                                 $s .= $prefix . $sk->makeKnownLinkObj( $nt, $text, '', $trail );
1434                                 continue;
1435                         }
1436                         if( $nt->isLocal() && $nt->isAlwaysKnown() ) {
1437                                 /**
1438                                  * Skip lookups for special pages and self-links.
1439                                  * External interwiki links are not included here because
1440                                  * the HTTP urls would break output in the next parse step;
1441                                  * they will have placeholders kept.
1442                                  */
1443                                 $s .= $sk->makeKnownLinkObj( $nt, $text, '', $trail, $prefix );
1444                         } else {
1445                                 /**
1446                                  * Add a link placeholder
1447                                  * Later, this will be replaced by a real link, after the existence or
1448                                  * non-existence of all the links is known
1449                                  */
1450                                 $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1451                         }
1452                 }
1453                 wfProfileOut( $fname );
1454                 return $s;
1455         }
1456
1457         /**
1458          * Make a link placeholder. The text returned can be later resolved to a real link with
1459          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1460          * parsing of interwiki links, and secondly to allow all extistence checks and
1461          * article length checks (for stub links) to be bundled into a single query.
1462          *
1463          */
1464         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1465                 if ( ! is_object($nt) ) {
1466                         # Fail gracefully
1467                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1468                 } else {
1469                         # Separate the link trail from the rest of the link
1470                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1471
1472                         if ( $nt->isExternal() ) {
1473                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1474                                 $this->mInterwikiLinkHolders['titles'][] =& $nt;
1475                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1476                         } else {
1477                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1478                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1479                                 $this->mLinkHolders['queries'][] = $query;
1480                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1481                                 $this->mLinkHolders['titles'][] =& $nt;
1482
1483                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1484                         }
1485                 }
1486                 return $retVal;
1487         }
1488
1489         /**
1490          * Return true if subpage links should be expanded on this page.
1491          * @return bool
1492          */
1493         function areSubpagesAllowed() {
1494                 # Some namespaces don't allow subpages
1495                 global $wgNamespacesWithSubpages;
1496                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1497         }
1498
1499         /**
1500          * Handle link to subpage if necessary
1501          * @param string $target the source of the link
1502          * @param string &$text the link text, modified as necessary
1503          * @return string the full name of the link
1504          * @access private
1505          */
1506         function maybeDoSubpageLink($target, &$text) {
1507                 # Valid link forms:
1508                 # Foobar -- normal
1509                 # :Foobar -- override special treatment of prefix (images, language links)
1510                 # /Foobar -- convert to CurrentPage/Foobar
1511                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1512                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1513                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1514
1515                 $fname = 'Parser::maybeDoSubpageLink';
1516                 wfProfileIn( $fname );
1517                 $ret = $target; # default return value is no change
1518
1519                 # Some namespaces don't allow subpages,
1520                 # so only perform processing if subpages are allowed
1521                 if( $this->areSubpagesAllowed() ) {
1522                         # Look at the first character
1523                         if( $target != '' && $target{0} == '/' ) {
1524                                 # / at end means we don't want the slash to be shown
1525                                 if( substr( $target, -1, 1 ) == '/' ) {
1526                                         $target = substr( $target, 1, -1 );
1527                                         $noslash = $target;
1528                                 } else {
1529                                         $noslash = substr( $target, 1 );
1530                                 }
1531
1532                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash);
1533                                 if( '' === $text ) {
1534                                         $text = $target;
1535                                 } # this might be changed for ugliness reasons
1536                         } else {
1537                                 # check for .. subpage backlinks
1538                                 $dotdotcount = 0;
1539                                 $nodotdot = $target;
1540                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1541                                         ++$dotdotcount;
1542                                         $nodotdot = substr( $nodotdot, 3 );
1543                                 }
1544                                 if($dotdotcount > 0) {
1545                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1546                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1547                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1548                                                 # / at the end means don't show full path
1549                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1550                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1551                                                         if( '' === $text ) {
1552                                                                 $text = $nodotdot;
1553                                                         }
1554                                                 }
1555                                                 $nodotdot = trim( $nodotdot );
1556                                                 if( $nodotdot != '' ) {
1557                                                         $ret .= '/' . $nodotdot;
1558                                                 }
1559                                         }
1560                                 }
1561                         }
1562                 }
1563
1564                 wfProfileOut( $fname );
1565                 return $ret;
1566         }
1567
1568         /**#@+
1569          * Used by doBlockLevels()
1570          * @access private
1571          */
1572         /* private */ function closeParagraph() {
1573                 $result = '';
1574                 if ( '' != $this->mLastSection ) {
1575                         $result = '</' . $this->mLastSection  . ">\n";
1576                 }
1577                 $this->mInPre = false;
1578                 $this->mLastSection = '';
1579                 return $result;
1580         }
1581         # getCommon() returns the length of the longest common substring
1582         # of both arguments, starting at the beginning of both.
1583         #
1584         /* private */ function getCommon( $st1, $st2 ) {
1585                 $fl = strlen( $st1 );
1586                 $shorter = strlen( $st2 );
1587                 if ( $fl < $shorter ) { $shorter = $fl; }
1588
1589                 for ( $i = 0; $i < $shorter; ++$i ) {
1590                         if ( $st1{$i} != $st2{$i} ) { break; }
1591                 }
1592                 return $i;
1593         }
1594         # These next three functions open, continue, and close the list
1595         # element appropriate to the prefix character passed into them.
1596         #
1597         /* private */ function openList( $char ) {
1598                 $result = $this->closeParagraph();
1599
1600                 if ( '*' == $char ) { $result .= '<ul><li>'; }
1601                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
1602                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
1603                 else if ( ';' == $char ) {
1604                         $result .= '<dl><dt>';
1605                         $this->mDTopen = true;
1606                 }
1607                 else { $result = '<!-- ERR 1 -->'; }
1608
1609                 return $result;
1610         }
1611
1612         /* private */ function nextItem( $char ) {
1613                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
1614                 else if ( ':' == $char || ';' == $char ) {
1615                         $close = '</dd>';
1616                         if ( $this->mDTopen ) { $close = '</dt>'; }
1617                         if ( ';' == $char ) {
1618                                 $this->mDTopen = true;
1619                                 return $close . '<dt>';
1620                         } else {
1621                                 $this->mDTopen = false;
1622                                 return $close . '<dd>';
1623                         }
1624                 }
1625                 return '<!-- ERR 2 -->';
1626         }
1627
1628         /* private */ function closeList( $char ) {
1629                 if ( '*' == $char ) { $text = '</li></ul>'; }
1630                 else if ( '#' == $char ) { $text = '</li></ol>'; }
1631                 else if ( ':' == $char ) {
1632                         if ( $this->mDTopen ) {
1633                                 $this->mDTopen = false;
1634                                 $text = '</dt></dl>';
1635                         } else {
1636                                 $text = '</dd></dl>';
1637                         }
1638                 }
1639                 else {  return '<!-- ERR 3 -->'; }
1640                 return $text."\n";
1641         }
1642         /**#@-*/
1643
1644         /**
1645          * Make lists from lines starting with ':', '*', '#', etc.
1646          *
1647          * @access private
1648          * @return string the lists rendered as HTML
1649          */
1650         function doBlockLevels( $text, $linestart ) {
1651                 $fname = 'Parser::doBlockLevels';
1652                 wfProfileIn( $fname );
1653
1654                 # Parsing through the text line by line.  The main thing
1655                 # happening here is handling of block-level elements p, pre,
1656                 # and making lists from lines starting with * # : etc.
1657                 #
1658                 $textLines = explode( "\n", $text );
1659
1660                 $lastPrefix = $output = '';
1661                 $this->mDTopen = $inBlockElem = false;
1662                 $prefixLength = 0;
1663                 $paragraphStack = false;
1664
1665                 if ( !$linestart ) {
1666                         $output .= array_shift( $textLines );
1667                 }
1668                 foreach ( $textLines as $oLine ) {
1669                         $lastPrefixLength = strlen( $lastPrefix );
1670                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
1671                         $preOpenMatch = preg_match('/<pre/i', $oLine );
1672                         if ( !$this->mInPre ) {
1673                                 # Multiple prefixes may abut each other for nested lists.
1674                                 $prefixLength = strspn( $oLine, '*#:;' );
1675                                 $pref = substr( $oLine, 0, $prefixLength );
1676
1677                                 # eh?
1678                                 $pref2 = str_replace( ';', ':', $pref );
1679                                 $t = substr( $oLine, $prefixLength );
1680                                 $this->mInPre = !empty($preOpenMatch);
1681                         } else {
1682                                 # Don't interpret any other prefixes in preformatted text
1683                                 $prefixLength = 0;
1684                                 $pref = $pref2 = '';
1685                                 $t = $oLine;
1686                         }
1687
1688                         # List generation
1689                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
1690                                 # Same as the last item, so no need to deal with nesting or opening stuff
1691                                 $output .= $this->nextItem( substr( $pref, -1 ) );
1692                                 $paragraphStack = false;
1693
1694                                 if ( substr( $pref, -1 ) == ';') {
1695                                         # The one nasty exception: definition lists work like this:
1696                                         # ; title : definition text
1697                                         # So we check for : in the remainder text to split up the
1698                                         # title and definition, without b0rking links.
1699                                         $term = $t2 = '';
1700                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
1701                                                 $t = $t2;
1702                                                 $output .= $term . $this->nextItem( ':' );
1703                                         }
1704                                 }
1705                         } elseif( $prefixLength || $lastPrefixLength ) {
1706                                 # Either open or close a level...
