includes/Parser.php

   1 <?php
   2 /**
   3  * File for Parser and related classes
   4  *
   5  * @package MediaWiki
   6  * @subpackage Parser
   7  */
   8
   9 /** */
  10 require_once( 'Sanitizer.php' );
  11 require_once( 'HttpFunctions.php' );
  12
  13 /**
  14  * Update this version number when the ParserOutput format
  15  * changes in an incompatible way, so the parser cache
  16  * can automatically discard old data.
  17  */
  18 define( 'MW_PARSER_VERSION', '1.5.0' );
  19
  20 /**
  21  * Variable substitution O(N^2) attack
  22  *
  23  * Without countermeasures, it would be possible to attack the parser by saving
  24  * a page filled with a large number of inclusions of large pages. The size of
  25  * the generated page would be proportional to the square of the input size.
  26  * Hence, we limit the number of inclusions of any given page, thus bringing any
  27  * attack back to O(N).
  28  */
  29
  30 define( 'MAX_INCLUDE_REPEAT', 100 );
  31 define( 'MAX_INCLUDE_SIZE', 1000000 ); // 1 Million
  32
  33 define( 'RLH_FOR_UPDATE', 1 );
  34
  35 # Allowed values for $mOutputType
  36 define( 'OT_HTML', 1 );
  37 define( 'OT_WIKI', 2 );
  38 define( 'OT_MSG' , 3 );
  39
  40 # string parameter for extractTags which will cause it
  41 # to strip HTML comments in addition to regular
  42 # <XML>-style tags. This should not be anything we
  43 # may want to use in wikisyntax
  44 define( 'STRIP_COMMENTS', 'HTMLCommentStrip' );
  45
  46 # prefix for escaping, used in two functions at least
  47 define( 'UNIQ_PREFIX', 'NaodW29');
  48
  49 # Constants needed for external link processing
  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',  '/\[(\b('.$wgUrlProtocols.')'.EXT_LINK_URL_CLASS.'+) *('.EXT_LINK_TEXT_CLASS.'*?)\]/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         var $mIWTransData = array();
 114         var $mAssocArgs = array();
 115
 116         /**#@-*/
 117
 118         /**
 119          * Constructor
 120          *
 121          * @access public
 122          */
 123         function Parser() {
 124                 global $wgContLang;
 125                 $this->mTemplates = array();
 126                 $this->mTemplatePath = array();
 127                 $this->mTagHooks = array();
 128                 $this->clearState();
 129         }
 130
 131         /**
 132          * Clear Parser state
 133          *
 134          * @access private
 135          */
 136         function clearState() {
 137                 $this->mOutput = new ParserOutput;
 138                 $this->mAutonumber = 0;
 139                 $this->mLastSection = '';
 140                 $this->mDTopen = false;
 141                 $this->mVariables = false;
 142                 $this->mIncludeCount = array();
 143                 $this->mStripState = array();
 144                 $this->mArgStack = array();
 145                 $this->mInPre = false;
 146                 $this->mInterwikiLinkHolders = array(
 147                         'texts' => array(),
 148                         'titles' => array()
 149                 );
 150                 $this->mLinkHolders = array(
 151                         'namespaces' => array(),
 152                         'dbkeys' => array(),
 153                         'queries' => array(),
 154                         'texts' => array(),
 155                         'titles' => array()
 156                 );
 157         }
 158
 159         /**
 160          * First pass--just handle <nowiki> sections, pass the rest off
 161          * to internalParse() which does all the real work.
 162          *
 163          * @access private
 164          * @param string $text Text we want to parse
 165          * @param Title &$title A title object
 166          * @param array $options
 167          * @param boolean $linestart
 168          * @param boolean $clearState
 169          * @return ParserOutput a ParserOutput
 170          */
 171         function parse( $text, &$title, $options, $linestart = true, $clearState = true ) {
 172                 global $wgUseTidy, $wgContLang;
 173                 $fname = 'Parser::parse';
 174                 wfProfileIn( $fname );
 175
 176                 if ( $clearState ) {
 177                         $this->clearState();
 178                 }
 179
 180                 $this->mOptions = $options;
 181                 $this->mTitle =& $title;
 182                 $this->mOutputType = OT_HTML;
 183
 184                 $this->mStripState = NULL;
 185
 186                 //$text = $this->strip( $text, $this->mStripState );
 187                 // VOODOO MAGIC FIX! Sometimes the above segfaults in PHP5.
 188                 $x =& $this->mStripState;
 189
 190                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$x ) );
 191                 $text = $this->strip( $text, $x );
 192                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$x ) );
 193
 194                 $text = $this->internalParse( $text );
 195
 196                 $text = $this->unstrip( $text, $this->mStripState );
 197
 198                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 199                 $fixtags = array(
 200                         # french spaces, last one Guillemet-left
 201                         # only if there is something before the space
 202                         '/(.) (?=\\?|:|;|!|\\302\\273)/' => '\\1&nbsp;\\2',
 203                         # french spaces, Guillemet-right
 204                         '/(\\302\\253) /' => '\\1&nbsp;',
 205                         '/<center *>(.*)<\\/center *>/i' => '<div class="center">\\1</div>',
 206                 );
 207                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 208
 209                 # only once and last
 210                 $text = $this->doBlockLevels( $text, $linestart );
 211
 212                 $this->replaceLinkHolders( $text );
 213
 214                 # the position of the convert() call should not be changed. it
 215                 # assumes that the links are all replaces and the only thing left
 216                 # is the <nowiki> mark.
 217                 $text = $wgContLang->convert($text);
 218                 $this->mOutput->setTitleText($wgContLang->getParsedTitle());
 219
 220                 $text = $this->unstripNoWiki( $text, $this->mStripState );
 221
 222                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 223
 224                 $text = Sanitizer::normalizeCharReferences( $text );
 225                 global $wgUseTidy;
 226                 if ($wgUseTidy) {
 227                         $text = Parser::tidy($text);
 228                 }
 229
 230                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 231
 232                 $this->mOutput->setText( $text );
 233                 wfProfileOut( $fname );
 234                 return $this->mOutput;
 235         }
 236
 237         /**
 238          * Get a random string
 239          *
 240          * @access private
 241          * @static
 242          */
 243         function getRandomString() {
 244                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 245         }
 246
 247         /**
 248          * Replaces all occurrences of <$tag>content</$tag> in the text
 249          * with a random marker and returns the new text. the output parameter
 250          * $content will be an associative array filled with data on the form
 251          * $unique_marker => content.
 252          *
 253          * If $content is already set, the additional entries will be appended
 254          * If $tag is set to STRIP_COMMENTS, the function will extract
 255          * <!-- HTML comments -->
 256          *
 257          * @access private
 258          * @static
 259          */
 260         function extractTagsAndParams($tag, $text, &$content, &$tags, &$params, $uniq_prefix = ''){
 261                 $rnd = $uniq_prefix . '-' . $tag . Parser::getRandomString();
 262                 if ( !$content ) {
 263                         $content = array( );
 264                 }
 265                 $n = 1;
 266                 $stripped = '';
 267
 268                 if ( !$tags ) {
 269                         $tags = array( );
 270                 }
 271
 272                 if ( !$params ) {
 273                         $params = array( );
 274                 }
 275
 276                 if( $tag == STRIP_COMMENTS ) {
 277                         $start = '/<!--()/';
 278                         $end   = '/-->/';
 279                 } else {
 280                         $start = "/<$tag(\\s+[^>]*|\\s*)>/i";
 281                         $end   = "/<\\/$tag\\s*>/i";
 282                 }
 283
 284                 while ( '' != $text ) {
 285                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 286                         $stripped .= $p[0];
 287                         if( count( $p ) < 3 ) {
 288                                 break;
 289                         }
 290                         $attributes = $p[1];
 291                         $inside     = $p[2];
 292
 293                         $marker = $rnd . sprintf('%08X', $n++);
 294                         $stripped .= $marker;
 295
 296                         $tags[$marker] = "<$tag$attributes>";
 297                         $params[$marker] = Sanitizer::decodeTagAttributes( $attributes );
 298
 299                         $q = preg_split( $end, $inside, 2 );
 300                         $content[$marker] = $q[0];
 301                         if( count( $q ) < 2 ) {
 302                                 # No end tag -- let it run out to the end of the text.
 303                                 break;
 304                         } else {
 305                                 $text = $q[1];
 306                         }
 307                 }
 308                 return $stripped;
 309         }
 310
 311         /**
 312          * Wrapper function for extractTagsAndParams
 313          * for cases where $tags and $params isn't needed
 314          * i.e. where tags will never have params, like <nowiki>
 315          *
 316          * @access private
 317          * @static
 318          */
 319         function extractTags( $tag, $text, &$content, $uniq_prefix = '' ) {
 320                 $dummy_tags = array();
 321                 $dummy_params = array();
 322
 323                 return Parser::extractTagsAndParams( $tag, $text, $content,
 324                         $dummy_tags, $dummy_params, $uniq_prefix );
 325         }
 326
 327         /**
 328          * Strips and renders nowiki, pre, math, hiero
 329          * If $render is set, performs necessary rendering operations on plugins
 330          * Returns the text, and fills an array with data needed in unstrip()
 331          * If the $state is already a valid strip state, it adds to the state
 332          *
 333          * @param bool $stripcomments when set, HTML comments <!-- like this -->
 334          *  will be stripped in addition to other tags. This is important
 335          *  for section editing, where these comments cause confusion when
 336          *  counting the sections in the wikisource
 337          *
 338          * @access private
 339          */
 340         function strip( $text, &$state, $stripcomments = false ) {
 341                 $render = ($this->mOutputType == OT_HTML);
 342                 $html_content = array();
 343                 $nowiki_content = array();
 344                 $math_content = array();
 345                 $pre_content = array();
 346                 $comment_content = array();
 347                 $ext_content = array();
 348                 $ext_tags = array();
 349                 $ext_params = array();
 350                 $gallery_content = array();
 351
 352                 # Replace any instances of the placeholders
 353                 $uniq_prefix = UNIQ_PREFIX;
 354                 #$text = str_replace( $uniq_prefix, wfHtmlEscapeFirst( $uniq_prefix ), $text );
 355
 356                 # html
 357                 global $wgRawHtml;
 358                 if( $wgRawHtml ) {
 359                         $text = Parser::extractTags('html', $text, $html_content, $uniq_prefix);
 360                         foreach( $html_content as $marker => $content ) {
 361                                 if ($render ) {
 362                                         # Raw and unchecked for validity.
 363                                         $html_content[$marker] = $content;
 364                                 } else {
 365                                         $html_content[$marker] = '<html>'.$content.'</html>';
 366                                 }
 367                         }
 368                 }
 369
 370                 # nowiki
 371                 $text = Parser::extractTags('nowiki', $text, $nowiki_content, $uniq_prefix);
 372                 foreach( $nowiki_content as $marker => $content ) {
 373                         if( $render ){
 374                                 $nowiki_content[$marker] = wfEscapeHTMLTagsOnly( $content );
 375                         } else {
 376                                 $nowiki_content[$marker] = '<nowiki>'.$content.'</nowiki>';
 377                         }
 378                 }
 379
 380                 # math
 381                 $text = Parser::extractTags('math', $text, $math_content, $uniq_prefix);
 382                 foreach( $math_content as $marker => $content ){
 383                         if( $render ) {
 384                                 if( $this->mOptions->getUseTeX() ) {
 385                                         $math_content[$marker] = renderMath( $content );
 386                                 } else {
 387                                         $math_content[$marker] = '&lt;math&gt;'.$content.'&lt;math&gt;';
 388                                 }
 389                         } else {
 390                                 $math_content[$marker] = '<math>'.$content.'</math>';
 391                         }
 392                 }
 393
 394                 # pre
 395                 $text = Parser::extractTags('pre', $text, $pre_content, $uniq_prefix);
 396                 foreach( $pre_content as $marker => $content ){
 397                         if( $render ){
 398                                 $pre_content[$marker] = '<pre>' . wfEscapeHTMLTagsOnly( $content ) . '</pre>';
 399                         } else {
 400                                 $pre_content[$marker] = '<pre>'.$content.'</pre>';
 401                         }
 402                 }
 403
 404                 # gallery
 405                 $text = Parser::extractTags('gallery', $text, $gallery_content, $uniq_prefix);
 406                 foreach( $gallery_content as $marker => $content ) {
 407                         require_once( 'ImageGallery.php' );
 408                         if ( $render ) {
 409                                 $gallery_content[$marker] = Parser::renderImageGallery( $content );
 410                         } else {
 411                                 $gallery_content[$marker] = '<gallery>'.$content.'</gallery>';
 412                         }
 413                 }
 414
 415                 # Comments
 416                 if($stripcomments) {
 417                         $text = Parser::extractTags(STRIP_COMMENTS, $text, $comment_content, $uniq_prefix);
 418                         foreach( $comment_content as $marker => $content ){
 419                                 $comment_content[$marker] = '<!--'.$content.'-->';
 420                         }
 421                 }
 422
 423                 # Extensions
 424                 foreach ( $this->mTagHooks as $tag => $callback ) {
 425                         $ext_content[$tag] = array();
 426                         $text = Parser::extractTagsAndParams( $tag, $text, $ext_content[$tag],
 427                                 $ext_tags[$tag], $ext_params[$tag], $uniq_prefix );
 428                         foreach( $ext_content[$tag] as $marker => $content ) {
 429                                 $full_tag = $ext_tags[$tag][$marker];
 430                                 $params = $ext_params[$tag][$marker];
 431                                 if ( $render ) {
 432                                         $ext_content[$tag][$marker] = $callback( $content, $params, $this );
 433                                 } else {
 434                                         $ext_content[$tag][$marker] = "$full_tag$content</$tag>";
 435                                 }
 436                         }
 437                 }
 438
 439                 # Merge state with the pre-existing state, if there is one
 440                 if ( $state ) {
 441                         $state['html'] = $state['html'] + $html_content;
 442                         $state['nowiki'] = $state['nowiki'] + $nowiki_content;
 443                         $state['math'] = $state['math'] + $math_content;
 444                         $state['pre'] = $state['pre'] + $pre_content;
 445                         $state['comment'] = $state['comment'] + $comment_content;
 446                         $state['gallery'] = $state['gallery'] + $gallery_content;
 447
 448                         foreach( $ext_content as $tag => $array ) {
 449                                 if ( array_key_exists( $tag, $state ) ) {
 450                                         $state[$tag] = $state[$tag] + $array;
 451                                 }
 452                         }
 453                 } else {
 454                         $state = array(
 455                           'html' => $html_content,
 456                           'nowiki' => $nowiki_content,
 457                           'math' => $math_content,
 458                           'pre' => $pre_content,
 459                           'comment' => $comment_content,
 460                           'gallery' => $gallery_content,
 461                         ) + $ext_content;
 462                 }
 463                 return $text;
 464         }
 465
 466         /**
 467          * restores pre, math, and hiero removed by strip()
 468          *
 469          * always call unstripNoWiki() after this one
 470          * @access private
 471          */
 472         function unstrip( $text, &$state ) {
 473                 if ( !is_array( $state ) ) {
 474                         return $text;
 475                 }
 476
 477                 # Must expand in reverse order, otherwise nested tags will be corrupted
 478                 $contentDict = end( $state );
 479                 for ( $contentDict = end( $state ); $contentDict !== false; $contentDict = prev( $state ) ) {
 480                         if( key($state) != 'nowiki' && key($state) != 'html') {
 481                                 for ( $content = end( $contentDict ); $content !== false; $content = prev( $contentDict ) ) {
 482                                         $text = str_replace( key( $contentDict ), $content, $text );
 483                                 }
 484                         }
 485                 }
 486
 487                 return $text;
 488         }
 489
 490         /**
 491          * always call this after unstrip() to preserve the order
 492          *
 493          * @access private
 494          */
 495         function unstripNoWiki( $text, &$state ) {
 496                 if ( !is_array( $state ) ) {
 497                         return $text;
 498                 }
 499
 500                 # Must expand in reverse order, otherwise nested tags will be corrupted
 501                 for ( $content = end($state['nowiki']); $content !== false; $content = prev( $state['nowiki'] ) ) {
 502                         $text = str_replace( key( $state['nowiki'] ), $content, $text );
 503                 }
 504
 505                 global $wgRawHtml;
 506                 if ($wgRawHtml) {
 507                         for ( $content = end($state['html']); $content !== false; $content = prev( $state['html'] ) ) {
 508                                 $text = str_replace( key( $state['html'] ), $content, $text );
 509                         }
 510                 }
 511
 512                 return $text;
 513         }
 514
 515         /**
 516          * Add an item to the strip state
 517          * Returns the unique tag which must be inserted into the stripped text
 518          * The tag will be replaced with the original text in unstrip()
 519          *
 520          * @access private
 521          */
 522         function insertStripItem( $text, &$state ) {
 523                 $rnd = UNIQ_PREFIX . '-item' . Parser::getRandomString();
 524                 if ( !$state ) {
 525                         $state = array(
 526                           'html' => array(),
 527                           'nowiki' => array(),
 528                           'math' => array(),
 529                           'pre' => array(),
 530                           'comment' => array(),
 531                           'gallery' => array(),
 532                         );
 533                 }
 534                 $state['item'][$rnd] = $text;
 535                 return $rnd;
 536         }
 537
 538         /**
 539          * Interface with html tidy, used if $wgUseTidy = true.
