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