1707                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
1708                                 $paragraphStack = false;
1709
1710                                 while( $commonPrefixLength < $lastPrefixLength ) {
1711                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
1712                                         --$lastPrefixLength;
1713                                 }
1714                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
1715                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
1716                                 }
1717                                 while ( $prefixLength > $commonPrefixLength ) {
1718                                         $char = substr( $pref, $commonPrefixLength, 1 );
1719                                         $output .= $this->openList( $char );
1720
1721                                         if ( ';' == $char ) {
1722                                                 # FIXME: This is dupe of code above
1723                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
1724                                                         $t = $t2;
1725                                                         $output .= $term . $this->nextItem( ':' );
1726                                                 }
1727                                         }
1728                                         ++$commonPrefixLength;
1729                                 }
1730                                 $lastPrefix = $pref2;
1731                         }
1732                         if( 0 == $prefixLength ) {
1733                                 wfProfileIn( "$fname-paragraph" );
1734                                 # No prefix (not in list)--go to paragraph mode
1735                                 $uniq_prefix = UNIQ_PREFIX;
1736                                 // XXX: use a stack for nestable elements like span, table and div
1737                                 $openmatch = preg_match('/(<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
1738                                 $closematch = preg_match(
1739                                         '/(<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
1740                                         '<td|<th|<div|<\\/div|<hr|<\\/pre|<\\/p|'.$uniq_prefix.'-pre|<\\/li|<\\/ul)/iS', $t );
1741                                 if ( $openmatch or $closematch ) {
1742                                         $paragraphStack = false;
1743                                         $output .= $this->closeParagraph();
1744                                         if($preOpenMatch and !$preCloseMatch) {
1745                                                 $this->mInPre = true;
1746                                         }
1747                                         if ( $closematch ) {
1748                                                 $inBlockElem = false;
1749                                         } else {
1750                                                 $inBlockElem = true;
1751                                         }
1752                                 } else if ( !$inBlockElem && !$this->mInPre ) {
1753                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
1754                                                 // pre
1755                                                 if ($this->mLastSection != 'pre') {
1756                                                         $paragraphStack = false;
1757                                                         $output .= $this->closeParagraph().'<pre>';
1758                                                         $this->mLastSection = 'pre';
1759                                                 }
1760                                                 $t = substr( $t, 1 );
1761                                         } else {
1762                                                 // paragraph
1763                                                 if ( '' == trim($t) ) {
1764                                                         if ( $paragraphStack ) {
1765                                                                 $output .= $paragraphStack.'<br />';
1766                                                                 $paragraphStack = false;
1767                                                                 $this->mLastSection = 'p';
1768                                                         } else {
1769                                                                 if ($this->mLastSection != 'p' ) {
1770                                                                         $output .= $this->closeParagraph();
1771                                                                         $this->mLastSection = '';
1772                                                                         $paragraphStack = '<p>';
1773                                                                 } else {
1774                                                                         $paragraphStack = '</p><p>';
1775                                                                 }
1776                                                         }
1777                                                 } else {
1778                                                         if ( $paragraphStack ) {
1779                                                                 $output .= $paragraphStack;
1780                                                                 $paragraphStack = false;
1781                                                                 $this->mLastSection = 'p';
1782                                                         } else if ($this->mLastSection != 'p') {
1783                                                                 $output .= $this->closeParagraph().'<p>';
1784                                                                 $this->mLastSection = 'p';
1785                                                         }
1786                                                 }
1787                                         }
1788                                 }
1789                                 wfProfileOut( "$fname-paragraph" );
1790                         }
1791                         if ($paragraphStack === false) {
1792                                 $output .= $t."\n";
1793                         }
1794                 }
1795                 while ( $prefixLength ) {
1796                         $output .= $this->closeList( $pref2{$prefixLength-1} );
1797                         --$prefixLength;
1798                 }
1799                 if ( '' != $this->mLastSection ) {
1800                         $output .= '</' . $this->mLastSection . '>';
1801                         $this->mLastSection = '';
1802                 }
1803
1804                 wfProfileOut( $fname );
1805                 return $output;
1806         }
1807
1808         /**
1809          * Split up a string on ':', ignoring any occurences inside
1810          * <a>..</a> or <span>...</span>
1811          * @param string $str the string to split
1812          * @param string &$before set to everything before the ':'
1813          * @param string &$after set to everything after the ':'
1814          * return string the position of the ':', or false if none found
1815          */
1816         function findColonNoLinks($str, &$before, &$after) {
1817                 # I wonder if we should make this count all tags, not just <a>
1818                 # and <span>. That would prevent us from matching a ':' that
1819                 # comes in the middle of italics other such formatting....
1820                 # -- Wil
1821                 $fname = 'Parser::findColonNoLinks';
1822                 wfProfileIn( $fname );
1823                 $pos = 0;
1824                 do {
1825                         $colon = strpos($str, ':', $pos);
1826
1827                         if ($colon !== false) {
1828                                 $before = substr($str, 0, $colon);
1829                                 $after = substr($str, $colon + 1);
1830
1831                                 # Skip any ':' within <a> or <span> pairs
1832                                 $a = substr_count($before, '<a');
1833                                 $s = substr_count($before, '<span');
1834                                 $ca = substr_count($before, '</a>');
1835                                 $cs = substr_count($before, '</span>');
1836
1837                                 if ($a <= $ca and $s <= $cs) {
1838                                         # Tags are balanced before ':'; ok
1839                                         break;
1840                                 }
1841                                 $pos = $colon + 1;
1842                         }
1843                 } while ($colon !== false);
1844                 wfProfileOut( $fname );
1845                 return $colon;
1846         }
1847
1848         /**
1849          * Return value of a magic variable (like PAGENAME)
1850          *
1851          * @access private
1852          */
1853         function getVariableValue( $index ) {
1854                 global $wgContLang, $wgSitename, $wgServer, $wgArticle;
1855
1856                 /**
1857                  * Some of these require message or data lookups and can be
1858                  * expensive to check many times.
1859                  */
1860                 static $varCache = array();
1861                 if( isset( $varCache[$index] ) ) return $varCache[$index];
1862
1863                 switch ( $index ) {
1864                         case MAG_CURRENTMONTH:
1865                                 return $varCache[$index] = $wgContLang->formatNum( date( 'm' ) );
1866                         case MAG_CURRENTMONTHNAME:
1867                                 return $varCache[$index] = $wgContLang->getMonthName( date('n') );
1868                         case MAG_CURRENTMONTHNAMEGEN:
1869                                 return $varCache[$index] = $wgContLang->getMonthNameGen( date('n') );
1870                         case MAG_CURRENTMONTHABBREV:
1871                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( date('n') );
1872                         case MAG_CURRENTDAY:
1873                                 return $varCache[$index] = $wgContLang->formatNum( date('j') );
1874                         case MAG_PAGENAME:
1875                                 return $this->mTitle->getText();
1876                         case MAG_PAGENAMEE:
1877                                 return $this->mTitle->getPartialURL();
1878                         case MAG_REVISIONID:
1879                                 return $wgArticle->getRevIdFetched();
1880                         case MAG_NAMESPACE:
1881                                 # return Namespace::getCanonicalName($this->mTitle->getNamespace());
1882                                 return $wgContLang->getNsText($this->mTitle->getNamespace()); # Patch by Dori
1883                         case MAG_CURRENTDAYNAME:
1884                                 return $varCache[$index] = $wgContLang->getWeekdayName( date('w')+1 );
1885                         case MAG_CURRENTYEAR:
1886                                 return $varCache[$index] = $wgContLang->formatNum( date( 'Y' ), true );
1887                         case MAG_CURRENTTIME:
1888                                 return $varCache[$index] = $wgContLang->time( wfTimestampNow(), false );
1889                         case MAG_CURRENTWEEK:
1890                                 return $varCache[$index] = $wgContLang->formatNum( date('W') );
1891                         case MAG_CURRENTDOW:
1892                                 return $varCache[$index] = $wgContLang->formatNum( date('w') );
1893                         case MAG_NUMBEROFARTICLES:
1894                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfArticles() );
1895                         case MAG_SITENAME:
1896                                 return $wgSitename;
1897                         case MAG_SERVER:
1898                                 return $wgServer;
1899                         default:
1900                                 return NULL;
1901                 }
1902         }
1903
1904         /**
1905          * initialise the magic variables (like CURRENTMONTHNAME)
1906          *
1907          * @access private
1908          */
1909         function initialiseVariables() {
1910                 $fname = 'Parser::initialiseVariables';
1911                 wfProfileIn( $fname );
1912                 global $wgVariableIDs;
1913                 $this->mVariables = array();
1914                 foreach ( $wgVariableIDs as $id ) {
1915                         $mw =& MagicWord::get( $id );
1916                         $mw->addToArray( $this->mVariables, $id );
1917                 }
1918                 wfProfileOut( $fname );
1919         }
1920
1921         /**
1922          * Replace magic variables, templates, and template arguments
1923          * with the appropriate text. Templates are substituted recursively,
1924          * taking care to avoid infinite loops.
1925          *
1926          * Note that the substitution depends on value of $mOutputType:
1927          *  OT_WIKI: only {{subst:}} templates
1928          *  OT_MSG: only magic variables
1929          *  OT_HTML: all templates and magic variables
1930          *
1931          * @param string $tex The text to transform
1932          * @param array $args Key-value pairs representing template parameters to substitute
1933          * @access private
1934          */
1935         function replaceVariables( $text, $args = array() ) {
1936
1937                 # Prevent too big inclusions
1938                 if( strlen( $text ) > MAX_INCLUDE_SIZE ) {
1939                         return $text;
1940                 }
1941
1942                 $fname = 'Parser::replaceVariables';
1943                 wfProfileIn( $fname );
1944
1945                 $titleChars = Title::legalChars();
1946
1947                 # This function is called recursively. To keep track of arguments we need a stack:
1948                 array_push( $this->mArgStack, $args );
1949
1950                 # Variable substitution
1951                 $text = preg_replace_callback( "/{{([$titleChars]*?)}}/", array( &$this, 'variableSubstitution' ), $text );
1952
1953                 if ( $this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI ) {
1954                         # Argument substitution
1955                         $text = preg_replace_callback( "/{{{([$titleChars]*?)}}}/", array( &$this, 'argSubstitution' ), $text );
1956                 }
1957                 # Template substitution
1958                 $regex = '/(\\n|{)?{{(['.$titleChars.']*)(\\|.*?|)}}/s';
1959                 $text = preg_replace_callback( $regex, array( &$this, 'braceSubstitution' ), $text );
1960
1961                 array_pop( $this->mArgStack );
1962
1963                 wfProfileOut( $fname );
1964                 return $text;
1965         }
1966
1967         /**
1968          * Replace magic variables
1969          * @access private
1970          */
1971         function variableSubstitution( $matches ) {
1972                 $fname = 'parser::variableSubstitution';
1973                 $varname = $matches[1];
1974                 wfProfileIn( $fname );
1975                 if ( !$this->mVariables ) {
1976                         $this->initialiseVariables();
1977                 }
1978                 $skip = false;
1979                 if ( $this->mOutputType == OT_WIKI ) {
1980                         # Do only magic variables prefixed by SUBST
1981                         $mwSubst =& MagicWord::get( MAG_SUBST );
1982                         if (!$mwSubst->matchStartAndRemove( $varname ))
1983                                 $skip = true;
1984                         # Note that if we don't substitute the variable below,
1985                         # we don't remove the {{subst:}} magic word, in case
1986                         # it is a template rather than a magic variable.