 540          * If tidy isn't able to correct the markup, the original will be
 541          * returned in all its glory with a warning comment appended.
 542          *
 543          * Either the external tidy program or the in-process tidy extension
 544          * will be used depending on availability. Override the default
 545          * $wgTidyInternal setting to disable the internal if it's not working.
 546          *
 547          * @param string $text Hideous HTML input
 548          * @return string Corrected HTML output
 549          * @access public
 550          * @static
 551          */
 552         function tidy( $text ) {
 553                 global $wgTidyInternal;
 554                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 555 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 556 '<head><title>test</title></head><body>'.$text.'</body></html>';
 557                 if( $wgTidyInternal ) {
 558                         $correctedtext = Parser::internalTidy( $wrappedtext );
 559                 } else {
 560                         $correctedtext = Parser::externalTidy( $wrappedtext );
 561                 }
 562                 if( is_null( $correctedtext ) ) {
 563                         wfDebug( "Tidy error detected!\n" );
 564                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 565                 }
 566                 return $correctedtext;
 567         }
 568
 569         /**
 570          * Spawn an external HTML tidy process and get corrected markup back from it.
 571          *
 572          * @access private
 573          * @static
 574          */
 575         function externalTidy( $text ) {
 576                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 577                 $fname = 'Parser::externalTidy';
 578                 wfProfileIn( $fname );
 579
 580                 $cleansource = '';
 581                 $opts = ' -utf8';
 582
 583                 $descriptorspec = array(
 584                         0 => array('pipe', 'r'),
 585                         1 => array('pipe', 'w'),
 586                         2 => array('file', '/dev/null', 'a')
 587                 );
 588                 $pipes = array();
 589                 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 590                 if (is_resource($process)) {
 591                         fwrite($pipes[0], $text);
 592                         fclose($pipes[0]);
 593                         while (!feof($pipes[1])) {
 594                                 $cleansource .= fgets($pipes[1], 1024);
 595                         }
 596                         fclose($pipes[1]);
 597                         proc_close($process);
 598                 }
 599
 600                 wfProfileOut( $fname );
 601
 602                 if( $cleansource == '' && $text != '') {
 603                         // Some kind of error happened, so we couldn't get the corrected text.
 604                         // Just give up; we'll use the source text and append a warning.
 605                         return null;
 606                 } else {
 607                         return $cleansource;
 608                 }
 609         }
 610
 611         /**
 612          * Use the HTML tidy PECL extension to use the tidy library in-process,
 613          * saving the overhead of spawning a new process. Currently written to
 614          * the PHP 4.3.x version of the extension, may not work on PHP 5.
 615          *
 616          * 'pear install tidy' should be able to compile the extension module.
 617          *
 618          * @access private
 619          * @static
 620          */
 621         function internalTidy( $text ) {
 622                 global $wgTidyConf;
 623                 $fname = 'Parser::internalTidy';
 624                 wfProfileIn( $fname );
 625
 626                 tidy_load_config( $wgTidyConf );
 627                 tidy_set_encoding( 'utf8' );
 628                 tidy_parse_string( $text );
 629                 tidy_clean_repair();
 630                 if( tidy_get_status() == 2 ) {
 631                         // 2 is magic number for fatal error
 632                         // http://www.php.net/manual/en/function.tidy-get-status.php
 633                         $cleansource = null;
 634                 } else {
 635                         $cleansource = tidy_get_output();
 636                 }
 637                 wfProfileOut( $fname );
 638                 return $cleansource;
 639         }
 640
 641         /**
 642          * parse the wiki syntax used to render tables
 643          *
 644          * @access private
 645          */
 646         function doTableStuff ( $t ) {
 647                 $fname = 'Parser::doTableStuff';
 648                 wfProfileIn( $fname );
 649
 650                 $t = explode ( "\n" , $t ) ;
 651                 $td = array () ; # Is currently a td tag open?
 652                 $ltd = array () ; # Was it TD or TH?
 653                 $tr = array () ; # Is currently a tr tag open?
 654                 $ltr = array () ; # tr attributes
 655                 $indent_level = 0; # indent level of the table
 656                 foreach ( $t AS $k => $x )
 657                 {
 658                         $x = trim ( $x ) ;
 659                         $fc = substr ( $x , 0 , 1 ) ;
 660                         if ( preg_match( '/^(:*)\{\|(.*)$/', $x, $matches ) ) {
 661                                 $indent_level = strlen( $matches[1] );
 662                                 $t[$k] = str_repeat( '<dl><dd>', $indent_level ) .
 663                                         '<table' . Sanitizer::fixTagAttributes ( $matches[2], 'table' ) . '>' ;
 664                                 array_push ( $td , false ) ;
 665                                 array_push ( $ltd , '' ) ;
 666                                 array_push ( $tr , false ) ;
 667                                 array_push ( $ltr , '' ) ;
 668                         }
 669                         else if ( count ( $td ) == 0 ) { } # Don't do any of the following
 670                         else if ( '|}' == substr ( $x , 0 , 2 ) ) {
 671                                 $z = "</table>" . substr ( $x , 2);
 672                                 $l = array_pop ( $ltd ) ;
 673                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 674                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 675                                 array_pop ( $ltr ) ;
 676                                 $t[$k] = $z . str_repeat( '</dd></dl>', $indent_level );
 677                         }
 678                         else if ( '|-' == substr ( $x , 0 , 2 ) ) { # Allows for |---------------
 679                                 $x = substr ( $x , 1 ) ;
 680                                 while ( $x != '' && substr ( $x , 0 , 1 ) == '-' ) $x = substr ( $x , 1 ) ;
 681                                 $z = '' ;
 682                                 $l = array_pop ( $ltd ) ;
 683                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 684                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 685                                 array_pop ( $ltr ) ;
 686                                 $t[$k] = $z ;
 687                                 array_push ( $tr , false ) ;
 688                                 array_push ( $td , false ) ;
 689                                 array_push ( $ltd , '' ) ;
 690                                 array_push ( $ltr , Sanitizer::fixTagAttributes ( $x, 'tr' ) ) ;
 691                         }
 692                         else if ( '|' == $fc || '!' == $fc || '|+' == substr ( $x , 0 , 2 ) ) { # Caption
 693                                 # $x is a table row
 694                                 if ( '|+' == substr ( $x , 0 , 2 ) ) {
 695                                         $fc = '+' ;
 696                                         $x = substr ( $x , 1 ) ;
 697                                 }
 698                                 $after = substr ( $x , 1 ) ;
 699                                 if ( $fc == '!' ) $after = str_replace ( '!!' , '||' , $after ) ;
 700                                 $after = explode ( '||' , $after ) ;
 701                                 $t[$k] = '' ;
 702
 703                                 # Loop through each table cell
 704                                 foreach ( $after AS $theline )
 705                                 {
 706                                         $z = '' ;
 707                                         if ( $fc != '+' )
 708                                         {
 709                                                 $tra = array_pop ( $ltr ) ;
 710                                                 if ( !array_pop ( $tr ) ) $z = '<tr'.$tra.">\n" ;
 711                                                 array_push ( $tr , true ) ;
 712                                                 array_push ( $ltr , '' ) ;
 713                                         }
 714
 715                                         $l = array_pop ( $ltd ) ;
 716                                         if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 717                                         if ( $fc == '|' ) $l = 'td' ;
 718                                         else if ( $fc == '!' ) $l = 'th' ;
 719                                         else if ( $fc == '+' ) $l = 'caption' ;
 720                                         else $l = '' ;
 721                                         array_push ( $ltd , $l ) ;
 722
 723                                         # Cell parameters
 724                                         $y = explode ( '|' , $theline , 2 ) ;
 725                                         # Note that a '|' inside an invalid link should not
 726                                         # be mistaken as delimiting cell parameters
 727                                         if ( strpos( $y[0], '[[' ) !== false ) {
 728                                                 $y = array ($theline);
 729                                         }
 730                                         if ( count ( $y ) == 1 )
 731                                                 $y = "{$z}<{$l}>{$y[0]}" ;
 732                                         else $y = $y = "{$z}<{$l}".Sanitizer::fixTagAttributes($y[0], $l).">{$y[1]}" ;
 733                                         $t[$k] .= $y ;
 734                                         array_push ( $td , true ) ;
 735                                 }
 736                         }
 737                 }
 738
 739                 # Closing open td, tr && table
 740                 while ( count ( $td ) > 0 )
 741                 {
 742                         if ( array_pop ( $td ) ) $t[] = '</td>' ;
 743                         if ( array_pop ( $tr ) ) $t[] = '</tr>' ;
 744                         $t[] = '</table>' ;
 745                 }
 746
 747                 $t = implode ( "\n" , $t ) ;
 748                 wfProfileOut( $fname );
 749                 return $t ;
 750         }
 751
 752         /**
 753          * Helper function for parse() that transforms wiki markup into
 754          * HTML. Only called for $mOutputType == OT_HTML.
 755          *
 756          * @access private
 757          */
 758         function internalParse( $text ) {
 759                 global $wgContLang;
 760                 $args = array();
 761                 $isMain = true;
 762                 $fname = 'Parser::internalParse';
 763                 wfProfileIn( $fname );
 764
 765                 # Remove <noinclude> tags and <includeonly> sections
 766                 $text = strtr( $text, array( '<noinclude>' => '', '</noinclude>' => '') );
 767                 $text = preg_replace( '/<includeonly>.*?<\/includeonly>/s', '', $text );
 768
 769                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'replaceVariables' ) );
 770                 $text = $this->replaceVariables( $text, $args );
 771
 772                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 773
 774                 $text = $this->doHeadings( $text );
 775                 if($this->mOptions->getUseDynamicDates()) {
 776                         $df =& DateFormatter::getInstance();
 777                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 778                 }
 779                 $text = $this->doAllQuotes( $text );
 780                 $text = $this->replaceInternalLinks( $text );
 781                 $text = $this->replaceExternalLinks( $text );
 782
 783                 # replaceInternalLinks may sometimes leave behind
 784                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 785                 $text = str_replace("http-noparse://","http://",$text);
 786
 787                 $text = $this->doMagicLinks( $text );
 788                 $text = $this->doTableStuff( $text );
 789                 $text = $this->formatHeadings( $text, $isMain );
 790
 791                 $regex = '/<!--IW_TRANSCLUDE (\d+)-->/';
 792                 $text = preg_replace_callback($regex, array(&$this, 'scarySubstitution'), $text);
 793
 794                 wfProfileOut( $fname );
 795                 return $text;
 796         }
 797
 798         function scarySubstitution($matches) {
 799 #               return "[[".$matches[0]."]]";
 800                 return $this->mIWTransData[(int)$matches[0]];
 801         }
 802
 803         /**
 804          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 805          * magic external links.