1987                 }
1988                 if ( !$skip && array_key_exists( $varname, $this->mVariables ) ) {
1989                         $id = $this->mVariables[$varname];
1990                         $text = $this->getVariableValue( $id );
1991                         $this->mOutput->mContainsOldMagic = true;
1992                 } else {
1993                         $text = $matches[0];
1994                 }
1995                 wfProfileOut( $fname );
1996                 return $text;
1997         }
1998
1999         # Split template arguments
2000         function getTemplateArgs( $argsString ) {
2001                 if ( $argsString === '' ) {
2002                         return array();
2003                 }
2004
2005                 $args = explode( '|', substr( $argsString, 1 ) );
2006
2007                 # If any of the arguments contains a '[[' but no ']]', it needs to be
2008                 # merged with the next arg because the '|' character between belongs
2009                 # to the link syntax and not the template parameter syntax.
2010                 $argc = count($args);
2011
2012                 for ( $i = 0; $i < $argc-1; $i++ ) {
2013                         if ( substr_count ( $args[$i], '[[' ) != substr_count ( $args[$i], ']]' ) ) {
2014                                 $args[$i] .= '|'.$args[$i+1];
2015                                 array_splice($args, $i+1, 1);
2016                                 $i--;
2017                                 $argc--;
2018                         }
2019                 }
2020
2021                 return $args;
2022         }
2023
2024         /**
2025          * Return the text of a template, after recursively
2026          * replacing any variables or templates within the template.
2027          *
2028          * @param array $matches The parts of the template
2029          *  $matches[1]: the title, i.e. the part before the |
2030          *  $matches[2]: the parameters (including a leading |), if  any
2031          * @return string the text of the template
2032          * @access private
2033          */
2034         function braceSubstitution( $matches ) {
2035                 global $wgLinkCache, $wgContLang;
2036                 $fname = 'Parser::braceSubstitution';
2037                 wfProfileIn( $fname );
2038
2039                 $found = false;
2040                 $nowiki = false;
2041                 $noparse = false;
2042
2043                 $title = NULL;
2044
2045                 # Need to know if the template comes at the start of a line,
2046                 # to treat the beginning of the template like the beginning
2047                 # of a line for tables and block-level elements.
2048                 $linestart = $matches[1];
2049
2050                 # $part1 is the bit before the first |, and must contain only title characters
2051                 # $args is a list of arguments, starting from index 0, not including $part1
2052
2053                 $part1 = $matches[2];
2054                 # If the third subpattern matched anything, it will start with |
2055
2056                 $args = $this->getTemplateArgs($matches[3]);
2057                 $argc = count( $args );
2058
2059                 # Don't parse {{{}}} because that's only for template arguments
2060                 if ( $linestart === '{' ) {
2061                         $text = $matches[0];
2062                         $found = true;
2063                         $noparse = true;
2064                 }
2065
2066                 # SUBST
2067                 if ( !$found ) {
2068                         $mwSubst =& MagicWord::get( MAG_SUBST );
2069                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor ($this->mOutputType == OT_WIKI) ) {
2070                                 # One of two possibilities is true:
2071                                 # 1) Found SUBST but not in the PST phase
2072                                 # 2) Didn't find SUBST and in the PST phase
2073                                 # In either case, return without further processing
2074                                 $text = $matches[0];
2075                                 $found = true;
2076                                 $noparse = true;
2077                         }
2078                 }
2079
2080                 # MSG, MSGNW and INT
2081                 if ( !$found ) {
2082                         # Check for MSGNW:
2083                         $mwMsgnw =& MagicWord::get( MAG_MSGNW );
2084                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2085                                 $nowiki = true;
2086                         }
2087
2088                         # int: is the wikitext equivalent of wfMsg()
2089                         $mwInt =& MagicWord::get( MAG_INT );
2090                         if ( $mwInt->matchStartAndRemove( $part1 ) ) {
2091                                 if ( $this->incrementIncludeCount( 'int:'.$part1 ) ) {
2092                                         $text = $linestart . wfMsgReal( $part1, $args, true );
2093                                         $found = true;
2094                                 }
2095                         }
2096
2097                         # msg: is the wikitext equivalent of wfMsgForContent()
2098                         $mwMsg =& MagicWord::get( MAG_MSG );
2099                         if ( $mwMsg->matchStartAndRemove( $part1 ) ) {
2100                                 if ( $this->incrementIncludeCount( 'msg:'.$part1 ) ) {
2101                                          $text = $linestart . wfMsgReal( $part1, $args, true, true );
2102                                          $found = true;
2103                                 }
2104                         }
2105                 }
2106
2107                 # NS
2108                 if ( !$found ) {
2109                         # Check for NS: (namespace expansion)
2110                         $mwNs = MagicWord::get( MAG_NS );
2111                         if ( $mwNs->matchStartAndRemove( $part1 ) ) {
2112                                 if ( intval( $part1 ) ) {
2113                                         $text = $linestart . $wgContLang->getNsText( intval( $part1 ) );
2114                                         $found = true;
2115                                 } else {
2116                                         $index = Namespace::getCanonicalIndex( strtolower( $part1 ) );
2117                                         if ( !is_null( $index ) ) {
2118                                                 $text = $linestart . $wgContLang->getNsText( $index );
2119                                                 $found = true;
2120                                         }
2121                                 }
2122                         }
2123                 }
2124
2125                 # LOCALURL and LOCALURLE
2126                 if ( !$found ) {
2127                         $mwLocal = MagicWord::get( MAG_LOCALURL );
2128                         $mwLocalE = MagicWord::get( MAG_LOCALURLE );
2129
2130                         if ( $mwLocal->matchStartAndRemove( $part1 ) ) {
2131                                 $func = 'getLocalURL';
2132                         } elseif ( $mwLocalE->matchStartAndRemove( $part1 ) ) {
2133                                 $func = 'escapeLocalURL';
2134                         } else {
2135                                 $func = '';
2136                         }
2137
2138                         if ( $func !== '' ) {
2139                                 $title = Title::newFromText( $part1 );
2140                                 if ( !is_null( $title ) ) {
2141                                         if ( $argc > 0 ) {
2142                                                 $text = $linestart . $title->$func( $args[0] );
2143                                         } else {
2144                                                 $text = $linestart . $title->$func();
2145                                         }
2146                                         $found = true;
2147                                 }
2148                         }
2149                 }
2150
2151                 # GRAMMAR
2152                 if ( !$found && $argc == 1 ) {
2153                         $mwGrammar =& MagicWord::get( MAG_GRAMMAR );
2154                         if ( $mwGrammar->matchStartAndRemove( $part1 ) ) {
2155                                 $text = $linestart . $wgContLang->convertGrammar( $args[0], $part1 );
2156                                 $found = true;
2157                         }
2158                 }
2159
2160                 # Template table test
2161
2162                 # Did we encounter this template already? If yes, it is in the cache
2163                 # and we need to check for loops.
2164                 if ( !$found && isset( $this->mTemplates[$part1] ) ) {
2165                         $found = true;
2166
2167                         # Infinite loop test
2168                         if ( isset( $this->mTemplatePath[$part1] ) ) {
2169                                 $noparse = true;
2170                                 $found = true;
2171                                 $text = $linestart .
2172                                         "\{\{$part1}}" .
2173                                         '<!-- WARNING: template loop detected -->';
2174                                 wfDebug( "$fname: template loop broken at '$part1'\n" );
2175                         } else {
2176                                 # set $text to cached message.