 806          *
 807          * @access private
 808          */
 809         function &doMagicLinks( &$text ) {
 810                 $text = $this->magicISBN( $text );
 811                 $text = $this->magicRFC( $text, 'RFC ', 'rfcurl' );
 812                 $text = $this->magicRFC( $text, 'PMID ', 'pubmedurl' );
 813                 return $text;
 814         }
 815
 816         /**
 817          * Parse ^^ tokens and return html
 818          *
 819          * @access private
 820          */
 821         function doExponent( $text ) {
 822                 $fname = 'Parser::doExponent';
 823                 wfProfileIn( $fname );
 824                 $text = preg_replace('/\^\^(.*)\^\^/','<small><sup>\\1</sup></small>', $text);
 825                 wfProfileOut( $fname );
 826                 return $text;
 827         }
 828
 829         /**
 830          * Parse headers and return html
 831          *
 832          * @access private
 833          */
 834         function doHeadings( $text ) {
 835                 $fname = 'Parser::doHeadings';
 836                 wfProfileIn( $fname );
 837                 for ( $i = 6; $i >= 1; --$i ) {
 838                         $h = substr( '======', 0, $i );
 839                         $text = preg_replace( "/^{$h}(.+){$h}(\\s|$)/m",
 840                           "<h{$i}>\\1</h{$i}>\\2", $text );
 841                 }
 842                 wfProfileOut( $fname );
 843                 return $text;
 844         }
 845
 846         /**
 847          * Replace single quotes with HTML markup
 848          * @access private
 849          * @return string the altered text
 850          */
 851         function doAllQuotes( $text ) {
 852                 $fname = 'Parser::doAllQuotes';
 853                 wfProfileIn( $fname );
 854                 $outtext = '';
 855                 $lines = explode( "\n", $text );
 856                 foreach ( $lines as $line ) {
 857                         $outtext .= $this->doQuotes ( $line ) . "\n";
 858                 }
 859                 $outtext = substr($outtext, 0,-1);
 860                 wfProfileOut( $fname );
 861                 return $outtext;
 862         }
 863
 864         /**
 865          * Helper function for doAllQuotes()
 866          * @access private
 867          */
 868         function doQuotes( $text ) {
 869                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
 870                 if ( count( $arr ) == 1 )
 871                         return $text;
 872                 else
 873                 {
 874                         # First, do some preliminary work. This may shift some apostrophes from
 875                         # being mark-up to being text. It also counts the number of occurrences
 876                         # of bold and italics mark-ups.
 877                         $i = 0;
 878                         $numbold = 0;
 879                         $numitalics = 0;
 880                         foreach ( $arr as $r )
 881                         {
 882                                 if ( ( $i % 2 ) == 1 )
 883                                 {
 884                                         # If there are ever four apostrophes, assume the first is supposed to
 885                                         # be text, and the remaining three constitute mark-up for bold text.
 886                                         if ( strlen( $arr[$i] ) == 4 )
 887                                         {
 888                                                 $arr[$i-1] .= "'";
 889                                                 $arr[$i] = "'''";
 890                                         }
 891                                         # If there are more than 5 apostrophes in a row, assume they're all
 892                                         # text except for the last 5.
 893                                         else if ( strlen( $arr[$i] ) > 5 )
 894                                         {
 895                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
 896                                                 $arr[$i] = "'''''";
 897                                         }
 898                                         # Count the number of occurrences of bold and italics mark-ups.
 899                                         # We are not counting sequences of five apostrophes.
 900                                         if ( strlen( $arr[$i] ) == 2 ) $numitalics++;  else
 901                                         if ( strlen( $arr[$i] ) == 3 ) $numbold++;     else
 902                                         if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
 903                                 }
 904                                 $i++;
 905                         }
 906
 907                         # If there is an odd number of both bold and italics, it is likely
 908                         # that one of the bold ones was meant to be an apostrophe followed
 909                         # by italics. Which one we cannot know for certain, but it is more
 910                         # likely to be one that has a single-letter word before it.
 911                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
 912                         {
 913                                 $i = 0;
 914                                 $firstsingleletterword = -1;
 915                                 $firstmultiletterword = -1;
 916                                 $firstspace = -1;
 917                                 foreach ( $arr as $r )
 918                                 {
 919                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
 920                                         {
 921                                                 $x1 = substr ($arr[$i-1], -1);
 922                                                 $x2 = substr ($arr[$i-1], -2, 1);
 923                                                 if ($x1 == ' ') {
 924                                                         if ($firstspace == -1) $firstspace = $i;
 925                                                 } else if ($x2 == ' ') {
 926                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
 927                                                 } else {
 928                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
 929                                                 }
 930                                         }
 931                                         $i++;
 932                                 }
 933
 934                                 # If there is a single-letter word, use it!
 935                                 if ($firstsingleletterword > -1)
 936                                 {
 937                                         $arr [ $firstsingleletterword ] = "''";
 938                                         $arr [ $firstsingleletterword-1 ] .= "'";
 939                                 }
 940                                 # If not, but there's a multi-letter word, use that one.
 941                                 else if ($firstmultiletterword > -1)
 942                                 {
 943                                         $arr [ $firstmultiletterword ] = "''";
 944                                         $arr [ $firstmultiletterword-1 ] .= "'";
 945                                 }
 946                                 # ... otherwise use the first one that has neither.
 947                                 # (notice that it is possible for all three to be -1 if, for example,
 948                                 # there is only one pentuple-apostrophe in the line)
 949                                 else if ($firstspace > -1)
 950                                 {
 951                                         $arr [ $firstspace ] = "''";
 952                                         $arr [ $firstspace-1 ] .= "'";
 953                                 }
 954                         }
 955
 956                         # Now let's actually convert our apostrophic mush to HTML!
 957                         $output = '';
 958                         $buffer = '';
 959                         $state = '';
 960                         $i = 0;
 961                         foreach ($arr as $r)
 962                         {
 963                                 if (($i % 2) == 0)
 964                                 {
 965                                         if ($state == 'both')
 966                                                 $buffer .= $r;
 967                                         else
 968                                                 $output .= $r;
 969                                 }
 970                                 else
 971                                 {
 972                                         if (strlen ($r) == 2)
 973                                         {
 974                                                 if ($state == 'i')
 975                                                 { $output .= '</i>'; $state = ''; }
 976                                                 else if ($state == 'bi')
 977                                                 { $output .= '</i>'; $state = 'b'; }
 978                                                 else if ($state == 'ib')
 979                                                 { $output .= '</b></i><b>'; $state = 'b'; }
 980                                                 else if ($state == 'both')
 981                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
 982                                                 else # $state can be 'b' or ''
 983                                                 { $output .= '<i>'; $state .= 'i'; }
 984                                         }
 985                                         else if (strlen ($r) == 3)
 986                                         {
 987                                                 if ($state == 'b')
 988                                                 { $output .= '</b>'; $state = ''; }
 989                                                 else if ($state == 'bi')
 990                                                 { $output .= '</i></b><i>'; $state = 'i'; }
 991                                                 else if ($state == 'ib')
 992                                                 { $output .= '</b>'; $state = 'i'; }
 993                                                 else if ($state == 'both')
 994                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
 995                                                 else # $state can be 'i' or ''
 996                                                 { $output .= '<b>'; $state .= 'b'; }
 997                                         }
 998                                         else if (strlen ($r) == 5)
 999                                         {
1000                                                 if ($state == 'b')
1001                                                 { $output .= '</b><i>'; $state = 'i'; }
1002                                                 else if ($state == 'i')
1003                                                 { $output .= '</i><b>'; $state = 'b'; }
1004                                                 else if ($state == 'bi')
1005                                                 { $output .= '</i></b>'; $state = ''; }
1006                                                 else if ($state == 'ib')
1007                                                 { $output .= '</b></i>'; $state = ''; }
1008                                                 else if ($state == 'both')
1009                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1010                                                 else # ($state == '')
1011                                                 { $buffer = ''; $state = 'both'; }
1012                                         }
1013                                 }
1014                                 $i++;
1015                         }
1016                         # Now close all remaining tags.  Notice that the order is important.
1017                         if ($state == 'b' || $state == 'ib')
1018                                 $output .= '</b>';
1019                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
1020                                 $output .= '</i>';
1021                         if ($state == 'bi')
1022                                 $output .= '</b>';
1023                         if ($state == 'both')
1024                                 $output .= '<b><i>'.$buffer.'</i></b>';
1025                         return $output;
1026                 }
1027         }
1028
1029         /**
1030          * Replace external links
1031          *
1032          * Note: this is all very hackish and the order of execution matters a lot.
1033          * Make sure to run maintenance/parserTests.php if you change this code.
1034          *
1035          * @access private
1036          */
1037         function replaceExternalLinks( $text ) {
1038                 global $wgContLang;
1039                 $fname = 'Parser::replaceExternalLinks';
1040                 wfProfileIn( $fname );
1041
1042                 $sk =& $this->mOptions->getSkin();
1043
1044                 $bits = preg_split( EXT_LINK_BRACKETED, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1045
1046                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
1047
1048                 $i = 0;
1049                 while ( $i<count( $bits ) ) {
1050                         $url = $bits[$i++];
1051                         $protocol = $bits[$i++];
1052                         $text = $bits[$i++];
1053                         $trail = $bits[$i++];
1054
1055                         # The characters '<' and '>' (which were escaped by
1056                         # removeHTMLtags()) should not be included in
1057                         # URLs, per RFC 2396.
1058                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1059                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1060                                 $url = substr($url, 0, $m2[0][1]);
1061                         }
1062
1063                         # If the link text is an image URL, replace it with an <img> tag
1064                         # This happened by accident in the original parser, but some people used it extensively
1065                         $img = $this->maybeMakeExternalImage( $text );
1066                         if ( $img !== false ) {
1067                                 $text = $img;
1068                         }
1069
1070                         $dtrail = '';
1071
1072                         # Set linktype for CSS - if URL==text, link is essentially free
1073                         $linktype = ($text == $url) ? 'free' : 'text';
1074
1075                         # No link text, e.g. [http://domain.tld/some.link]
1076                         if ( $text == '' ) {
1077                                 # Autonumber if allowed
1078                                 if ( strpos( HTTP_PROTOCOLS, str_replace('/','\/', $protocol) ) !== false ) {
1079                                         $text = '[' . ++$this->mAutonumber . ']';
1080                                         $linktype = 'autonumber';
1081                                 } else {
1082                                         # Otherwise just use the URL
1083                                         $text = htmlspecialchars( $url );
1084                                         $linktype = 'free';
1085                                 }
1086                         } else {
1087                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1088                                 # Check for trail
1089                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1090                         }
1091
1092                         $text = $wgContLang->markNoConversion($text);
1093
1094                         # Replace &amp; from obsolete syntax with &.
1095                         # All HTML entities will be escaped by makeExternalLink()
1096                         # or maybeMakeExternalImage()
1097                         $url = str_replace( '&amp;', '&', $url );
1098
1099                         # Process the trail (i.e. everything after this link up until start of the next link),
1100                         # replacing any non-bracketed links
1101                         $trail = $this->replaceFreeExternalLinks( $trail );
1102
1103
1104                         # Use the encoded URL
1105                         # This means that users can paste URLs directly into the text
1106                         # Funny characters like &ouml; aren't valid in URLs anyway
1107                         # This was changed in August 2004
1108                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype ) . $dtrail . $trail;
1109                 }
1110
1111                 wfProfileOut( $fname );
1112                 return $s;
1113         }
1114
1115         /**
1116          * Replace anything that looks like a URL with a link
1117          * @access private
1118          */
1119         function replaceFreeExternalLinks( $text ) {
1120                 global $wgUrlProtocols;
1121                 global $wgContLang;
1122                 $fname = 'Parser::replaceFreeExternalLinks';
1123                 wfProfileIn( $fname );
1124
1125                 $bits = preg_split( '/(\b(?:'.$wgUrlProtocols.'))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1126                 $s = array_shift( $bits );
1127                 $i = 0;
1128
1129                 $sk =& $this->mOptions->getSkin();
1130
1131                 while ( $i < count( $bits ) ){
1132                         $protocol = $bits[$i++];
1133                         $remainder = $bits[$i++];
1134
1135                         if ( preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1136                                 # Found some characters after the protocol that look promising
1137                                 $url = $protocol . $m[1];
1138                                 $trail = $m[2];
1139
1140                                 # The characters '<' and '>' (which were escaped by
1141                                 # removeHTMLtags()) should not be included in
1142                                 # URLs, per RFC 2396.
1143                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1144                                         $trail = substr($url, $m2[0][1]) . $trail;
1145                                         $url = substr($url, 0, $m2[0][1]);
1146                                 }
1147
1148                                 # Move trailing punctuation to $trail
1149                                 $sep = ',;\.:!?';
1150                                 # If there is no left bracket, then consider right brackets fair game too
1151                                 if ( strpos( $url, '(' ) === false ) {
1152                                         $sep .= ')';
1153                                 }
1154
1155                                 $numSepChars = strspn( strrev( $url ), $sep );
1156                                 if ( $numSepChars ) {
1157                                         $trail = substr( $url, -$numSepChars ) . $trail;
1158                                         $url = substr( $url, 0, -$numSepChars );
1159                                 }
1160
1161                                 # Replace &amp; from obsolete syntax with &.
1162                                 # All HTML entities will be escaped by makeExternalLink()
1163                                 # or maybeMakeExternalImage()
1164                                 $url = str_replace( '&amp;', '&', $url );
1165
1166                                 # Is this an external image?