2177                                 $text = $linestart . $this->mTemplates[$part1];
2178                         }
2179                 }
2180
2181                 # Load from database
2182                 $replaceHeadings = false;
2183                 $isHTML = false;
2184                 $lastPathLevel = $this->mTemplatePath;
2185                 if ( !$found ) {
2186                         $ns = NS_TEMPLATE;
2187                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage='' );
2188                         if ($subpage !== '') {
2189                                 $ns = $this->mTitle->getNamespace();
2190                         }
2191                         $title = Title::newFromText( $part1, $ns );
2192                         if ( !is_null( $title ) && !$title->isExternal() ) {
2193                                 # Check for excessive inclusion
2194                                 $dbk = $title->getPrefixedDBkey();
2195                                 if ( $this->incrementIncludeCount( $dbk ) ) {
2196                                         if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() ) {
2197                                                 # Capture special page output
2198                                                 $text = SpecialPage::capturePath( $title );
2199                                                 if ( $text && !is_object( $text ) ) {
2200                                                         $found = true;
2201                                                         $noparse = true;
2202                                                         $isHTML = true;
2203                                                         $this->mOutput->setCacheTime( -1 );
2204                                                 }
2205                                         } else {
2206                                                 $article = new Article( $title );
2207                                                 $articleContent = $article->getContentWithoutUsingSoManyDamnGlobals();
2208                                                 if ( $articleContent !== false ) {
2209                                                         $found = true;
2210                                                         $text = $articleContent;
2211                                                         $replaceHeadings = true;
2212                                                 }
2213                                         }
2214                                 }
2215
2216                                 # If the title is valid but undisplayable, make a link to it
2217                                 if ( $this->mOutputType == OT_HTML && !$found ) {
2218                                         $text = '[['.$title->getPrefixedText().']]';
2219                                         $found = true;
2220                                 }
2221
2222                                 # Template cache array insertion
2223                                 if( $found ) {
2224                                         $this->mTemplates[$part1] = $text;
2225                                         $text = $linestart . $text;
2226                                 }
2227                         }
2228                 }
2229
2230                 # Recursive parsing, escaping and link table handling
2231                 # Only for HTML output
2232                 if ( $nowiki && $found && $this->mOutputType == OT_HTML ) {
2233                         $text = wfEscapeWikiText( $text );
2234                 } elseif ( ($this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI) && $found && !$noparse) {
2235                         # Clean up argument array
2236                         $assocArgs = array();
2237                         $index = 1;
2238                         foreach( $args as $arg ) {
2239                                 $eqpos = strpos( $arg, '=' );
2240                                 if ( $eqpos === false ) {
2241                                         $assocArgs[$index++] = $arg;
2242                                 } else {
2243                                         $name = trim( substr( $arg, 0, $eqpos ) );
2244                                         $value = trim( substr( $arg, $eqpos+1 ) );
2245                                         if ( $value === false ) {
2246                                                 $value = '';
2247                                         }
2248                                         if ( $name !== false ) {
2249                                                 $assocArgs[$name] = $value;
2250                                         }
2251                                 }
2252                         }
2253
2254                         # Add a new element to the templace recursion path
2255                         $this->mTemplatePath[$part1] = 1;
2256
2257                         if( $this->mOutputType == OT_HTML ) {
2258                                 $text = $this->strip( $text, $this->mStripState );
2259                                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'replaceVariables' ), $assocArgs );
2260                         }
2261                         $text = $this->replaceVariables( $text, $assocArgs );
2262
2263                         # Resume the link cache and register the inclusion as a link
2264                         if ( $this->mOutputType == OT_HTML && !is_null( $title ) ) {
2265                                 $wgLinkCache->addLinkObj( $title );
2266                         }
2267
2268                         # If the template begins with a table or block-level
2269                         # element, it should be treated as beginning a new line.
2270                         if ($linestart !== '\n' && preg_match('/^({\\||:|;|#|\*)/', $text)) {
2271                                 $text = "\n" . $text;
2272                         }
2273                 }
2274                 # Prune lower levels off the recursion check path
2275                 $this->mTemplatePath = $lastPathLevel;
2276
2277                 if ( !$found ) {
2278                         wfProfileOut( $fname );
2279                         return $matches[0];
2280                 } else {
2281                         if ( $isHTML ) {
2282                                 # Replace raw HTML by a placeholder
2283                                 # Add a blank line preceding, to prevent it from mucking up
2284                                 # immediately preceding headings
2285                                 $text = "\n\n" . $this->insertStripItem( $text, $this->mStripState );
2286                         } else {
2287                                 # replace ==section headers==
2288                                 # XXX this needs to go away once we have a better parser.
2289                                 if ( $this->mOutputType != OT_WIKI && $replaceHeadings ) {
2290                                         if( !is_null( $title ) )
2291                                                 $encodedname = base64_encode($title->getPrefixedDBkey());
2292                                         else
2293                                                 $encodedname = base64_encode("");
2294                                         $m = preg_split('/(^={1,6}.*?={1,6}\s*?$)/m', $text, -1,
2295                                                 PREG_SPLIT_DELIM_CAPTURE);
2296                                         $text = '';
2297                                         $nsec = 0;
2298                                         for( $i = 0; $i < count($m); $i += 2 ) {
2299                                                 $text .= $m[$i];
2300                                                 if (!isset($m[$i + 1]) || $m[$i + 1] == "") continue;
2301                                                 $hl = $m[$i + 1];
2302                                                 if( strstr($hl, "<!--MWTEMPLATESECTION") ) {
2303                                                         $text .= $hl;
2304                                                         continue;
2305                                                 }
2306                                                 preg_match('/^(={1,6})(.*?)(={1,6})\s*?$/m', $hl, $m2);
2307                                                 $text .= $m2[1] . $m2[2] . "<!--MWTEMPLATESECTION="
2308                                                         . $encodedname . "&" . base64_encode("$nsec") . "-->" . $m2[3];
2309
2310                                                 $nsec++;
2311                                         }
2312                                 }
2313                         }
2314                 }
2315
2316                 # Prune lower levels off the recursion check path
2317                 $this->mTemplatePath = $lastPathLevel;
2318
2319                 if ( !$found ) {
2320                         wfProfileOut( $fname );
2321                         return $matches[0];
2322                 } else {
2323                         wfProfileOut( $fname );
2324                         return $text;
2325                 }
2326         }
2327
2328         /**
2329          * Triple brace replacement -- used for template arguments
2330          * @access private
2331          */
2332         function argSubstitution( $matches ) {
2333                 $arg = trim( $matches[1] );
2334                 $text = $matches[0];
2335                 $inputArgs = end( $this->mArgStack );
2336
2337                 if ( array_key_exists( $arg, $inputArgs ) ) {
2338                         $text = $inputArgs[$arg];
2339                 }
2340
2341                 return $text;
2342         }
2343
2344         /**
2345          * Returns true if the function is allowed to include this entity
2346          * @access private
2347          */
2348         function incrementIncludeCount( $dbk ) {
2349                 if ( !array_key_exists( $dbk, $this->mIncludeCount ) ) {
2350                         $this->mIncludeCount[$dbk] = 0;
2351                 }
2352                 if ( ++$this->mIncludeCount[$dbk] <= MAX_INCLUDE_REPEAT ) {
2353                         return true;
2354                 } else {
2355                         return false;
2356                 }
2357         }
2358
2359         /**
2360          * This function accomplishes several tasks:
2361          * 1) Auto-number headings if that option is enabled
2362          * 2) Add an [edit] link to sections for logged in users who have enabled the option
2363          * 3) Add a Table of contents on the top for users who have enabled the option
2364          * 4) Auto-anchor headings
2365          *
2366          * It loops through all headlines, collects the necessary data, then splits up the
2367          * string and re-inserts the newly formatted headlines.
2368          *
2369          * @param string $text
2370          * @param boolean $isMain
2371          * @access private
2372          */
2373         function formatHeadings( $text, $isMain=true ) {
2374                 global $wgMaxTocLevel, $wgContLang, $wgLinkHolders, $wgInterwikiLinkHolders;
2375
2376                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
2377                 $doShowToc = true;
2378                 $forceTocHere = false;
2379                 if( !$this->mTitle->userCanEdit() ) {
2380                         $showEditLink = 0;
2381                 } else {
2382                         $showEditLink = $this->mOptions->getEditSection();
2383                 }
2384
2385                 # Inhibit editsection links if requested in the page
2386                 $esw =& MagicWord::get( MAG_NOEDITSECTION );
2387                 if( $esw->matchAndRemove( $text ) ) {
2388                         $showEditLink = 0;
2389                 }
2390                 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
2391                 # do not add TOC
2392                 $mw =& MagicWord::get( MAG_NOTOC );
2393                 if( $mw->matchAndRemove( $text ) ) {
2394                         $doShowToc = false;
2395                 }
2396
2397                 # Get all headlines for numbering them and adding funky stuff like [edit]
2398                 # links - this is for later, but we need the number of headlines right now
2399                 $numMatches = preg_match_all( '/<H([1-6])(.*?'.'>)(.*?)<\/H[1-6] *>/i', $text, $matches );
2400
2401                 # if there are fewer than 4 headlines in the article, do not show TOC
2402                 if( $numMatches < 4 ) {
2403                         $doShowToc = false;
2404                 }
2405
2406                 # if the string __TOC__ (not case-sensitive) occurs in the HTML,
2407                 # override above conditions and always show TOC at that place
2408
2409                 $mw =& MagicWord::get( MAG_TOC );
2410                 if($mw->match( $text ) ) {
2411                         $doShowToc = true;
2412                         $forceTocHere = true;
2413                 } else {
2414                         # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
2415                         # override above conditions and always show TOC above first header
2416                         $mw =& MagicWord::get( MAG_FORCETOC );
2417                         if ($mw->matchAndRemove( $text ) ) {
2418                                 $doShowToc = true;
2419                         }
2420                 }
2421
2422                 # Never ever show TOC if no headers
2423                 if( $numMatches < 1 ) {
2424                         $doShowToc = false;
2425                 }
2426
2427                 # We need this to perform operations on the HTML
2428                 $sk =& $this->mOptions->getSkin();
2429
2430                 # headline counter
2431                 $headlineCount = 0;
2432                 $sectionCount = 0; # headlineCount excluding template sections
2433
2434                 # Ugh .. the TOC should have neat indentation levels which can be
2435                 # passed to the skin functions. These are determined here
2436                 $toc = '';
2437                 $full = '';
2438                 $head = array();
2439                 $sublevelCount = array();
2440                 $levelCount = array();
2441                 $toclevel = 0;
2442                 $level = 0;
2443                 $prevlevel = 0;
2444                 $toclevel = 0;
2445                 $prevtoclevel = 0;
2446
2447                 foreach( $matches[3] as $headline ) {
2448                         $istemplate = 0;
2449                         $templatetitle = '';
2450                         $templatesection = 0;
2451                         $numbering = '';
2452
2453                         if (preg_match("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", $headline, $mat)) {
2454                                 $istemplate = 1;
2455                                 $templatetitle = base64_decode($mat[1]);
2456                                 $templatesection = 1 + (int)base64_decode($mat[2]);
2457                                 $headline = preg_replace("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", "", $headline);
2458                         }
2459
2460                         if( $toclevel ) {
2461                                 $prevlevel = $level;
2462                                 $prevtoclevel = $toclevel;
2463                         }
2464                         $level = $matches[1][$headlineCount];
2465
2466                         if( $doNumberHeadings || $doShowToc ) {
2467
2468                                 if ( $level > $prevlevel ) {
2469                                         # Increase TOC level
2470                                         $toclevel++;
2471                                         $sublevelCount[$toclevel] = 0;
2472                                         $toc .= $sk->tocIndent();
2473                                 }
2474                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
2475                                         # Decrease TOC level, find level to jump to
2476
2477                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
2478                                                 # Can only go down to level 1
2479                                                 $toclevel = 1;
2480                                         } else {
2481                                                 for ($i = $toclevel; $i > 0; $i--) {
2482                                                         if ( $levelCount[$i] == $level ) {
2483                                                                 # Found last matching level
2484                                                                 $toclevel = $i;
2485                                                                 break;
2486                                                         }
2487                                                         elseif ( $levelCount[$i] < $level ) {
2488                                                                 # Found first matching level below current level
2489                                                                 $toclevel = $i + 1;
2490                                                                 break;
2491                                                         }
2492                                                 }
2493                                         }
2494
2495                                         $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
2496                                 }
2497                                 else {
2498                                         # No change in level, end TOC line
2499                                         $toc .= $sk->tocLineEnd();
2500                                 }
2501
2502                                 $levelCount[$toclevel] = $level;
2503
2504                                 # count number of headlines for each level
2505                                 @$sublevelCount[$toclevel]++;
2506                                 $dot = 0;
2507                                 for( $i = 1; $i <= $toclevel; $i++ ) {
2508                                         if( !empty( $sublevelCount[$i] ) ) {
2509                                                 if( $dot ) {
2510                                                         $numbering .= '.';
2511                                                 }
2512                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
2513                                                 $dot = 1;
2514                                         }
2515                                 }
2516                         }
2517
2518                         # The canonized header is a version of the header text safe to use for links
2519                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
2520                         $canonized_headline = $this->unstrip( $headline, $this->mStripState );
2521                         $canonized_headline = $this->unstripNoWiki( $canonized_headline, $this->mStripState );
2522
2523                         # Remove link placeholders by the link text.
2524                         #     <!--LINK number-->
2525                         # turns into
2526                         #     link text with suffix
2527                         $canonized_headline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
2528                                                             "\$this->mLinkHolders['texts'][\$1]",
2529                                                             $canonized_headline );
2530                         $canonized_headline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
2531                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
2532                                                             $canonized_headline );
2533
2534                         # strip out HTML
2535                         $canonized_headline = preg_replace( '/<.*?' . '>/','',$canonized_headline );
2536                         $tocline = trim( $canonized_headline );
2537                         $canonized_headline = urlencode( Sanitizer::decodeCharReferences( str_replace(' ', '_', $tocline) ) );
2538                         $replacearray = array(
2539                                 '%3A' => ':',
2540                                 '%' => '.'
2541                         );
2542                         $canonized_headline = str_replace(array_keys($replacearray),array_values($replacearray),$canonized_headline);
2543                         $refers[$headlineCount] = $canonized_headline;
2544
2545                         # count how many in assoc. array so we can track dupes in anchors
2546                         @$refers[$canonized_headline]++;
2547                         $refcount[$headlineCount]=$refers[$canonized_headline];
2548
2549                         # Don't number the heading if it is the only one (looks silly)
2550                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
2551                                 # the two are different if the line contains a link
2552                                 $headline=$numbering . ' ' . $headline;
2553                         }
2554
2555                         # Create the anchor for linking from the TOC to the section
2556                         $anchor = $canonized_headline;
2557                         if($refcount[$headlineCount] > 1 ) {
2558                                 $anchor .= '_' . $refcount[$headlineCount];
2559                         }
2560                         if( $doShowToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
2561                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
2562                         }
2563                         if( $showEditLink && ( !$istemplate || $templatetitle !== "" ) ) {
2564                                 if ( empty( $head[$headlineCount] ) ) {
2565                                         $head[$headlineCount] = '';
2566                                 }
2567                                 if( $istemplate )
2568                                         $head[$headlineCount] .= $sk->editSectionLinkForOther($templatetitle, $templatesection);
2569                                 else
2570                                         $head[$headlineCount] .= $sk->editSectionLink($this->mTitle, $sectionCount+1);
2571                         }
2572
2573                         # give headline the correct <h#> tag
2574                         @$head[$headlineCount] .= "<a name=\"$anchor\"></a><h".$level.$matches[2][$headlineCount] .$headline.'</h'.$level.'>';
2575
2576                         $headlineCount++;
2577                         if( !$istemplate )
2578                                 $sectionCount++;
2579                 }
2580
2581                 if( $doShowToc ) {
2582                         $toc .= $sk->tocUnindent( $toclevel - 1 );
2583                         $toc = $sk->tocList( $toc );
2584                 }
2585
2586                 # split up and insert constructed headlines
2587
2588                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
2589                 $i = 0;
2590
2591                 foreach( $blocks as $block ) {
2592                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
2593                                 # This is the [edit] link that appears for the top block of text when
2594                                 # section editing is enabled
2595
2596                                 # Disabled because it broke block formatting
2597                                 # For example, a bullet point in the top line
2598                                 # $full .= $sk->editSectionLink(0);
2599                         }
2600                         $full .= $block;
2601                         if( $doShowToc && !$i && $isMain && !$forceTocHere) {
2602                         # Top anchor now in skin
2603                                 $full = $full.$toc;
2604                         }
2605
2606                         if( !empty( $head[$i] ) ) {
2607                                 $full .= $head[$i];
2608                         }
2609                         $i++;
2610                 }
2611                 if($forceTocHere) {
2612                         $mw =& MagicWord::get( MAG_TOC );
2613                         return $mw->replace( $toc, $full );
2614                 } else {
2615                         return $full;
2616                 }
2617         }
2618
2619         /**
2620          * Return an HTML link for the "ISBN 123456" text
2621          * @access private
2622          */
2623         function magicISBN( $text ) {
2624                 $fname = 'Parser::magicISBN';
2625                 wfProfileIn( $fname );
2626
2627                 $a = split( 'ISBN ', ' '.$text );
2628                 if ( count ( $a ) < 2 ) {
2629                         wfProfileOut( $fname );
2630                         return $text;
2631                 }
2632                 $text = substr( array_shift( $a ), 1);
2633                 $valid = '0123456789-Xx';
2634
2635                 foreach ( $a as $x ) {
2636                         $isbn = $blank = '' ;
2637                         while ( ' ' == $x{0} ) {
2638                                 $blank .= ' ';
2639                                 $x = substr( $x, 1 );
2640                         }
2641                         if ( $x == '' ) { # blank isbn
2642                                 $text .= "ISBN $blank";
2643                                 continue;
2644                         }
2645                         while ( strstr( $valid, $x{0} ) != false ) {
2646                                 $isbn .= $x{0};
2647                                 $x = substr( $x, 1 );
2648                         }
2649                         $num = str_replace( '-', '', $isbn );
2650                         $num = str_replace( ' ', '', $num );
2651                         $num = str_replace( 'x', 'X', $num );
2652
2653                         if ( '' == $num ) {
2654                                 $text .= "ISBN $blank$x";
2655                         } else {
2656                                 $titleObj = Title::makeTitle( NS_SPECIAL, 'Booksources' );
2657                                 $text .= '<a href="' .
2658                                 $titleObj->escapeLocalUrl( 'isbn='.$num ) .
2659                                         "\" class=\"internal\">ISBN $isbn</a>";
2660                                 $text .= $x;
2661                         }
2662                 }
2663                 wfProfileOut( $fname );
2664                 return $text;
2665         }
2666
2667         /**
2668          * Return an HTML link for the "RFC 1234" text
2669          *
2670          * @access private
2671          * @param string $text     Text to be processed
2672          * @param string $keyword  Magic keyword to use (default RFC)
2673          * @param string $urlmsg   Interface message to use (default rfcurl)
2674          * @return string
2675          */
2676         function magicRFC( $text, $keyword='RFC ', $urlmsg='rfcurl'  ) {
2677
2678                 $valid = '0123456789';
2679                 $internal = false;
2680
2681                 $a = split( $keyword, ' '.$text );
2682                 if ( count ( $a ) < 2 ) {
2683                         return $text;
2684                 }
2685                 $text = substr( array_shift( $a ), 1);
2686
2687                 /* Check if keyword is preceed by [[.
2688                  * This test is made here cause of the array_shift above
2689                  * that prevent the test to be done in the foreach.
2690                  */
2691                 if ( substr( $text, -2 ) == '[[' ) {
2692                         $internal = true;
2693                 }
2694
2695                 foreach ( $a as $x ) {
2696                         /* token might be empty if we have RFC RFC 1234 */
2697                         if ( $x=='' ) {
2698                                 $text.=$keyword;
2699                                 continue;
2700                                 }
2701
2702                         $id = $blank = '' ;
2703
2704                         /** remove and save whitespaces in $blank */
2705                         while ( $x{0} == ' ' ) {
2706                                 $blank .= ' ';
2707                                 $x = substr( $x, 1 );
2708                         }
2709
2710                         /** remove and save the rfc number in $id */
2711                         while ( strstr( $valid, $x{0} ) != false ) {
2712                                 $id .= $x{0};
2713                                 $x = substr( $x, 1 );
2714                         }
2715
2716                         if ( $id == '' ) {
2717                                 /* call back stripped spaces*/
2718                                 $text .= $keyword.$blank.$x;
2719                         } elseif( $internal ) {
2720                                 /* normal link */
2721                                 $text .= $keyword.$id.$x;
2722                         } else {
2723                                 /* build the external link*/
2724                                 $url = wfMsg( $urlmsg, $id);
2725                                 $sk =& $this->mOptions->getSkin();
2726                                 $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
2727                                 $text .= "<a href='{$url}'{$la}>{$keyword}{$id}</a>{$x}";
2728                         }
2729
2730                         /* Check if the next RFC keyword is preceed by [[ */
2731                         $internal = ( substr($x,-2) == '[[' );
2732                 }
2733                 return $text;
2734         }
2735
2736         /**
2737          * Transform wiki markup when saving a page by doing \r\n -> \n
2738          * conversion, substitting signatures, {{subst:}} templates, etc.