1167                                 $text = $this->maybeMakeExternalImage( $url );
1168                                 if ( $text === false ) {
1169                                         # Not an image, make a link
1170                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free' );
1171                                 }
1172                                 $s .= $text . $trail;
1173                         } else {
1174                                 $s .= $protocol . $remainder;
1175                         }
1176                 }
1177                 wfProfileOut( $fname );
1178                 return $s;
1179         }
1180
1181         /**
1182          * make an image if it's allowed
1183          * @access private
1184          */
1185         function maybeMakeExternalImage( $url ) {
1186                 $sk =& $this->mOptions->getSkin();
1187                 $text = false;
1188                 if ( $this->mOptions->getAllowExternalImages() ) {
1189                         if ( preg_match( EXT_IMAGE_REGEX, $url ) ) {
1190                                 # Image found
1191                                 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
1192                         }
1193                 }
1194                 return $text;
1195         }
1196
1197         /**
1198          * Process [[ ]] wikilinks
1199          *
1200          * @access private
1201          */
1202         function replaceInternalLinks( $s ) {
1203                 global $wgContLang, $wgLinkCache, $wgUrlProtocols;
1204                 static $fname = 'Parser::replaceInternalLinks' ;
1205
1206                 wfProfileIn( $fname );
1207
1208                 wfProfileIn( $fname.'-setup' );
1209                 static $tc = FALSE;
1210                 # the % is needed to support urlencoded titles as well
1211                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1212
1213                 $sk =& $this->mOptions->getSkin();
1214
1215                 #split the entire text string on occurences of [[
1216                 $a = explode( '[[', ' ' . $s );
1217                 #get the first element (all text up to first [[), and remove the space we added
1218                 $s = array_shift( $a );
1219                 $s = substr( $s, 1 );
1220
1221                 # Match a link having the form [[namespace:link|alternate]]trail
1222                 static $e1 = FALSE;
1223                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1224                 # Match cases where there is no "]]", which might still be images
1225                 static $e1_img = FALSE;
1226                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1227                 # Match the end of a line for a word that's not followed by whitespace,
1228                 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1229                 $e2 = wfMsgForContent( 'linkprefix' );
1230
1231                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1232
1233                 if( is_null( $this->mTitle ) ) {
1234                         wfDebugDieBacktrace( 'nooo' );
1235                 }
1236                 $nottalk = !$this->mTitle->isTalkPage();
1237
1238                 if ( $useLinkPrefixExtension ) {
1239                         if ( preg_match( $e2, $s, $m ) ) {
1240                                 $first_prefix = $m[2];
1241                                 $s = $m[1];
1242                         } else {
1243                                 $first_prefix = false;
1244                         }
1245                 } else {
1246                         $prefix = '';
1247                 }
1248
1249                 $selflink = $this->mTitle->getPrefixedText();
1250                 wfProfileOut( $fname.'-setup' );
1251
1252                 $checkVariantLink = sizeof($wgContLang->getVariants())>1;
1253                 $useSubpages = $this->areSubpagesAllowed();
1254
1255                 # Loop for each link
1256                 for ($k = 0; isset( $a[$k] ); $k++) {
1257                         $line = $a[$k];
1258                         if ( $useLinkPrefixExtension ) {
1259                                 wfProfileIn( $fname.'-prefixhandling' );
1260                                 if ( preg_match( $e2, $s, $m ) ) {
1261                                         $prefix = $m[2];
1262                                         $s = $m[1];
1263                                 } else {
1264                                         $prefix='';
1265                                 }
1266                                 # first link
1267                                 if($first_prefix) {
1268                                         $prefix = $first_prefix;
1269                                         $first_prefix = false;
1270                                 }
1271                                 wfProfileOut( $fname.'-prefixhandling' );
1272                         }
1273
1274                         $might_be_img = false;
1275
1276                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1277                                 $text = $m[2];
1278                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1279                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1280                                 # the real problem is with the $e1 regex
1281                                 # See bug 1300.
1282                                 #
1283                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1284                                 # and no image is in sight. See bug 2095.
1285                                 #
1286                                 if( $text !== '' && preg_match( "/^\](.*)/s", $m[3], $n ) ) {
1287                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1288                                         $m[3] = $n[1];
1289                                 }
1290                                 # fix up urlencoded title texts
1291                                 if(preg_match('/%/', $m[1] )) $m[1] = urldecode($m[1]);
1292                                 $trail = $m[3];
1293                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1294                                 $might_be_img = true;
1295                                 $text = $m[2];
1296                                 if(preg_match('/%/', $m[1] )) $m[1] = urldecode($m[1]);
1297                                 $trail = "";
1298                         } else { # Invalid form; output directly
1299                                 $s .= $prefix . '[[' . $line ;
1300                                 continue;
1301                         }
1302
1303                         # Don't allow internal links to pages containing
1304                         # PROTO: where PROTO is a valid URL protocol; these
1305                         # should be external links.
1306                         if (preg_match('/^(\b(?:'.$wgUrlProtocols.'))/', $m[1])) {
1307                                 $s .= $prefix . '[[' . $line ;
1308                                 continue;
1309                         }
1310
1311                         # Make subpage if necessary
1312                         if( $useSubpages ) {
1313                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1314                         } else {
1315                                 $link = $m[1];
1316                         }
1317
1318                         $noforce = (substr($m[1], 0, 1) != ':');
1319                         if (!$noforce) {
1320                                 # Strip off leading ':'
1321                                 $link = substr($link, 1);
1322                         }
1323
1324                         $nt = Title::newFromText( $this->unstripNoWiki($link, $this->mStripState) );
1325                         if( !$nt ) {
1326                                 $s .= $prefix . '[[' . $line;
1327                                 continue;
1328                         }
1329
1330                         #check other language variants of the link
1331                         #if the article does not exist
1332                         if( $checkVariantLink
1333                             && $nt->getArticleID() == 0 ) {
1334                                 $wgContLang->findVariantLink($link, $nt);
1335                         }
1336
1337                         $ns = $nt->getNamespace();
1338                         $iw = $nt->getInterWiki();
1339
1340                         if ($might_be_img) { # if this is actually an invalid link
1341                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1342                                         $found = false;
1343                                         while (isset ($a[$k+1]) ) {
1344                                                 #look at the next 'line' to see if we can close it there
1345                                                 $next_line =  array_shift(array_splice( $a, $k + 1, 1) );
1346                                                 if( preg_match("/^(.*?]].*?)]](.*)$/sD", $next_line, $m) ) {
1347                                                 # the first ]] closes the inner link, the second the image
1348                                                         $found = true;
1349                                                         $text .= '[[' . $m[1];
1350                                                         $trail = $m[2];
1351                                                         break;
1352                                                 } elseif( preg_match("/^.*?]].*$/sD", $next_line, $m) ) {
1353                                                         #if there's exactly one ]] that's fine, we'll keep looking
1354                                                         $text .= '[[' . $m[0];
1355                                                 } else {
1356                                                         #if $next_line is invalid too, we need look no further
1357                                                         $text .= '[[' . $next_line;
1358                                                         break;
1359                                                 }
1360                                         }
1361                                         if ( !$found ) {
1362                                                 # we couldn't find the end of this imageLink, so output it raw
1363                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1364                                                 $text = $this->replaceInternalLinks($text);
1365                                                 $s .= $prefix . '[[' . $link . '|' . $text;
1366                                                 # note: no $trail, because without an end, there *is* no trail
1367                                                 continue;
1368                                         }
1369                                 } else { #it's not an image, so output it raw
1370                                         $s .= $prefix . '[[' . $link . '|' . $text;
1371                                         # note: no $trail, because without an end, there *is* no trail
1372                                         continue;
1373                                 }
1374                         }
1375
1376                         $wasblank = ( '' == $text );
1377                         if( $wasblank ) $text = $link;
1378
1379
1380                         # Link not escaped by : , create the various objects
1381                         if( $noforce ) {
1382
1383                                 # Interwikis
1384                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1385                                         array_push( $this->mOutput->mLanguageLinks, $nt->getFullText() );
1386                                         $s = rtrim($s . "\n");
1387                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1388                                         continue;
1389                                 }
1390
1391                                 if ( $ns == NS_IMAGE ) {
1392                                         wfProfileIn( "$fname-image" );
1393                                         if ( !wfIsBadImage( $nt->getDBkey() ) ) {
1394                                                 # recursively parse links inside the image caption
1395                                                 # actually, this will parse them in any other parameters, too,
1396                                                 # but it might be hard to fix that, and it doesn't matter ATM
1397                                                 $text = $this->replaceExternalLinks($text);
1398                                                 $text = $this->replaceInternalLinks($text);
1399
1400                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1401                                                 $s .= $prefix . str_replace('http://', 'http-noparse://', $this->makeImage( $nt, $text ) ) . $trail;
1402                                                 $wgLinkCache->addImageLinkObj( $nt );
1403
1404                                                 wfProfileOut( "$fname-image" );
1405                                                 continue;
1406                                         }
1407                                         wfProfileOut( "$fname-image" );
1408
1409                                 }
1410
1411                                 if ( $ns == NS_CATEGORY ) {
1412                                         wfProfileIn( "$fname-category" );
1413                                         $t = $wgContLang->convertHtml( $nt->getText() );
1414                                         $s = rtrim($s . "\n"); # bug 87
1415
1416                                         $wgLinkCache->suspend(); # Don't save in links/brokenlinks
1417                                         $t = $sk->makeLinkObj( $nt, $t, '', '' , $prefix );
1418                                         $wgLinkCache->resume();
1419
1420                                         if ( $wasblank ) {
1421                                                 if ( $this->mTitle->getNamespace() == NS_CATEGORY ) {
1422                                                         $sortkey = $this->mTitle->getText();
1423                                                 } else {
1424                                                         $sortkey = $this->mTitle->getPrefixedText();
1425                                                 }
1426                                         } else {
1427                                                 $sortkey = $text;
1428                                         }
1429                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1430                                         $wgLinkCache->addCategoryLinkObj( $nt, $sortkey );
1431                                         $this->mOutput->addCategoryLink( $t );
1432
1433                                         /**
1434                                          * Strip the whitespace Category links produce, see bug 87
1435                                          * @todo We might want to use trim($tmp, "\n") here.
1436                                          */
1437                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1438
1439                                         wfProfileOut( "$fname-category" );
1440                                         continue;
1441                                 }
1442                         }
1443
1444                         if( ( $nt->getPrefixedText() === $selflink ) &&
1445                             ( $nt->getFragment() === '' ) ) {
1446                                 # Self-links are handled specially; generally de-link and change to bold.
1447                                 $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1448                                 continue;
1449                         }
1450
1451                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1452                         if( $ns == NS_MEDIA ) {
1453                                 $s .= $prefix . $sk->makeMediaLinkObj( $nt, $text, true ) . $trail;
1454                                 $wgLinkCache->addImageLinkObj( $nt );
1455                                 continue;
1456                         } elseif( $ns == NS_SPECIAL ) {
1457                                 $s .= $prefix . $sk->makeKnownLinkObj( $nt, $text, '', $trail );
1458                                 continue;
1459                         }
1460                         if( !$nt->isExternal() && $nt->isAlwaysKnown() ) {
1461                                 /**
1462                                  * Skip lookups for special pages and self-links.
1463                                  * External interwiki links are not included here because
1464                                  * the HTTP urls would break output in the next parse step;
1465                                  * they will have placeholders kept.
1466                                  */
1467                                 $s .= $sk->makeKnownLinkObj( $nt, $text, '', $trail, $prefix );
1468                         } else {
1469                                 /**
1470                                  * Add a link placeholder
1471                                  * Later, this will be replaced by a real link, after the existence or
1472                                  * non-existence of all the links is known
1473                                  */
1474                                 $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1475                         }
1476                 }
1477                 wfProfileOut( $fname );
1478                 return $s;
1479         }
1480
1481         /**
1482          * Make a link placeholder. The text returned can be later resolved to a real link with
1483          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1484          * parsing of interwiki links, and secondly to allow all extistence checks and
1485          * article length checks (for stub links) to be bundled into a single query.
1486          *
1487          */
1488         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1489                 if ( ! is_object($nt) ) {
1490                         # Fail gracefully
1491                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1492                 } else {
1493                         # Separate the link trail from the rest of the link
1494                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1495
1496                         if ( $nt->isExternal() ) {
1497                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1498                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1499                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1500                         } else {
1501                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1502                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1503                                 $this->mLinkHolders['queries'][] = $query;
1504                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1505                                 $this->mLinkHolders['titles'][] = $nt;
1506
1507                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1508                         }
1509                 }
1510                 return $retVal;
1511         }
1512
1513         /**
1514          * Return true if subpage links should be expanded on this page.
1515          * @return bool
1516          */
1517         function areSubpagesAllowed() {
1518                 # Some namespaces don't allow subpages
1519                 global $wgNamespacesWithSubpages;
1520                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1521         }
1522
1523         /**
1524          * Handle link to subpage if necessary
1525          * @param string $target the source of the link
1526          * @param string &$text the link text, modified as necessary
1527          * @return string the full name of the link
1528          * @access private
1529          */
1530         function maybeDoSubpageLink($target, &$text) {
1531                 # Valid link forms:
1532                 # Foobar -- normal
1533                 # :Foobar -- override special treatment of prefix (images, language links)
1534                 # /Foobar -- convert to CurrentPage/Foobar
1535                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1536                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1537                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1538
1539                 $fname = 'Parser::maybeDoSubpageLink';
1540                 wfProfileIn( $fname );
1541                 $ret = $target; # default return value is no change
1542
1543                 # Some namespaces don't allow subpages,
1544                 # so only perform processing if subpages are allowed
1545                 if( $this->areSubpagesAllowed() ) {
1546                         # Look at the first character
1547                         if( $target != '' && $target{0} == '/' ) {
1548                                 # / at end means we don't want the slash to be shown
1549                                 if( substr( $target, -1, 1 ) == '/' ) {
1550                                         $target = substr( $target, 1, -1 );
1551                                         $noslash = $target;
1552                                 } else {
1553                                         $noslash = substr( $target, 1 );
1554                                 }
1555
1556                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash);
1557                                 if( '' === $text ) {
1558                                         $text = $target;
1559                                 } # this might be changed for ugliness reasons
1560                         } else {
1561                                 # check for .. subpage backlinks
1562                                 $dotdotcount = 0;
1563                                 $nodotdot = $target;
1564                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1565                                         ++$dotdotcount;
1566                                         $nodotdot = substr( $nodotdot, 3 );
1567                                 }
1568                                 if($dotdotcount > 0) {
1569                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1570                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1571                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1572                                                 # / at the end means don't show full path
1573                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1574                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1575                                                         if( '' === $text ) {
1576                                                                 $text = $nodotdot;
1577                                                         }
1578                                                 }
1579                                                 $nodotdot = trim( $nodotdot );
1580                                                 if( $nodotdot != '' ) {
1581                                                         $ret .= '/' . $nodotdot;
1582                                                 }
1583                                         }
1584                                 }
1585                         }
1586                 }
1587
1588                 wfProfileOut( $fname );
1589                 return $ret;
1590         }
1591
1592         /**#@+
1593          * Used by doBlockLevels()
1594          * @access private
1595          */
1596         /* private */ function closeParagraph() {
1597                 $result = '';
1598                 if ( '' != $this->mLastSection ) {
1599                         $result = '</' . $this->mLastSection  . ">\n";
1600                 }
1601                 $this->mInPre = false;
1602                 $this->mLastSection = '';
1603                 return $result;
1604         }
1605         # getCommon() returns the length of the longest common substring
1606         # of both arguments, starting at the beginning of both.
1607         #
1608         /* private */ function getCommon( $st1, $st2 ) {
1609                 $fl = strlen( $st1 );
1610                 $shorter = strlen( $st2 );
1611                 if ( $fl < $shorter ) { $shorter = $fl; }
1612
1613                 for ( $i = 0; $i < $shorter; ++$i ) {
1614                         if ( $st1{$i} != $st2{$i} ) { break; }
1615                 }
1616                 return $i;
1617         }
1618         # These next three functions open, continue, and close the list
1619         # element appropriate to the prefix character passed into them.