2739          *
2740          * @param string $text the text to transform
2741          * @param Title &$title the Title object for the current article
2742          * @param User &$user the User object describing the current user
2743          * @param ParserOptions $options parsing options
2744          * @param bool $clearState whether to clear the parser state first
2745          * @return string the altered wiki markup
2746          * @access public
2747          */
2748         function preSaveTransform( $text, &$title, &$user, $options, $clearState = true ) {
2749                 $this->mOptions = $options;
2750                 $this->mTitle =& $title;
2751                 $this->mOutputType = OT_WIKI;
2752
2753                 if ( $clearState ) {
2754                         $this->clearState();
2755                 }
2756
2757                 $stripState = false;
2758                 $pairs = array(
2759                         "\r\n" => "\n",
2760                 );
2761                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
2762                 $text = $this->strip( $text, $stripState, false );
2763                 $text = $this->pstPass2( $text, $user );
2764                 $text = $this->unstrip( $text, $stripState );
2765                 $text = $this->unstripNoWiki( $text, $stripState );
2766                 return $text;
2767         }
2768
2769         /**
2770          * Pre-save transform helper function
2771          * @access private
2772          */
2773         function pstPass2( $text, &$user ) {
2774                 global $wgContLang, $wgLocaltimezone;
2775
2776                 # Variable replacement
2777                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
2778                 $text = $this->replaceVariables( $text );
2779
2780                 # Signatures
2781                 #
2782                 $n = $user->getName();
2783                 $k = $user->getOption( 'nickname' );
2784                 if ( '' == $k ) { $k = $n; }
2785                 if ( isset( $wgLocaltimezone ) ) {
2786                         $oldtz = getenv( 'TZ' );
2787                         putenv( 'TZ='.$wgLocaltimezone );
2788                 }
2789
2790                 /* Note: This is the timestamp saved as hardcoded wikitext to
2791                  * the database, we use $wgContLang here in order to give
2792                  * everyone the same signiture and use the default one rather
2793                  * than the one selected in each users preferences.
2794                  */
2795                 $d = $wgContLang->timeanddate( wfTimestampNow(), false, false) .
2796                   ' (' . date( 'T' ) . ')';
2797                 if ( isset( $wgLocaltimezone ) ) {
2798                         putenv( 'TZ='.$oldtz );
2799                 }
2800
2801                 if( $user->getOption( 'fancysig' ) ) {
2802                         $sigText = $k;
2803                 } else {
2804                         $sigText = '[[' . $wgContLang->getNsText( NS_USER ) . ":$n|$k]]";
2805                 }
2806                 $text = preg_replace( '/~~~~~/', $d, $text );
2807                 $text = preg_replace( '/~~~~/', "$sigText $d", $text );
2808                 $text = preg_replace( '/~~~/', $sigText, $text );
2809
2810                 # Context links: [[|name]] and [[name (context)|]]
2811                 #
2812                 $tc = "[&;%\\-,.\\(\\)' _0-9A-Za-z\\/:\\x80-\\xff]";
2813                 $np = "[&;%\\-,.' _0-9A-Za-z\\/:\\x80-\\xff]"; # No parens
2814                 $namespacechar = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
2815                 $conpat = "/^({$np}+) \\(({$tc}+)\\)$/";
2816
2817                 $p1 = "/\[\[({$np}+) \\(({$np}+)\\)\\|]]/";             # [[page (context)|]]
2818                 $p2 = "/\[\[\\|({$tc}+)]]/";                                    # [[|page]]
2819                 $p3 = "/\[\[(:*$namespacechar+):({$np}+)\\|]]/";                # [[namespace:page|]] and [[:namespace:page|]]
2820                 $p4 = "/\[\[(:*$namespacechar+):({$np}+) \\(({$np}+)\\)\\|]]/"; # [[ns:page (cont)|]] and [[:ns:page (cont)|]]
2821                 $context = '';
2822                 $t = $this->mTitle->getText();
2823                 if ( preg_match( $conpat, $t, $m ) ) {
2824                         $context = $m[2];
2825                 }
2826                 $text = preg_replace( $p4, '[[\\1:\\2 (\\3)|\\2]]', $text );
2827                 $text = preg_replace( $p1, '[[\\1 (\\2)|\\1]]', $text );
2828                 $text = preg_replace( $p3, '[[\\1:\\2|\\2]]', $text );
2829
2830                 if ( '' == $context ) {
2831                         $text = preg_replace( $p2, '[[\\1]]', $text );
2832                 } else {
2833                         $text = preg_replace( $p2, "[[\\1 ({$context})|\\1]]", $text );
2834                 }
2835
2836                 # Trim trailing whitespace
2837                 # MAG_END (__END__) tag allows for trailing
2838                 # whitespace to be deliberately included
2839                 $text = rtrim( $text );
2840                 $mw =& MagicWord::get( MAG_END );
2841                 $mw->matchAndRemove( $text );
2842
2843                 return $text;
2844         }
2845
2846         /**
2847          * Set up some variables which are usually set up in parse()
2848          * so that an external function can call some class members with confidence
2849          * @access public
2850          */
2851         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
2852                 $this->mTitle =& $title;
2853                 $this->mOptions = $options;
2854                 $this->mOutputType = $outputType;
2855                 if ( $clearState ) {
2856                         $this->clearState();
2857                 }
2858         }
2859
2860         /**
2861          * Transform a MediaWiki message by replacing magic variables.
2862          *
2863          * @param string $text the text to transform
2864          * @param ParserOptions $options  options
2865          * @return string the text with variables substituted
2866          * @access public
2867          */
2868         function transformMsg( $text, $options ) {
2869                 global $wgTitle;
2870                 static $executing = false;
2871
2872                 # Guard against infinite recursion
2873                 if ( $executing ) {
2874                         return $text;
2875                 }
2876                 $executing = true;
2877
2878                 $this->mTitle = $wgTitle;
2879                 $this->mOptions = $options;
2880                 $this->mOutputType = OT_MSG;
2881                 $this->clearState();
2882                 $text = $this->replaceVariables( $text );
2883
2884                 $executing = false;
2885                 return $text;
2886         }
2887
2888         /**
2889          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
2890          * Callback will be called with the text within
2891          * Transform and return the text within
2892          * @access public
2893          */
2894         function setHook( $tag, $callback ) {
2895                 $oldVal = @$this->mTagHooks[$tag];
2896                 $this->mTagHooks[$tag] = $callback;
2897                 return $oldVal;
2898         }
2899
2900         /**
2901          * Replace <!--LINK--> link placeholders with actual links, in the buffer
2902          * Placeholders created in Skin::makeLinkObj()
2903          * Returns an array of links found, indexed by PDBK:
2904          *  0 - broken
2905          *  1 - normal link
2906          *  2 - stub
2907          * $options is a bit field, RLH_FOR_UPDATE to select for update
2908          */
2909         function replaceLinkHolders( &$text, $options = 0 ) {
2910                 global $wgUser, $wgLinkCache;
2911                 global $wgOutputReplace;
2912
2913                 $fname = 'Parser::replaceLinkHolders';
2914                 wfProfileIn( $fname );
2915
2916                 $pdbks = array();
2917                 $colours = array();
2918                 $sk = $this->mOptions->getSkin();
2919
2920                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
2921                         wfProfileIn( $fname.'-check' );
2922                         $dbr =& wfGetDB( DB_SLAVE );
2923                         $page = $dbr->tableName( 'page' );
2924                         $threshold = $wgUser->getOption('stubthreshold');
2925
2926                         # Sort by namespace
2927                         asort( $this->mLinkHolders['namespaces'] );
2928
2929                         # Generate query
2930                         $query = false;
2931                         foreach ( $this->mLinkHolders['namespaces'] as $key => $val ) {
2932                                 # Make title object
2933                                 $title = $this->mLinkHolders['titles'][$key];
2934
2935                                 # Skip invalid entries.
2936                                 # Result will be ugly, but prevents crash.
2937                                 if ( is_null( $title ) ) {
2938                                         continue;
2939                                 }
2940                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
2941
2942                                 # Check if it's in the link cache already
2943                                 if ( $wgLinkCache->getGoodLinkID( $pdbk ) ) {
2944                                         $colours[$pdbk] = 1;
2945                                 } elseif ( $wgLinkCache->isBadLink( $pdbk ) ) {
2946                                         $colours[$pdbk] = 0;
2947                                 } else {
2948                                         # Not in the link cache, add it to the query
2949                                         if ( !isset( $current ) ) {
2950                                                 $current = $val;
2951                                                 $query =  "SELECT page_id, page_namespace, page_title";
2952                                                 if ( $threshold > 0 ) {
2953                                                         $query .= ', page_len, page_is_redirect';
2954                                                 }
2955                                                 $query .= " FROM $page WHERE (page_namespace=$val AND page_title IN(";
2956                                         } elseif ( $current != $val ) {
2957                                                 $current = $val;
2958                                                 $query .= ")) OR (page_namespace=$val AND page_title IN(";
2959                                         } else {
2960                                                 $query .= ', ';
2961                                         }
2962
2963                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
2964                                 }
2965                         }
2966                         if ( $query ) {
2967                                 $query .= '))';
2968                                 if ( $options & RLH_FOR_UPDATE ) {
2969                                         $query .= ' FOR UPDATE';
2970                                 }
2971
2972                                 $res = $dbr->query( $query, $fname );
2973
2974                                 # Fetch data and form into an associative array
2975                                 # non-existent = broken
2976                                 # 1 = known
2977                                 # 2 = stub
2978                                 while ( $s = $dbr->fetchObject($res) ) {
2979                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
2980                                         $pdbk = $title->getPrefixedDBkey();
2981                                         $wgLinkCache->addGoodLinkObj( $s->page_id, $title );
2982
2983                                         if ( $threshold >  0 ) {
2984                                                 $size = $s->page_len;
2985                                                 if ( $s->page_is_redirect || $s->page_namespace != 0 || $size >= $threshold ) {
2986                                                         $colours[$pdbk] = 1;
2987                                                 } else {
2988                                                         $colours[$pdbk] = 2;
2989                                                 }
2990                                         } else {
2991                                                 $colours[$pdbk] = 1;
2992                                         }
2993                                 }
2994                         }
2995                         wfProfileOut( $fname.'-check' );
2996
2997                         # Construct search and replace arrays
2998                         wfProfileIn( $fname.'-construct' );
2999                         $wgOutputReplace = array();
3000                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3001                                 $pdbk = $pdbks[$key];
3002                                 $searchkey = "<!--LINK $key-->";
3003                                 $title = $this->mLinkHolders['titles'][$key];
3004                                 if ( empty( $colours[$pdbk] ) ) {
3005                                         $wgLinkCache->addBadLinkObj( $title );
3006                                         $colours[$pdbk] = 0;
3007                                         $wgOutputReplace[$searchkey] = $sk->makeBrokenLinkObj( $title,
3008                                                                         $this->mLinkHolders['texts'][$key],
3009                                                                         $this->mLinkHolders['queries'][$key] );
3010                                 } elseif ( $colours[$pdbk] == 1 ) {
3011                                         $wgOutputReplace[$searchkey] = $sk->makeKnownLinkObj( $title,
3012                                                                         $this->mLinkHolders['texts'][$key],
3013                                                                         $this->mLinkHolders['queries'][$key] );
3014                                 } elseif ( $colours[$pdbk] == 2 ) {
3015                                         $wgOutputReplace[$searchkey] = $sk->makeStubLinkObj( $title,
3016                                                                         $this->mLinkHolders['texts'][$key],
3017                                                                         $this->mLinkHolders['queries'][$key] );
3018                                 }
3019                         }
3020                         wfProfileOut( $fname.'-construct' );
3021
3022                         # Do the thing
3023                         wfProfileIn( $fname.'-replace' );
3024
3025                         $text = preg_replace_callback(
3026                                 '/(<!--LINK .*?-->)/',
3027                                 "wfOutputReplaceMatches",
3028                                 $text);
3029
3030                         wfProfileOut( $fname.'-replace' );
3031                 }
3032
3033                 # Now process interwiki link holders
3034                 # This is quite a bit simpler than internal links
3035                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
3036                         wfProfileIn( $fname.'-interwiki' );
3037                         # Make interwiki link HTML
3038                         $wgOutputReplace = array();
3039                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
3040                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
3041                                 $wgOutputReplace[$key] = $sk->makeLinkObj( $title, $link );
3042                         }
3043
3044                         $text = preg_replace_callback(
3045                                 '/<!--IWLINK (.*?)-->/',
3046                                 "wfOutputReplaceMatches",
3047                                 $text );
3048                         wfProfileOut( $fname.'-interwiki' );
3049                 }
3050
3051                 wfProfileOut( $fname );
3052                 return $colours;
3053         }
3054
3055         /**
3056          * Replace <!--LINK--> link placeholders with plain text of links
3057          * (not HTML-formatted).
3058          * @param string $text
3059          * @return string
3060          */
3061         function replaceLinkHoldersText( $text ) {
3062                 global $wgUser, $wgLinkCache;
3063                 global $wgOutputReplace;
3064
3065                 $fname = 'Parser::replaceLinkHoldersText';
3066                 wfProfileIn( $fname );
3067
3068                 $text = preg_replace_callback(
3069                         '/<!--(LINK|IWLINK) (.*?)-->/',
3070                         array( &$this, 'replaceLinkHoldersTextCallback' ),
3071                         $text );
3072
3073                 wfProfileOut( $fname );
3074                 return $text;
3075         }
3076
3077         /**
3078          * @param array $matches
3079          * @return string
3080          * @access private
3081          */
3082         function replaceLinkHoldersTextCallback( $matches ) {
3083                 $type = $matches[1];
3084                 $key  = $matches[2];
3085                 if( $type == 'LINK' ) {
3086                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
3087                                 return $this->mLinkHolders['texts'][$key];
3088                         }
3089                 } elseif( $type == 'IWLINK' ) {
3090                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
3091                                 return $this->mInterwikiLinkHolders['texts'][$key];
3092                         }
3093                 }
3094                 return $matches[0];
3095         }
3096
3097         /**
3098          * Renders an image gallery from a text with one line per image.
3099          * text labels may be given by using |-style alternative text. E.g.
3100          *   Image:one.jpg|The number "1"
3101          *   Image:tree.jpg|A tree
3102          * given as text will return the HTML of a gallery with two images,
3103          * labeled 'The number "1"' and
3104          * 'A tree'.
3105          *
3106          * @static
3107          */
3108         function renderImageGallery( $text ) {
3109                 # Setup the parser
3110                 global $wgUser, $wgParser, $wgTitle;
3111                 $parserOptions = ParserOptions::newFromUser( $wgUser );
3112
3113                 global $wgLinkCache;
3114                 $ig = new ImageGallery();
3115                 $ig->setShowBytes( false );
3116                 $ig->setShowFilename( false );
3117                 $lines = explode( "\n", $text );
3118
3119                 foreach ( $lines as $line ) {
3120                         # match lines like these:
3121                         # Image:someimage.jpg|This is some image
3122                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
3123                         # Skip empty lines
3124                         if ( count( $matches ) == 0 ) {
3125                                 continue;
3126                         }
3127                         $nt = Title::newFromURL( $matches[1] );
3128                         if( is_null( $nt ) ) {
3129                                 # Bogus title. Ignore these so we don't bomb out later.
3130                                 continue;
3131                         }
3132                         if ( isset( $matches[3] ) ) {
3133                                 $label = $matches[3];
3134                         } else {
3135                                 $label = '';
3136                         }
3137
3138                         $html = $wgParser->parse( $label , $wgTitle, $parserOptions );
3139                         $html = $html->mText;
3140
3141                         $ig->add( new Image( $nt ), $html );
3142                         $wgLinkCache->addImageLinkObj( $nt );
3143                 }
3144                 return $ig->toHTML();
3145         }
3146
3147         /**
3148          * Parse image options text and use it to make an image
3149          */
3150         function makeImage( &$nt, $options ) {
3151                 global $wgContLang, $wgUseImageResize;
3152                 global $wgUser, $wgThumbLimits;
3153
3154                 $align = '';
3155
3156                 # Check if the options text is of the form "options|alt text"
3157                 # Options are:
3158                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
3159                 #  * left               no resizing, just left align. label is used for alt= only
3160                 #  * right              same, but right aligned
3161                 #  * none               same, but not aligned
3162                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
3163                 #  * center             center the image
3164                 #  * framed             Keep original image size, no magnify-button.
3165
3166                 $part = explode( '|', $options);
3167
3168                 $mwThumb  =& MagicWord::get( MAG_IMG_THUMBNAIL );
3169                 $mwLeft   =& MagicWord::get( MAG_IMG_LEFT );
3170                 $mwRight  =& MagicWord::get( MAG_IMG_RIGHT );
3171                 $mwNone   =& MagicWord::get( MAG_IMG_NONE );
3172                 $mwWidth  =& MagicWord::get( MAG_IMG_WIDTH );
3173                 $mwCenter =& MagicWord::get( MAG_IMG_CENTER );
3174                 $mwFramed =& MagicWord::get( MAG_IMG_FRAMED );
3175                 $caption = '';
3176
3177                 $width = $height = $framed = $thumb = false;
3178                 $manual_thumb = "" ;
3179
3180                 foreach( $part as $key => $val ) {
3181                         $val_parts = explode ( "=" , $val , 2 ) ;
3182                         $left_part = array_shift ( $val_parts ) ;
3183                         if ( $wgUseImageResize && ! is_null( $mwThumb->matchVariableStartToEnd($val) ) ) {
3184                                 $thumb=true;
3185                         } elseif ( $wgUseImageResize && count ( $val_parts ) == 1 && ! is_null( $mwThumb->matchVariableStartToEnd($left_part) ) ) {
3186                                 # use manually specified thumbnail
3187                                 $thumb=true;
3188                                 $manual_thumb = array_shift ( $val_parts ) ;
3189                         } elseif ( ! is_null( $mwRight->matchVariableStartToEnd($val) ) ) {
3190                                 # remember to set an alignment, don't render immediately
3191                                 $align = 'right';
3192                         } elseif ( ! is_null( $mwLeft->matchVariableStartToEnd($val) ) ) {
3193                                 # remember to set an alignment, don't render immediately
3194                                 $align = 'left';
3195                         } elseif ( ! is_null( $mwCenter->matchVariableStartToEnd($val) ) ) {
3196                                 # remember to set an alignment, don't render immediately
3197                                 $align = 'center';
3198                         } elseif ( ! is_null( $mwNone->matchVariableStartToEnd($val) ) ) {
3199                                 # remember to set an alignment, don't render immediately
3200                                 $align = 'none';
3201                         } elseif ( $wgUseImageResize && ! is_null( $match = $mwWidth->matchVariableStartToEnd($val) ) ) {
3202                                 wfDebug( "MAG_IMG_WIDTH match: $match\n" );
3203                                 # $match is the image width in pixels
3204                                 if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $match, $m ) ) {
3205                                         $width = intval( $m[1] );
3206                                         $height = intval( $m[2] );
3207                                 } else {
3208                                         $width = intval($match);
3209                                 }
3210                         } elseif ( ! is_null( $mwFramed->matchVariableStartToEnd($val) ) ) {
3211                                 $framed=true;
3212                         } else {
3213                                 $caption = $val;
3214                         }
3215                 }
3216                 # Strip bad stuff out of the alt text
3217                 $alt = $this->replaceLinkHoldersText( $caption );
3218                 $alt = Sanitizer::stripAllTags( $alt );
3219
3220                 # Linker does the rest
3221                 $sk =& $this->mOptions->getSkin();
3222                 return $sk->makeImageLinkObj( $nt, $caption, $alt, $align, $width, $height, $framed, $thumb, $manual_thumb );
3223         }
3224 }
3225
3226 /**
3227  * @todo document
3228  * @package MediaWiki
3229  */
3230 class ParserOutput
3231 {
3232         var $mText, $mLanguageLinks, $mCategoryLinks, $mContainsOldMagic;
3233         var $mCacheTime; # Timestamp on this article, or -1 for uncacheable. Used in ParserCache.