1620         #
1621         /* private */ function openList( $char ) {
1622                 $result = $this->closeParagraph();
1623
1624                 if ( '*' == $char ) { $result .= '<ul><li>'; }
1625                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
1626                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
1627                 else if ( ';' == $char ) {
1628                         $result .= '<dl><dt>';
1629                         $this->mDTopen = true;
1630                 }
1631                 else { $result = '<!-- ERR 1 -->'; }
1632
1633                 return $result;
1634         }
1635
1636         /* private */ function nextItem( $char ) {
1637                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
1638                 else if ( ':' == $char || ';' == $char ) {
1639                         $close = '</dd>';
1640                         if ( $this->mDTopen ) { $close = '</dt>'; }
1641                         if ( ';' == $char ) {
1642                                 $this->mDTopen = true;
1643                                 return $close . '<dt>';
1644                         } else {
1645                                 $this->mDTopen = false;
1646                                 return $close . '<dd>';
1647                         }
1648                 }
1649                 return '<!-- ERR 2 -->';
1650         }
1651
1652         /* private */ function closeList( $char ) {
1653                 if ( '*' == $char ) { $text = '</li></ul>'; }
1654                 else if ( '#' == $char ) { $text = '</li></ol>'; }
1655                 else if ( ':' == $char ) {
1656                         if ( $this->mDTopen ) {
1657                                 $this->mDTopen = false;
1658                                 $text = '</dt></dl>';
1659                         } else {
1660                                 $text = '</dd></dl>';
1661                         }
1662                 }
1663                 else {  return '<!-- ERR 3 -->'; }
1664                 return $text."\n";
1665         }
1666         /**#@-*/
1667
1668         /**
1669          * Make lists from lines starting with ':', '*', '#', etc.
1670          *
1671          * @access private
1672          * @return string the lists rendered as HTML
1673          */
1674         function doBlockLevels( $text, $linestart ) {
1675                 $fname = 'Parser::doBlockLevels';
1676                 wfProfileIn( $fname );
1677
1678                 # Parsing through the text line by line.  The main thing
1679                 # happening here is handling of block-level elements p, pre,
1680                 # and making lists from lines starting with * # : etc.
1681                 #
1682                 $textLines = explode( "\n", $text );
1683
1684                 $lastPrefix = $output = '';
1685                 $this->mDTopen = $inBlockElem = false;
1686                 $prefixLength = 0;
1687                 $paragraphStack = false;
1688
1689                 if ( !$linestart ) {
1690                         $output .= array_shift( $textLines );
1691                 }
1692                 foreach ( $textLines as $oLine ) {
1693                         $lastPrefixLength = strlen( $lastPrefix );
1694                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
1695                         $preOpenMatch = preg_match('/<pre/i', $oLine );
1696                         if ( !$this->mInPre ) {
1697                                 # Multiple prefixes may abut each other for nested lists.
1698                                 $prefixLength = strspn( $oLine, '*#:;' );
1699                                 $pref = substr( $oLine, 0, $prefixLength );
1700
1701                                 # eh?
1702                                 $pref2 = str_replace( ';', ':', $pref );
1703                                 $t = substr( $oLine, $prefixLength );
1704                                 $this->mInPre = !empty($preOpenMatch);
1705                         } else {
1706                                 # Don't interpret any other prefixes in preformatted text
1707                                 $prefixLength = 0;
1708                                 $pref = $pref2 = '';
1709                                 $t = $oLine;
1710                         }
1711
1712                         # List generation
1713                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
1714                                 # Same as the last item, so no need to deal with nesting or opening stuff
1715                                 $output .= $this->nextItem( substr( $pref, -1 ) );
1716                                 $paragraphStack = false;
1717
1718                                 if ( substr( $pref, -1 ) == ';') {
1719                                         # The one nasty exception: definition lists work like this:
1720                                         # ; title : definition text
1721                                         # So we check for : in the remainder text to split up the
1722                                         # title and definition, without b0rking links.
1723                                         $term = $t2 = '';
1724                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
1725                                                 $t = $t2;
1726                                                 $output .= $term . $this->nextItem( ':' );
1727                                         }
1728                                 }
1729                         } elseif( $prefixLength || $lastPrefixLength ) {
1730                                 # Either open or close a level...
1731                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
1732                                 $paragraphStack = false;
1733
1734                                 while( $commonPrefixLength < $lastPrefixLength ) {
1735                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
1736                                         --$lastPrefixLength;
1737                                 }
1738                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
1739                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
1740                                 }
1741                                 while ( $prefixLength > $commonPrefixLength ) {
1742                                         $char = substr( $pref, $commonPrefixLength, 1 );
1743                                         $output .= $this->openList( $char );
1744
1745                                         if ( ';' == $char ) {
1746                                                 # FIXME: This is dupe of code above
1747                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
1748                                                         $t = $t2;
1749                                                         $output .= $term . $this->nextItem( ':' );
1750                                                 }
1751                                         }
1752                                         ++$commonPrefixLength;
1753                                 }
1754                                 $lastPrefix = $pref2;
1755                         }
1756                         if( 0 == $prefixLength ) {
1757                                 wfProfileIn( "$fname-paragraph" );
1758                                 # No prefix (not in list)--go to paragraph mode
1759                                 $uniq_prefix = UNIQ_PREFIX;
1760                                 // XXX: use a stack for nestable elements like span, table and div
1761                                 $openmatch = preg_match('/(<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
1762                                 $closematch = preg_match(
1763                                         '/(<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
1764                                         '<td|<th|<div|<\\/div|<hr|<\\/pre|<\\/p|'.$uniq_prefix.'-pre|<\\/li|<\\/ul)/iS', $t );
1765                                 if ( $openmatch or $closematch ) {
1766                                         $paragraphStack = false;
1767                                         $output .= $this->closeParagraph();
1768                                         if ( $preOpenMatch and !$preCloseMatch ) {
1769                                                 $this->mInPre = true;
1770                                         }
1771                                         if ( $closematch ) {
1772                                                 $inBlockElem = false;
1773                                         } else {
1774                                                 $inBlockElem = true;
1775                                         }
1776                                 } else if ( !$inBlockElem && !$this->mInPre ) {
1777                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
1778                                                 // pre
1779                                                 if ($this->mLastSection != 'pre') {
1780                                                         $paragraphStack = false;
1781                                                         $output .= $this->closeParagraph().'<pre>';
1782                                                         $this->mLastSection = 'pre';
1783                                                 }
1784                                                 $t = substr( $t, 1 );
1785                                         } else {
1786                                                 // paragraph
1787                                                 if ( '' == trim($t) ) {
1788                                                         if ( $paragraphStack ) {
1789                                                                 $output .= $paragraphStack.'<br />';
1790                                                                 $paragraphStack = false;
1791                                                                 $this->mLastSection = 'p';
1792                                                         } else {
1793                                                                 if ($this->mLastSection != 'p' ) {
1794                                                                         $output .= $this->closeParagraph();
1795                                                                         $this->mLastSection = '';
1796                                                                         $paragraphStack = '<p>';
1797                                                                 } else {
1798                                                                         $paragraphStack = '</p><p>';
1799                                                                 }
1800                                                         }
1801                                                 } else {
1802                                                         if ( $paragraphStack ) {
1803                                                                 $output .= $paragraphStack;
1804                                                                 $paragraphStack = false;
1805                                                                 $this->mLastSection = 'p';
1806                                                         } else if ($this->mLastSection != 'p') {
1807                                                                 $output .= $this->closeParagraph().'<p>';
1808                                                                 $this->mLastSection = 'p';
1809                                                         }
1810                                                 }
1811                                         }
1812                                 }
1813                                 wfProfileOut( "$fname-paragraph" );
1814                         }
1815                         // somewhere above we forget to get out of pre block (bug 785)
1816                         if($preCloseMatch && $this->mInPre) {
1817                                 $this->mInPre = false;
1818                         }
1819                         if ($paragraphStack === false) {
1820                                 $output .= $t."\n";
1821                         }
1822                 }
1823                 while ( $prefixLength ) {
1824                         $output .= $this->closeList( $pref2{$prefixLength-1} );
1825                         --$prefixLength;
1826                 }
1827                 if ( '' != $this->mLastSection ) {
1828                         $output .= '</' . $this->mLastSection . '>';
1829                         $this->mLastSection = '';
1830                 }
1831
1832                 wfProfileOut( $fname );
1833                 return $output;
1834         }
1835
1836         /**
1837          * Split up a string on ':', ignoring any occurences inside
1838          * <a>..</a> or <span>...</span>
1839          * @param string $str the string to split
1840          * @param string &$before set to everything before the ':'
1841          * @param string &$after set to everything after the ':'
1842          * return string the position of the ':', or false if none found
1843          */
1844         function findColonNoLinks($str, &$before, &$after) {
1845                 # I wonder if we should make this count all tags, not just <a>
1846                 # and <span>. That would prevent us from matching a ':' that
1847                 # comes in the middle of italics other such formatting....
1848                 # -- Wil
1849                 $fname = 'Parser::findColonNoLinks';
1850                 wfProfileIn( $fname );
1851                 $pos = 0;
1852                 do {
1853                         $colon = strpos($str, ':', $pos);
1854
1855                         if ($colon !== false) {
1856                                 $before = substr($str, 0, $colon);
1857                                 $after = substr($str, $colon + 1);
1858
1859                                 # Skip any ':' within <a> or <span> pairs
1860                                 $a = substr_count($before, '<a');
1861                                 $s = substr_count($before, '<span');
1862                                 $ca = substr_count($before, '</a>');
1863                                 $cs = substr_count($before, '</span>');
1864
1865                                 if ($a <= $ca and $s <= $cs) {
1866                                         # Tags are balanced before ':'; ok
1867                                         break;
1868                                 }
1869                                 $pos = $colon + 1;
1870                         }
1871                 } while ($colon !== false);
1872                 wfProfileOut( $fname );
1873                 return $colon;
1874         }
1875
1876         /**
1877          * Return value of a magic variable (like PAGENAME)
1878          *
1879          * @access private
1880          */
1881         function getVariableValue( $index ) {
1882                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgArticle, $wgScriptPath;
1883
1884                 /**
1885                  * Some of these require message or data lookups and can be
1886                  * expensive to check many times.
1887                  */
1888                 static $varCache = array();
1889                 if( isset( $varCache[$index] ) ) return $varCache[$index];
1890
1891                 switch ( $index ) {
1892                         case MAG_CURRENTMONTH:
1893                                 return $varCache[$index] = $wgContLang->formatNum( date( 'm' ) );
1894                         case MAG_CURRENTMONTHNAME:
1895                                 return $varCache[$index] = $wgContLang->getMonthName( date('n') );
1896                         case MAG_CURRENTMONTHNAMEGEN:
1897                                 return $varCache[$index] = $wgContLang->getMonthNameGen( date('n') );
1898                         case MAG_CURRENTMONTHABBREV:
1899                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( date('n') );
1900                         case MAG_CURRENTDAY:
1901                                 return $varCache[$index] = $wgContLang->formatNum( date('j') );
1902                         case MAG_PAGENAME:
1903                                 return $this->mTitle->getText();
1904                         case MAG_PAGENAMEE:
1905                                 return $this->mTitle->getPartialURL();
1906                         case MAG_REVISIONID:
1907                                 return $wgArticle->getRevIdFetched();
1908                         case MAG_NAMESPACE:
1909                                 # return Namespace::getCanonicalName($this->mTitle->getNamespace());
1910                                 return $wgContLang->getNsText($this->mTitle->getNamespace()); # Patch by Dori
1911                         case MAG_CURRENTDAYNAME:
1912                                 return $varCache[$index] = $wgContLang->getWeekdayName( date('w')+1 );
1913                         case MAG_CURRENTYEAR:
1914                                 return $varCache[$index] = $wgContLang->formatNum( date( 'Y' ), true );
1915                         case MAG_CURRENTTIME:
1916                                 return $varCache[$index] = $wgContLang->time( wfTimestampNow(), false );
1917                         case MAG_CURRENTWEEK:
1918                                 return $varCache[$index] = $wgContLang->formatNum( date('W') );
1919                         case MAG_CURRENTDOW:
1920                                 return $varCache[$index] = $wgContLang->formatNum( date('w') );
1921                         case MAG_NUMBEROFARTICLES:
1922                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfArticles() );
1923                         case MAG_NUMBEROFFILES:
1924                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfFiles() );
1925                         case MAG_SITENAME:
1926                                 return $wgSitename;
1927                         case MAG_SERVER:
1928                                 return $wgServer;
1929                         case MAG_SERVERNAME:
1930                                 return $wgServerName;
1931                         case MAG_SCRIPTPATH:
1932                                 return $wgScriptPath;
1933                         default:
1934                                 return NULL;
1935                 }
1936         }
1937
1938         /**
1939          * initialise the magic variables (like CURRENTMONTHNAME)
1940          *
1941          * @access private
1942          */
1943         function initialiseVariables() {
1944                 $fname = 'Parser::initialiseVariables';
1945                 wfProfileIn( $fname );
1946                 global $wgVariableIDs;
1947                 $this->mVariables = array();
1948                 foreach ( $wgVariableIDs as $id ) {
1949                         $mw =& MagicWord::get( $id );
1950                         $mw->addToArray( $this->mVariables, $id );
1951                 }
1952                 wfProfileOut( $fname );
1953         }
1954
1955         /**
1956          * Replace magic variables, templates, and template arguments
1957          * with the appropriate text. Templates are substituted recursively,
1958          * taking care to avoid infinite loops.