3234         var $mVersion;   # Compatibility check
3235         var $mTitleText; # title text of the chosen language variant
3236
3237         function ParserOutput( $text = '', $languageLinks = array(), $categoryLinks = array(),
3238                 $containsOldMagic = false, $titletext = '' )
3239         {
3240                 $this->mText = $text;
3241                 $this->mLanguageLinks = $languageLinks;
3242                 $this->mCategoryLinks = $categoryLinks;
3243                 $this->mContainsOldMagic = $containsOldMagic;
3244                 $this->mCacheTime = '';
3245                 $this->mVersion = MW_PARSER_VERSION;
3246                 $this->mTitleText = $titletext;
3247         }
3248
3249         function getText()                   { return $this->mText; }
3250         function getLanguageLinks()          { return $this->mLanguageLinks; }
3251         function getCategoryLinks()          { return array_keys( $this->mCategoryLinks ); }
3252         function getCacheTime()              { return $this->mCacheTime; }
3253         function getTitleText()              { return $this->mTitleText; }
3254         function containsOldMagic()          { return $this->mContainsOldMagic; }
3255         function setText( $text )            { return wfSetVar( $this->mText, $text ); }
3256         function setLanguageLinks( $ll )     { return wfSetVar( $this->mLanguageLinks, $ll ); }
3257         function setCategoryLinks( $cl )     { return wfSetVar( $this->mCategoryLinks, $cl ); }
3258         function setContainsOldMagic( $com ) { return wfSetVar( $this->mContainsOldMagic, $com ); }
3259         function setCacheTime( $t )          { return wfSetVar( $this->mCacheTime, $t ); }
3260         function setTitleText( $t )          { return wfSetVar ($this->mTitleText, $t); }
3261
3262         function addCategoryLink( $c )       { $this->mCategoryLinks[$c] = 1; }
3263
3264         function merge( $other ) {
3265                 $this->mLanguageLinks = array_merge( $this->mLanguageLinks, $other->mLanguageLinks );
3266                 $this->mCategoryLinks = array_merge( $this->mCategoryLinks, $this->mLanguageLinks );
3267                 $this->mContainsOldMagic = $this->mContainsOldMagic || $other->mContainsOldMagic;
3268         }
3269
3270         /**
3271          * Return true if this cached output object predates the global or
3272          * per-article cache invalidation timestamps, or if it comes from
3273          * an incompatible older version.
3274          *
3275          * @param string $touched the affected article's last touched timestamp
3276          * @return bool
3277          * @access public
3278          */
3279         function expired( $touched ) {
3280                 global $wgCacheEpoch;
3281                 return $this->getCacheTime() == -1 || // parser says it's uncacheable
3282                        $this->getCacheTime() <= $touched ||
3283                        $this->getCacheTime() <= $wgCacheEpoch ||
3284                        !isset( $this->mVersion ) ||
3285                        version_compare( $this->mVersion, MW_PARSER_VERSION, "lt" );
3286         }
3287 }
3288
3289 /**
3290  * Set options of the Parser
3291  * @todo document
3292  * @package MediaWiki
3293  */
3294 class ParserOptions
3295 {
3296         # All variables are private
3297         var $mUseTeX;                    # Use texvc to expand <math> tags
3298         var $mUseDynamicDates;           # Use DateFormatter to format dates
3299         var $mInterwikiMagic;            # Interlanguage links are removed and returned in an array
3300         var $mAllowExternalImages;       # Allow external images inline
3301         var $mSkin;                      # Reference to the preferred skin
3302         var $mDateFormat;                # Date format index
3303         var $mEditSection;               # Create "edit section" links
3304         var $mNumberHeadings;            # Automatically number headings
3305         var $mAllowSpecialInclusion;     # Allow inclusion of special pages
3306
3307         function getUseTeX()                        { return $this->mUseTeX; }
3308         function getUseDynamicDates()               { return $this->mUseDynamicDates; }
3309         function getInterwikiMagic()                { return $this->mInterwikiMagic; }
3310         function getAllowExternalImages()           { return $this->mAllowExternalImages; }
3311         function getSkin()                          { return $this->mSkin; }
3312         function getDateFormat()                    { return $this->mDateFormat; }
3313         function getEditSection()                   { return $this->mEditSection; }
3314         function getNumberHeadings()                { return $this->mNumberHeadings; }
3315         function getAllowSpecialInclusion()         { return $this->mAllowSpecialInclusion; }
3316
3317
3318         function setUseTeX( $x )                    { return wfSetVar( $this->mUseTeX, $x ); }
3319         function setUseDynamicDates( $x )           { return wfSetVar( $this->mUseDynamicDates, $x ); }
3320         function setInterwikiMagic( $x )            { return wfSetVar( $this->mInterwikiMagic, $x ); }
3321         function setAllowExternalImages( $x )       { return wfSetVar( $this->mAllowExternalImages, $x ); }
3322         function setDateFormat( $x )                { return wfSetVar( $this->mDateFormat, $x ); }
3323         function setEditSection( $x )               { return wfSetVar( $this->mEditSection, $x ); }
3324         function setNumberHeadings( $x )            { return wfSetVar( $this->mNumberHeadings, $x ); }
3325         function setAllowSpecialInclusion( $x )     { return wfSetVar( $this->mAllowSpecialInclusion, $x ); }
3326
3327         function setSkin( &$x ) { $this->mSkin =& $x; }
3328
3329         function ParserOptions() {
3330                 global $wgUser;
3331                 $this->initialiseFromUser( $wgUser );
3332         }
3333
3334         /**
3335          * Get parser options
3336          * @static
3337          */
3338         function newFromUser( &$user ) {
3339                 $popts = new ParserOptions;
3340                 $popts->initialiseFromUser( $user );
3341                 return $popts;
3342         }
3343
3344         /** Get user options */
3345         function initialiseFromUser( &$userInput ) {
3346                 global $wgUseTeX, $wgUseDynamicDates, $wgInterwikiMagic, $wgAllowExternalImages,
3347                        $wgAllowSpecialInclusion;
3348                 $fname = 'ParserOptions::initialiseFromUser';
3349                 wfProfileIn( $fname );
3350                 if ( !$userInput ) {
3351                         $user = new User;
3352                         $user->setLoaded( true );
3353                 } else {
3354                         $user =& $userInput;
3355                 }
3356
3357                 $this->mUseTeX = $wgUseTeX;
3358                 $this->mUseDynamicDates = $wgUseDynamicDates;
3359                 $this->mInterwikiMagic = $wgInterwikiMagic;
3360                 $this->mAllowExternalImages = $wgAllowExternalImages;
3361                 wfProfileIn( $fname.'-skin' );
3362                 $this->mSkin =& $user->getSkin();
3363                 wfProfileOut( $fname.'-skin' );
3364                 $this->mDateFormat = $user->getOption( 'date' );
3365                 $this->mEditSection = $user->getOption( 'editsection' );
3366                 $this->mNumberHeadings = $user->getOption( 'numberheadings' );
3367                 $this->mAllowSpecialInclusion = $wgAllowSpecialInclusion;
3368                 wfProfileOut( $fname );
3369         }
3370 }
3371
3372 /**
3373  * Callback function used by Parser::replaceLinkHolders()
3374  * to substitute link placeholders.
3375  */
3376 function &wfOutputReplaceMatches( $matches ) {
3377         global $wgOutputReplace;
3378         return $wgOutputReplace[$matches[1]];
3379 }
3380
3381 /**
3382  * Return the total number of articles
3383  */
3384 function wfNumberOfArticles() {
3385         global $wgNumberOfArticles;
3386
3387         wfLoadSiteStats();
3388         return $wgNumberOfArticles;
3389 }
3390
3391 /**
3392  * Get various statistics from the database
3393  * @private
3394  */
3395 function wfLoadSiteStats() {
3396         global $wgNumberOfArticles, $wgTotalViews, $wgTotalEdits;
3397         $fname = 'wfLoadSiteStats';
3398
3399         if ( -1 != $wgNumberOfArticles ) return;
3400         $dbr =& wfGetDB( DB_SLAVE );
3401         $s = $dbr->selectRow( 'site_stats',
3402                 array( 'ss_total_views', 'ss_total_edits', 'ss_good_articles' ),
3403                 array( 'ss_row_id' => 1 ), $fname
3404         );
3405
3406         if ( $s === false ) {
3407                 return;
3408         } else {
3409                 $wgTotalViews = $s->ss_total_views;
3410                 $wgTotalEdits = $s->ss_total_edits;
3411                 $wgNumberOfArticles = $s->ss_good_articles;
3412         }
3413 }
3414
3415 /**
3416  * Escape html tags
3417  * Basicly replacing " > and < with HTML entities ( &quot;, &gt;, &lt;)
3418  *
3419  * @param string $in Text that might contain HTML tags
3420  * @return string Escaped string
3421  */
3422 function wfEscapeHTMLTagsOnly( $in ) {
3423         return str_replace(
3424                 array( '"', '>', '<' ),
3425                 array( '&quot;', '&gt;', '&lt;' ),
3426                 $in );
3427 }
3428
3429 ?>