1959          *
1960          * Note that the substitution depends on value of $mOutputType:
1961          *  OT_WIKI: only {{subst:}} templates
1962          *  OT_MSG: only magic variables
1963          *  OT_HTML: all templates and magic variables
1964          *
1965          * @param string $tex The text to transform
1966          * @param array $args Key-value pairs representing template parameters to substitute
1967          * @access private
1968          */
1969         function replaceVariables( $text, $args = array() ) {
1970
1971                 # Prevent too big inclusions
1972                 if( strlen( $text ) > MAX_INCLUDE_SIZE ) {
1973                         return $text;
1974                 }
1975
1976                 $fname = 'Parser::replaceVariables';
1977                 wfProfileIn( $fname );
1978
1979                 $titleChars = Title::legalChars();
1980
1981                 # This function is called recursively. To keep track of arguments we need a stack:
1982                 array_push( $this->mArgStack, $args );
1983
1984                 # Variable substitution
1985                 $text = preg_replace_callback( "/{{([$titleChars]*?)}}/", array( &$this, 'variableSubstitution' ), $text );
1986
1987                 if ( $this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI ) {
1988                         # Argument substitution
1989                         $text = preg_replace_callback( "/{{{([$titleChars]*?)}}}/", array( &$this, 'argSubstitution' ), $text );
1990                 }
1991                 # Template substitution
1992                 $regex = '/(\\n|{)?{{(['.$titleChars.']*)(\\|.*?|)}}/s';
1993                 $text = preg_replace_callback( $regex, array( &$this, 'braceSubstitution' ), $text );
1994
1995                 array_pop( $this->mArgStack );
1996
1997                 wfProfileOut( $fname );
1998                 return $text;
1999         }
2000
2001         /**
2002          * Replace magic variables
2003          * @access private
2004          */
2005         function variableSubstitution( $matches ) {
2006                 $fname = 'parser::variableSubstitution';
2007                 $varname = $matches[1];
2008                 wfProfileIn( $fname );
2009                 if ( !$this->mVariables ) {
2010                         $this->initialiseVariables();
2011                 }
2012                 $skip = false;
2013                 if ( $this->mOutputType == OT_WIKI ) {
2014                         # Do only magic variables prefixed by SUBST
2015                         $mwSubst =& MagicWord::get( MAG_SUBST );
2016                         if (!$mwSubst->matchStartAndRemove( $varname ))
2017                                 $skip = true;
2018                         # Note that if we don't substitute the variable below,
2019                         # we don't remove the {{subst:}} magic word, in case
2020                         # it is a template rather than a magic variable.
2021                 }
2022                 if ( !$skip && array_key_exists( $varname, $this->mVariables ) ) {
2023                         $id = $this->mVariables[$varname];
2024                         $text = $this->getVariableValue( $id );
2025                         $this->mOutput->mContainsOldMagic = true;
2026                 } else {
2027                         $text = $matches[0];
2028                 }
2029                 wfProfileOut( $fname );
2030                 return $text;
2031         }
2032
2033         # Split template arguments
2034         function getTemplateArgs( $argsString ) {
2035                 if ( $argsString === '' ) {
2036                         return array();
2037                 }
2038
2039                 $args = explode( '|', substr( $argsString, 1 ) );
2040
2041                 # If any of the arguments contains a '[[' but no ']]', it needs to be
2042                 # merged with the next arg because the '|' character between belongs
2043                 # to the link syntax and not the template parameter syntax.
2044                 $argc = count($args);
2045
2046                 for ( $i = 0; $i < $argc-1; $i++ ) {
2047                         if ( substr_count ( $args[$i], '[[' ) != substr_count ( $args[$i], ']]' ) ) {
2048                                 $args[$i] .= '|'.$args[$i+1];
2049                                 array_splice($args, $i+1, 1);
2050                                 $i--;
2051                                 $argc--;
2052                         }
2053                 }
2054
2055                 return $args;
2056         }
2057
2058         /**
2059          * Return the text of a template, after recursively
2060          * replacing any variables or templates within the template.
2061          *
2062          * @param array $matches The parts of the template
2063          *  $matches[1]: the title, i.e. the part before the |
2064          *  $matches[2]: the parameters (including a leading |), if  any
2065          * @return string the text of the template
2066          * @access private
2067          */
2068         function braceSubstitution( $matches ) {
2069                 global $wgLinkCache, $wgContLang;
2070                 $fname = 'Parser::braceSubstitution';
2071                 wfProfileIn( $fname );
2072
2073                 $found = false;
2074                 $nowiki = false;
2075                 $noparse = false;
2076
2077                 $title = NULL;
2078
2079                 # Need to know if the template comes at the start of a line,
2080                 # to treat the beginning of the template like the beginning
2081                 # of a line for tables and block-level elements.
2082                 $linestart = $matches[1];
2083
2084                 # $part1 is the bit before the first |, and must contain only title characters
2085                 # $args is a list of arguments, starting from index 0, not including $part1
2086
2087                 $part1 = $matches[2];
2088                 # If the third subpattern matched anything, it will start with |
2089
2090                 $args = $this->getTemplateArgs($matches[3]);
2091                 $argc = count( $args );
2092
2093                 # Don't parse {{{}}} because that's only for template arguments
2094                 if ( $linestart === '{' ) {
2095                         $text = $matches[0];
2096                         $found = true;
2097                         $noparse = true;
2098                 }
2099
2100                 # SUBST
2101                 if ( !$found ) {
2102                         $mwSubst =& MagicWord::get( MAG_SUBST );
2103                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor ($this->mOutputType == OT_WIKI) ) {
2104                                 # One of two possibilities is true:
2105                                 # 1) Found SUBST but not in the PST phase
2106                                 # 2) Didn't find SUBST and in the PST phase
2107                                 # In either case, return without further processing
2108                                 $text = $matches[0];
2109                                 $found = true;
2110                                 $noparse = true;
2111                         }
2112                 }
2113
2114                 # MSG, MSGNW and INT
2115                 if ( !$found ) {
2116                         # Check for MSGNW:
2117                         $mwMsgnw =& MagicWord::get( MAG_MSGNW );
2118                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2119                                 $nowiki = true;
2120                         } else {
2121                                 # Remove obsolete MSG:
2122                                 $mwMsg =& MagicWord::get( MAG_MSG );
2123                                 $mwMsg->matchStartAndRemove( $part1 );
2124                         }
2125
2126                         # Check if it is an internal message
2127                         $mwInt =& MagicWord::get( MAG_INT );
2128                         if ( $mwInt->matchStartAndRemove( $part1 ) ) {
2129                                 if ( $this->incrementIncludeCount( 'int:'.$part1 ) ) {
2130                                         $text = $linestart . wfMsgReal( $part1, $args, true );
2131                                         $found = true;
2132                                 }
2133                         }
2134                 }
2135
2136                 # NS
2137                 if ( !$found ) {
2138                         # Check for NS: (namespace expansion)
2139                         $mwNs = MagicWord::get( MAG_NS );
2140                         if ( $mwNs->matchStartAndRemove( $part1 ) ) {
2141                                 if ( intval( $part1 ) ) {
2142                                         $text = $linestart . $wgContLang->getNsText( intval( $part1 ) );
2143                                         $found = true;
2144                                 } else {
2145                                         $index = Namespace::getCanonicalIndex( strtolower( $part1 ) );
2146                                         if ( !is_null( $index ) ) {
2147                                                 $text = $linestart . $wgContLang->getNsText( $index );
2148                                                 $found = true;
2149                                         }
2150                                 }
2151                         }
2152                 }
2153
2154                 # LOCALURL and LOCALURLE
2155                 if ( !$found ) {
2156                         $mwLocal = MagicWord::get( MAG_LOCALURL );
2157                         $mwLocalE = MagicWord::get( MAG_LOCALURLE );
2158
2159                         if ( $mwLocal->matchStartAndRemove( $part1 ) ) {
2160                                 $func = 'getLocalURL';
2161                         } elseif ( $mwLocalE->matchStartAndRemove( $part1 ) ) {
2162                                 $func = 'escapeLocalURL';
2163                         } else {
2164                                 $func = '';
2165                         }
2166
2167                         if ( $func !== '' ) {
2168                                 $title = Title::newFromText( $part1 );
2169                                 if ( !is_null( $title ) ) {
2170                                         if ( $argc > 0 ) {
2171                                                 $text = $linestart . $title->$func( $args[0] );
2172                                         } else {
2173                                                 $text = $linestart . $title->$func();
2174                                         }
2175                                         $found = true;
2176                                 }
2177                         }
2178                 }
2179
2180                 # GRAMMAR
2181                 if ( !$found && $argc == 1 ) {
2182                         $mwGrammar =& MagicWord::get( MAG_GRAMMAR );
2183                         if ( $mwGrammar->matchStartAndRemove( $part1 ) ) {
2184                                 $text = $linestart . $wgContLang->convertGrammar( $args[0], $part1 );
2185                                 $found = true;
2186                         }
2187                 }
2188
2189                 # Template table test
2190
2191                 # Did we encounter this template already? If yes, it is in the cache
2192                 # and we need to check for loops.
2193                 if ( !$found && isset( $this->mTemplates[$part1] ) ) {
2194                         $found = true;
2195
2196                         # Infinite loop test
2197                         if ( isset( $this->mTemplatePath[$part1] ) ) {
2198                                 $noparse = true;
2199                                 $found = true;
2200                                 $text = $linestart .
2201                                         "\{\{$part1}}" .
2202                                         '<!-- WARNING: template loop detected -->';
2203                                 wfDebug( "$fname: template loop broken at '$part1'\n" );
2204                         } else {
2205                                 # set $text to cached message.
2206                                 $text = $linestart . $this->mTemplates[$part1];
2207                         }
2208                 }
2209
2210                 # Load from database
2211                 $replaceHeadings = false;
2212                 $isHTML = false;
2213                 $lastPathLevel = $this->mTemplatePath;
2214                 if ( !$found ) {
2215                         $ns = NS_TEMPLATE;
2216                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage='' );
2217                         if ($subpage !== '') {
2218                                 $ns = $this->mTitle->getNamespace();
2219                         }
2220                         $title = Title::newFromText( $part1, $ns );
2221
2222                         if ($title) {
2223                                 $interwiki = Title::getInterwikiLink($title->getInterwiki());
2224                                 if ($interwiki != '' && $title->isTrans()) {
2225                                         return $this->scarytransclude($title, $interwiki);
2226                                 }
2227                         }
2228
2229                         if ( !is_null( $title ) && !$title->isExternal() ) {
2230                                 # Check for excessive inclusion
2231                                 $dbk = $title->getPrefixedDBkey();
2232                                 if ( $this->incrementIncludeCount( $dbk ) ) {
2233                                         if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() ) {
2234                                                 # Capture special page output
2235                                                 $text = SpecialPage::capturePath( $title );
2236                                                 if ( is_string( $text ) ) {
2237                                                         $found = true;
2238                                                         $noparse = true;
2239                                                         $isHTML = true;
2240                                                         $this->disableCache();
2241                                                 }
2242                                         } else {
2243                                                 $article = new Article( $title );
2244                                                 $articleContent = $article->fetchContent();
2245                                                 if ( $articleContent !== false ) {
2246                                                         $found = true;
2247                                                         $text = $articleContent;
2248                                                         $replaceHeadings = true;
2249                                                 }
2250                                         }
2251                                 }
2252
2253                                 # If the title is valid but undisplayable, make a link to it
2254                                 if ( $this->mOutputType == OT_HTML && !$found ) {
2255                                         $text = '[['.$title->getPrefixedText().']]';
2256                                         $found = true;
2257                                 }
2258
2259                                 # Template cache array insertion
2260                                 if( $found ) {
2261                                         $this->mTemplates[$part1] = $text;
2262                                         $text = $linestart . $text;
2263                                 }
2264                         }
2265                 }
2266
2267                 # Recursive parsing, escaping and link table handling
2268                 # Only for HTML output
2269                 if ( $nowiki && $found && $this->mOutputType == OT_HTML ) {
2270                         $text = wfEscapeWikiText( $text );
2271                 } elseif ( ($this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI) && $found && !$noparse) {
2272                         # Clean up argument array
2273                         $this->mAssocArgs = array();
2274                         $index = 1;
2275                         foreach( $args as $arg ) {
2276                                 $eqpos = strpos( $arg, '=' );
2277                                 if ( $eqpos === false ) {
2278                                         $this->mAssocArgs[$index++] = $arg;
2279                                 } else {
2280                                         $name = trim( substr( $arg, 0, $eqpos ) );
2281                                         $value = trim( substr( $arg, $eqpos+1 ) );
2282                                         if ( $value === false ) {
2283                                                 $value = '';
2284                                         }
2285                                         if ( $name !== false ) {
2286                                                 $this->mAssocArgs[$name] = $value;
2287                                         }
2288                                 }
2289                         }
2290
2291                         # Add a new element to the templace recursion path
2292                         $this->mTemplatePath[$part1] = 1;
2293
2294                         if( $this->mOutputType == OT_HTML ) {
2295                                 # Remove <noinclude> sections and <includeonly> tags
2296                                 $text = preg_replace( '/<noinclude>.*?<\/noinclude>/s', '', $text );
2297                                 $text = strtr( $text, array( '<includeonly>' => '' , '</includeonly>' => '' ) );
2298                                 # Strip <nowiki>, <pre>, etc.
2299                                 $text = $this->strip( $text, $this->mStripState );
2300                                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'replaceVariables' ), $this->mAssocArgs );
2301                         }
2302                         $text = $this->replaceVariables( $text, $this->mAssocArgs );
2303
2304                         # Resume the link cache and register the inclusion as a link
2305                         if ( $this->mOutputType == OT_HTML && !is_null( $title ) ) {
2306                                 $wgLinkCache->addLinkObj( $title );
2307                         }
2308
2309                         # If the template begins with a table or block-level
2310                         # element, it should be treated as beginning a new line.
2311                         if ($linestart !== '\n' && preg_match('/^({\\||:|;|#|\*)/', $text)) {
2312                                 $text = "\n" . $text;
2313                         }
2314                 }
2315                 # Prune lower levels off the recursion check path
2316                 $this->mTemplatePath = $lastPathLevel;
2317
2318                 if ( !$found ) {
2319                         wfProfileOut( $fname );
2320                         return $matches[0];
2321                 } else {
2322                         if ( $isHTML ) {
2323                                 # Replace raw HTML by a placeholder
2324                                 # Add a blank line preceding, to prevent it from mucking up
2325                                 # immediately preceding headings
2326                                 $text = "\n\n" . $this->insertStripItem( $text, $this->mStripState );
2327                         } else {
2328                                 # replace ==section headers==
2329                                 # XXX this needs to go away once we have a better parser.
2330                                 if ( $this->mOutputType != OT_WIKI && $replaceHeadings ) {
2331                                         if( !is_null( $title ) )
2332                                                 $encodedname = base64_encode($title->getPrefixedDBkey());
2333                                         else
2334                                                 $encodedname = base64_encode("");
2335                                         $m = preg_split('/(^={1,6}.*?={1,6}\s*?$)/m', $text, -1,
2336                                                 PREG_SPLIT_DELIM_CAPTURE);
2337                                         $text = '';
2338                                         $nsec = 0;
2339                                         for( $i = 0; $i < count($m); $i += 2 ) {
2340                                                 $text .= $m[$i];
2341                                                 if (!isset($m[$i + 1]) || $m[$i + 1] == "") continue;
2342                                                 $hl = $m[$i + 1];
2343                                                 if( strstr($hl, "<!--MWTEMPLATESECTION") ) {
2344                                                         $text .= $hl;
2345                                                         continue;
2346                                                 }
2347                                                 preg_match('/^(={1,6})(.*?)(={1,6})\s*?$/m', $hl, $m2);
2348                                                 $text .= $m2[1] . $m2[2] . "<!--MWTEMPLATESECTION="
2349                                                         . $encodedname . "&" . base64_encode("$nsec") . "-->" . $m2[3];
2350
2351                                                 $nsec++;
2352                                         }
2353                                 }
2354                         }
2355                 }
2356
2357                 # Prune lower levels off the recursion check path
2358                 $this->mTemplatePath = $lastPathLevel;
2359
2360                 if ( !$found ) {
2361                         wfProfileOut( $fname );
2362                         return $matches[0];
2363                 } else {
2364                         wfProfileOut( $fname );
2365                         return $text;
2366                 }
2367         }
2368
2369         /**
2370          * Translude an interwiki link.
2371          */
2372         function scarytransclude($title, $interwiki) {
2373                 global $wgEnableScaryTranscluding;
2374
2375                 if (!$wgEnableScaryTranscluding)
2376                         return wfMsg('scarytranscludedisabled');
2377
2378                 $articlename = "Template:" . $title->getDBkey();
2379                 $url = str_replace('$1', urlencode($articlename), $interwiki);
2380                 if (strlen($url) > 255)
2381                         return wfMsg('scarytranscludetoolong');
2382                 $text = $this->fetchScaryTemplateMaybeFromCache($url);
2383                 $this->mIWTransData[] = $text;
2384                 return "<!--IW_TRANSCLUDE ".(count($this->mIWTransData) - 1)."-->";
2385         }
2386
2387         function fetchScaryTemplateMaybeFromCache($url) {
2388                 $dbr = wfGetDB(DB_SLAVE);
2389                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
2390                                 array('tc_url' => $url));
2391                 if ($obj) {
2392                         $time = $obj->tc_time;
2393                         $text = $obj->tc_contents;
2394                         if ($time && $time < (time() + (60*60))) {
2395                                 return $text;
2396                         }
2397                 }
2398
2399                 $text = wfGetHTTP($url . '?action=render');
2400                 if (!$text)
2401                         return wfMsg('scarytranscludefailed', $url);
2402
2403                 $dbw = wfGetDB(DB_MASTER);
2404                 $dbw->replace('transcache', array(), array(
2405                         'tc_url' => $url,
2406                         'tc_time' => time(),
2407                         'tc_contents' => $text));
2408                 return $text;
2409         }
2410
2411
2412         /**
2413          * Triple brace replacement -- used for template arguments
2414          * @access private
2415          */
2416         function argSubstitution( $matches ) {
2417                 $arg = trim( $matches[1] );
2418                 $text = $matches[0];
2419                 $inputArgs = end( $this->mArgStack );
2420
2421                 if ( array_key_exists( $arg, $inputArgs ) ) {
2422                         $text = $inputArgs[$arg];
2423                 }
2424
2425                 return $text;
2426         }
2427
2428         /**
2429          * Returns true if the function is allowed to include this entity
2430          * @access private
2431          */
2432         function incrementIncludeCount( $dbk ) {
2433                 if ( !array_key_exists( $dbk, $this->mIncludeCount ) ) {
2434                         $this->mIncludeCount[$dbk] = 0;
2435                 }
2436                 if ( ++$this->mIncludeCount[$dbk] <= MAX_INCLUDE_REPEAT ) {
2437                         return true;
2438                 } else {
2439                         return false;
2440                 }
2441         }
2442
2443         /**
2444          * This function accomplishes several tasks:
2445          * 1) Auto-number headings if that option is enabled
2446          * 2) Add an [edit] link to sections for logged in users who have enabled the option
2447          * 3) Add a Table of contents on the top for users who have enabled the option
2448          * 4) Auto-anchor headings
2449          *
2450          * It loops through all headlines, collects the necessary data, then splits up the
2451          * string and re-inserts the newly formatted headlines.
2452          *
2453          * @param string $text
2454          * @param boolean $isMain
2455          * @access private
2456          */
2457         function formatHeadings( $text, $isMain=true ) {
2458                 global $wgMaxTocLevel, $wgContLang, $wgLinkHolders, $wgInterwikiLinkHolders;
2459
2460                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
2461                 $doShowToc = true;
2462                 $forceTocHere = false;
2463                 if( !$this->mTitle->userCanEdit() ) {
2464                         $showEditLink = 0;
2465                 } else {
2466                         $showEditLink = $this->mOptions->getEditSection();
2467                 }
2468
2469                 # Inhibit editsection links if requested in the page
2470                 $esw =& MagicWord::get( MAG_NOEDITSECTION );
2471                 if( $esw->matchAndRemove( $text ) ) {
2472                         $showEditLink = 0;
2473                 }
2474                 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
2475                 # do not add TOC
2476                 $mw =& MagicWord::get( MAG_NOTOC );
2477                 if( $mw->matchAndRemove( $text ) ) {
2478                         $doShowToc = false;
2479                 }
2480
2481                 # Get all headlines for numbering them and adding funky stuff like [edit]
2482                 # links - this is for later, but we need the number of headlines right now
2483                 $numMatches = preg_match_all( '/<H([1-6])(.*?'.'>)(.*?)<\/H[1-6] *>/i', $text, $matches );
2484
2485                 # if there are fewer than 4 headlines in the article, do not show TOC
2486                 if( $numMatches < 4 ) {
2487                         $doShowToc = false;
2488                 }
2489
2490                 # if the string __TOC__ (not case-sensitive) occurs in the HTML,
2491                 # override above conditions and always show TOC at that place
2492
2493                 $mw =& MagicWord::get( MAG_TOC );
2494                 if($mw->match( $text ) ) {
2495                         $doShowToc = true;
2496                         $forceTocHere = true;
2497                 } else {
2498                         # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
2499                         # override above conditions and always show TOC above first header
2500                         $mw =& MagicWord::get( MAG_FORCETOC );
2501                         if ($mw->matchAndRemove( $text ) ) {
2502                                 $doShowToc = true;
2503                         }
2504                 }
2505
2506                 # Never ever show TOC if no headers
2507                 if( $numMatches < 1 ) {
2508                         $doShowToc = false;
2509                 }
2510
2511                 # We need this to perform operations on the HTML
2512                 $sk =& $this->mOptions->getSkin();
2513
2514                 # headline counter
2515                 $headlineCount = 0;
2516                 $sectionCount = 0; # headlineCount excluding template sections
2517
2518                 # Ugh .. the TOC should have neat indentation levels which can be
2519                 # passed to the skin functions. These are determined here
2520                 $toc = '';
2521                 $full = '';
2522                 $head = array();
2523                 $sublevelCount = array();
2524                 $levelCount = array();
2525                 $toclevel = 0;
2526                 $level = 0;
2527                 $prevlevel = 0;
2528                 $toclevel = 0;
2529                 $prevtoclevel = 0;
2530
2531                 foreach( $matches[3] as $headline ) {
2532                         $istemplate = 0;
2533                         $templatetitle = '';
2534                         $templatesection = 0;
2535                         $numbering = '';
2536
2537                         if (preg_match("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", $headline, $mat)) {
2538                                 $istemplate = 1;
2539                                 $templatetitle = base64_decode($mat[1]);
2540                                 $templatesection = 1 + (int)base64_decode($mat[2]);
2541                                 $headline = preg_replace("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", "", $headline);
2542                         }
2543
2544                         if( $toclevel ) {
2545                                 $prevlevel = $level;
2546                                 $prevtoclevel = $toclevel;
2547                         }
2548                         $level = $matches[1][$headlineCount];
2549
2550                         if( $doNumberHeadings || $doShowToc ) {
2551
2552                                 if ( $level > $prevlevel ) {
2553                                         # Increase TOC level
2554                                         $toclevel++;
2555                                         $sublevelCount[$toclevel] = 0;
2556                                         $toc .= $sk->tocIndent();
2557                                 }
2558                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
2559                                         # Decrease TOC level, find level to jump to
2560
2561                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
2562                                                 # Can only go down to level 1
2563                                                 $toclevel = 1;
2564                                         } else {
2565                                                 for ($i = $toclevel; $i > 0; $i--) {
2566                                                         if ( $levelCount[$i] == $level ) {
2567                                                                 # Found last matching level
2568                                                                 $toclevel = $i;
2569                                                                 break;
2570                                                         }
2571                                                         elseif ( $levelCount[$i] < $level ) {
2572                                                                 # Found first matching level below current level
2573                                                                 $toclevel = $i + 1;
2574                                                                 break;
2575                                                         }
2576                                                 }
2577                                         }
2578
2579                                         $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
2580                                 }
2581                                 else {
2582                                         # No change in level, end TOC line
2583                                         $toc .= $sk->tocLineEnd();
2584                                 }
2585
2586                                 $levelCount[$toclevel] = $level;
2587
2588                                 # count number of headlines for each level
2589                                 @$sublevelCount[$toclevel]++;
2590                                 $dot = 0;
2591                                 for( $i = 1; $i <= $toclevel; $i++ ) {
2592                                         if( !empty( $sublevelCount[$i] ) ) {
2593                                                 if( $dot ) {
2594                                                         $numbering .= '.';
2595                                                 }
2596                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
2597                                                 $dot = 1;
2598                                         }
2599                                 }
2600                         }
2601
2602                         # The canonized header is a version of the header text safe to use for links
2603                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
2604                         $canonized_headline = $this->unstrip( $headline, $this->mStripState );
2605                         $canonized_headline = $this->unstripNoWiki( $canonized_headline, $this->mStripState );
2606
2607                         # Remove link placeholders by the link text.
2608                         #     <!--LINK number-->
2609                         # turns into
2610                         #     link text with suffix
2611                         $canonized_headline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
2612                                                             "\$this->mLinkHolders['texts'][\$1]",
2613                                                             $canonized_headline );
2614                         $canonized_headline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
2615                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
2616                                                             $canonized_headline );
2617
2618                         # strip out HTML
2619                         $canonized_headline = preg_replace( '/<.*?' . '>/','',$canonized_headline );
2620                         $tocline = trim( $canonized_headline );
2621                         $canonized_headline = urlencode( Sanitizer::decodeCharReferences( str_replace(' ', '_', $tocline) ) );
2622                         $replacearray = array(
2623                                 '%3A' => ':',
2624                                 '%' => '.'
2625                         );
2626                         $canonized_headline = str_replace(array_keys($replacearray),array_values($replacearray),$canonized_headline);
2627                         $refers[$headlineCount] = $canonized_headline;
2628
2629                         # count how many in assoc. array so we can track dupes in anchors
2630                         @$refers[$canonized_headline]++;
2631                         $refcount[$headlineCount]=$refers[$canonized_headline];
2632
2633                         # Don't number the heading if it is the only one (looks silly)
2634                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
2635                                 # the two are different if the line contains a link
2636                                 $headline=$numbering . ' ' . $headline;
2637                         }
2638
2639                         # Create the anchor for linking from the TOC to the section
2640                         $anchor = $canonized_headline;
2641                         if($refcount[$headlineCount] > 1 ) {
2642                                 $anchor .= '_' . $refcount[$headlineCount];
2643                         }
2644                         if( $doShowToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
2645                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
2646                         }
2647                         if( $showEditLink && ( !$istemplate || $templatetitle !== "" ) ) {
2648                                 if ( empty( $head[$headlineCount] ) ) {
2649                                         $head[$headlineCount] = '';
2650                                 }
2651                                 if( $istemplate )
2652                                         $head[$headlineCount] .= $sk->editSectionLinkForOther($templatetitle, $templatesection);
2653                                 else
2654                                         $head[$headlineCount] .= $sk->editSectionLink($this->mTitle, $sectionCount+1);
2655                         }
2656
2657                         # give headline the correct <h#> tag
2658                         @$head[$headlineCount] .= "<a name=\"$anchor\"></a><h".$level.$matches[2][$headlineCount] .$headline.'</h'.$level.'>';
2659
2660                         $headlineCount++;
2661                         if( !$istemplate )
2662                                 $sectionCount++;
2663                 }
2664
2665                 if( $doShowToc ) {
2666                         $toc .= $sk->tocUnindent( $toclevel - 1 );
2667                         $toc = $sk->tocList( $toc );
2668                 }
2669
2670                 # split up and insert constructed headlines
2671
2672                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
2673                 $i = 0;
2674
2675                 foreach( $blocks as $block ) {
2676                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
2677                                 # This is the [edit] link that appears for the top block of text when
2678                                 # section editing is enabled
2679
2680                                 # Disabled because it broke block formatting
2681                                 # For example, a bullet point in the top line
2682                                 # $full .= $sk->editSectionLink(0);
2683                         }
2684                         $full .= $block;
2685                         if( $doShowToc && !$i && $isMain && !$forceTocHere) {
2686                         # Top anchor now in skin
2687                                 $full = $full.$toc;
2688                         }
2689
2690                         if( !empty( $head[$i] ) ) {
2691                                 $full .= $head[$i];
2692                         }
2693                         $i++;
2694                 }
2695                 if($forceTocHere) {
2696                         $mw =& MagicWord::get( MAG_TOC );
2697                         return $mw->replace( $toc, $full );
2698                 } else {
2699                         return $full;
2700                 }
2701         }
2702
2703         /**
2704          * Return an HTML link for the "ISBN 123456" text
2705          * @access private
2706          */
2707         function magicISBN( $text ) {
2708                 $fname = 'Parser::magicISBN';
2709                 wfProfileIn( $fname );
2710
2711                 $a = split( 'ISBN ', ' '.$text );
2712                 if ( count ( $a ) < 2 ) {
2713                         wfProfileOut( $fname );
2714                         return $text;
2715                 }
2716                 $text = substr( array_shift( $a ), 1);
2717                 $valid = '0123456789-Xx';
2718
2719                 foreach ( $a as $x ) {
2720                         $isbn = $blank = '' ;
2721                         while ( ' ' == $x{0} ) {
2722                                 $blank .= ' ';
2723                                 $x = substr( $x, 1 );
2724                         }
2725                         if ( $x == '' ) { # blank isbn
2726                                 $text .= "ISBN $blank";
2727                                 continue;
2728                         }
2729                         while ( strstr( $valid, $x{0} ) != false ) {
2730                                 $isbn .= $x{0};
2731                                 $x = substr( $x, 1 );
2732                         }
2733                         $num = str_replace( '-', '', $isbn );
2734                         $num = str_replace( ' ', '', $num );
2735                         $num = str_replace( 'x', 'X', $num );
2736
2737                         if ( '' == $num ) {
2738                                 $text .= "ISBN $blank$x";
2739                         } else {
2740                                 $titleObj = Title::makeTitle( NS_SPECIAL, 'Booksources' );
2741                                 $text .= '<a href="' .
2742                                 $titleObj->escapeLocalUrl( 'isbn='.$num ) .
2743                                         "\" class=\"internal\">ISBN $isbn</a>";
2744                                 $text .= $x;
2745                         }
2746                 }
2747                 wfProfileOut( $fname );
2748                 return $text;
2749         }
2750
2751         /**
2752          * Return an HTML link for the "RFC 1234" text
2753          *
2754          * @access private
2755          * @param string $text     Text to be processed
2756          * @param string $keyword  Magic keyword to use (default RFC)
2757          * @param string $urlmsg   Interface message to use (default rfcurl)
2758          * @return string
2759          */
2760         function magicRFC( $text, $keyword='RFC ', $urlmsg='rfcurl'  ) {
2761
2762                 $valid = '0123456789';
2763                 $internal = false;
2764
2765                 $a = split( $keyword, ' '.$text );
2766                 if ( count ( $a ) < 2 ) {
2767                         return $text;
2768                 }
2769                 $text = substr( array_shift( $a ), 1);
2770
2771                 /* Check if keyword is preceed by [[.
2772                  * This test is made here cause of the array_shift above
2773                  * that prevent the test to be done in the foreach.
2774                  */
2775                 if ( substr( $text, -2 ) == '[[' ) {
2776                         $internal = true;
2777                 }
2778
2779                 foreach ( $a as $x ) {
2780                         /* token might be empty if we have RFC RFC 1234 */
2781                         if ( $x=='' ) {
2782                                 $text.=$keyword;
2783                                 continue;
2784                                 }
2785
2786                         $id = $blank = '' ;
2787
2788                         /** remove and save whitespaces in $blank */
2789                         while ( $x{0} == ' ' ) {
2790                                 $blank .= ' ';
2791                                 $x = substr( $x, 1 );
2792                         }
2793
2794                         /** remove and save the rfc number in $id */
2795                         while ( strstr( $valid, $x{0} ) != false ) {
2796                                 $id .= $x{0};
2797                                 $x = substr( $x, 1 );
2798                         }
2799
2800                         if ( $id == '' ) {
2801                                 /* call back stripped spaces*/
2802                                 $text .= $keyword.$blank.$x;
2803                         } elseif( $internal ) {
2804                                 /* normal link */
2805                                 $text .= $keyword.$id.$x;
2806                         } else {
2807                                 /* build the external link*/
2808                                 $url = wfMsg( $urlmsg, $id);
2809                                 $sk =& $this->mOptions->getSkin();
2810                                 $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
2811                                 $text .= "<a href='{$url}'{$la}>{$keyword}{$id}</a>{$x}";
2812                         }
2813
2814                         /* Check if the next RFC keyword is preceed by [[ */
2815                         $internal = ( substr($x,-2) == '[[' );
2816                 }
2817                 return $text;
2818         }
2819
2820         /**
2821          * Transform wiki markup when saving a page by doing \r\n -> \n
2822          * conversion, substitting signatures, {{subst:}} templates, etc.
2823          *
2824          * @param string $text the text to transform
2825          * @param Title &$title the Title object for the current article
2826          * @param User &$user the User object describing the current user
2827          * @param ParserOptions $options parsing options
2828          * @param bool $clearState whether to clear the parser state first
2829          * @return string the altered wiki markup
2830          * @access public
2831          */
2832         function preSaveTransform( $text, &$title, &$user, $options, $clearState = true ) {
2833                 $this->mOptions = $options;
2834                 $this->mTitle =& $title;
2835                 $this->mOutputType = OT_WIKI;
2836
2837                 if ( $clearState ) {
2838                         $this->clearState();
2839                 }
2840
2841                 $stripState = false;
2842                 $pairs = array(
2843                         "\r\n" => "\n",
2844                 );
2845                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
2846                 $text = $this->strip( $text, $stripState, false );
2847                 $text = $this->pstPass2( $text, $user );
2848                 $text = $this->unstrip( $text, $stripState );
2849                 $text = $this->unstripNoWiki( $text, $stripState );
2850                 return $text;
2851         }
2852
2853         /**
2854          * Pre-save transform helper function
2855          * @access private
2856          */
2857         function pstPass2( $text, &$user ) {
2858                 global $wgContLang, $wgLocaltimezone;
2859
2860                 # Variable replacement
2861                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
2862                 $text = $this->replaceVariables( $text );
2863
2864                 # Signatures
2865                 #
2866                 $n = $user->getName();
2867                 $k = $user->getOption( 'nickname' );
2868                 if ( '' == $k ) { $k = $n; }
2869                 if ( isset( $wgLocaltimezone ) ) {
2870                         $oldtz = getenv( 'TZ' );
2871                         putenv( 'TZ='.$wgLocaltimezone );
2872                 }
2873
2874                 /* Note: This is the timestamp saved as hardcoded wikitext to
2875                  * the database, we use $wgContLang here in order to give
2876                  * everyone the same signiture and use the default one rather
2877                  * than the one selected in each users preferences.
2878                  */
2879                 $d = $wgContLang->timeanddate( date( 'YmdHis' ), false, false) .
2880                   ' (' . date( 'T' ) . ')';
2881                 if ( isset( $wgLocaltimezone ) ) {
2882                         putenv( 'TZ='.$oldtz );
2883                 }
2884
2885                 if( $user->getOption( 'fancysig' ) ) {
2886                         $sigText = $k;
2887                 } else {
2888                         $sigText = '[[' . $wgContLang->getNsText( NS_USER ) . ":$n|$k]]";
2889                 }
2890                 $text = preg_replace( '/~~~~~/', $d, $text );
2891                 $text = preg_replace( '/~~~~/', "$sigText $d", $text );
2892                 $text = preg_replace( '/~~~/', $sigText, $text );
2893
2894                 # Context links: [[|name]] and [[name (context)|]]
2895                 #
2896                 $tc = "[&;%\\-,.\\(\\)' _0-9A-Za-z\\/:\\x80-\\xff]";
2897                 $np = "[&;%\\-,.' _0-9A-Za-z\\/:\\x80-\\xff]"; # No parens
2898                 $namespacechar = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
2899                 $conpat = "/^({$np}+) \\(({$tc}+)\\)$/";
2900
2901                 $p1 = "/\[\[({$np}+) \\(({$np}+)\\)\\|]]/";             # [[page (context)|]]
2902                 $p2 = "/\[\[\\|({$tc}+)]]/";                                    # [[|page]]
2903                 $p3 = "/\[\[(:*$namespacechar+):({$np}+)\\|]]/";                # [[namespace:page|]] and [[:namespace:page|]]
2904                 $p4 = "/\[\[(:*$namespacechar+):({$np}+) \\(({$np}+)\\)\\|]]/"; # [[ns:page (cont)|]] and [[:ns:page (cont)|]]
2905                 $context = '';
2906                 $t = $this->mTitle->getText();
2907                 if ( preg_match( $conpat, $t, $m ) ) {
2908                         $context = $m[2];
2909                 }
2910                 $text = preg_replace( $p4, '[[\\1:\\2 (\\3)|\\2]]', $text );
2911                 $text = preg_replace( $p1, '[[\\1 (\\2)|\\1]]', $text );
2912                 $text = preg_replace( $p3, '[[\\1:\\2|\\2]]', $text );
2913
2914                 if ( '' == $context ) {
2915                         $text = preg_replace( $p2, '[[\\1]]', $text );
2916                 } else {
2917                         $text = preg_replace( $p2, "[[\\1 ({$context})|\\1]]", $text );
2918                 }
2919
2920                 # Trim trailing whitespace
2921                 # MAG_END (__END__) tag allows for trailing
2922                 # whitespace to be deliberately included
2923                 $text = rtrim( $text );
2924                 $mw =& MagicWord::get( MAG_END );
2925                 $mw->matchAndRemove( $text );
2926
2927                 return $text;
2928         }
2929
2930         /**
2931          * Set up some variables which are usually set up in parse()
2932          * so that an external function can call some class members with confidence
2933          * @access public
2934          */
2935         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
2936                 $this->mTitle =& $title;
2937                 $this->mOptions = $options;
2938                 $this->mOutputType = $outputType;
2939                 if ( $clearState ) {
2940                         $this->clearState();
2941                 }
2942         }
2943
2944         /**
2945          * Transform a MediaWiki message by replacing magic variables.
2946          *
2947          * @param string $text the text to transform
2948          * @param ParserOptions $options  options
2949          * @return string the text with variables substituted
2950          * @access public
2951          */
2952         function transformMsg( $text, $options ) {
2953                 global $wgTitle;
2954                 static $executing = false;
2955
2956                 # Guard against infinite recursion
2957                 if ( $executing ) {
2958                         return $text;
2959                 }
2960                 $executing = true;
2961
2962                 $this->mTitle = $wgTitle;
2963                 $this->mOptions = $options;
2964                 $this->mOutputType = OT_MSG;
2965                 $this->clearState();
2966                 $text = $this->replaceVariables( $text );
2967
2968                 $executing = false;
2969                 return $text;
2970         }
2971
2972         /**
2973          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
2974          * Callback will be called with the text within
2975          * Transform and return the text within
2976          * @access public
2977          */
2978         function setHook( $tag, $callback ) {
2979                 $oldVal = @$this->mTagHooks[$tag];
2980                 $this->mTagHooks[$tag] = $callback;
2981                 return $oldVal;
2982         }
2983
2984         /**
2985          * Replace <!--LINK--> link placeholders with actual links, in the buffer
2986          * Placeholders created in Skin::makeLinkObj()
2987          * Returns an array of links found, indexed by PDBK:
2988          *  0 - broken
2989          *  1 - normal link
2990          *  2 - stub
2991          * $options is a bit field, RLH_FOR_UPDATE to select for update
2992          */
2993         function replaceLinkHolders( &$text, $options = 0 ) {
2994                 global $wgUser, $wgLinkCache;
2995                 global $wgOutputReplace;
2996
2997                 $fname = 'Parser::replaceLinkHolders';
2998                 wfProfileIn( $fname );
2999
3000                 $pdbks = array();
3001                 $colours = array();
3002                 $sk = $this->mOptions->getSkin();
3003
3004                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
3005                         wfProfileIn( $fname.'-check' );
3006                         $dbr =& wfGetDB( DB_SLAVE );
3007                         $page = $dbr->tableName( 'page' );
3008                         $threshold = $wgUser->getOption('stubthreshold');
3009
3010                         # Sort by namespace
3011                         asort( $this->mLinkHolders['namespaces'] );
3012
3013                         # Generate query
3014                         $query = false;
3015                         foreach ( $this->mLinkHolders['namespaces'] as $key => $val ) {
3016                                 # Make title object
3017                                 $title = $this->mLinkHolders['titles'][$key];
3018
3019                                 # Skip invalid entries.
3020                                 # Result will be ugly, but prevents crash.
3021                                 if ( is_null( $title ) ) {
3022                                         continue;
3023                                 }
3024                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
3025
3026                                 # Check if it's in the link cache already
3027                                 if ( $wgLinkCache->getGoodLinkID( $pdbk ) ) {
3028                                         $colours[$pdbk] = 1;
3029                                 } elseif ( $wgLinkCache->isBadLink( $pdbk ) ) {
3030                                         $colours[$pdbk] = 0;
3031                                 } else {
3032                                         # Not in the link cache, add it to the query
3033                                         if ( !isset( $current ) ) {
3034                                                 $current = $val;
3035                                                 $query =  "SELECT page_id, page_namespace, page_title";
3036                                                 if ( $threshold > 0 ) {
3037                                                         $query .= ', page_len, page_is_redirect';
3038                                                 }
3039                                                 $query .= " FROM $page WHERE (page_namespace=$val AND page_title IN(";
3040                                         } elseif ( $current != $val ) {
3041                                                 $current = $val;
3042                                                 $query .= ")) OR (page_namespace=$val AND page_title IN(";
3043                                         } else {
3044                                                 $query .= ', ';
3045                                         }
3046
3047                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
3048                                 }
3049                         }
3050                         if ( $query ) {
3051                                 $query .= '))';
3052                                 if ( $options & RLH_FOR_UPDATE ) {
3053                                         $query .= ' FOR UPDATE';
3054                                 }
3055
3056                                 $res = $dbr->query( $query, $fname );
3057
3058                                 # Fetch data and form into an associative array
3059                                 # non-existent = broken
3060                                 # 1 = known
3061                                 # 2 = stub
3062                                 while ( $s = $dbr->fetchObject($res) ) {
3063                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
3064                                         $pdbk = $title->getPrefixedDBkey();
3065                                         $wgLinkCache->addGoodLinkObj( $s->page_id, $title );
3066
3067                                         if ( $threshold >  0 ) {
3068                                                 $size = $s->page_len;
3069                                                 if ( $s->page_is_redirect || $s->page_namespace != 0 || $size >= $threshold ) {
3070                                                         $colours[$pdbk] = 1;
3071                                                 } else {
3072                                                         $colours[$pdbk] = 2;
3073                                                 }
3074                                         } else {
3075                                                 $colours[$pdbk] = 1;
3076                                         }
3077                                 }
3078                         }
3079                         wfProfileOut( $fname.'-check' );
3080
3081                         # Construct search and replace arrays
3082                         wfProfileIn( $fname.'-construct' );
3083                         $wgOutputReplace = array();
3084                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3085                                 $pdbk = $pdbks[$key];
3086                                 $searchkey = "<!--LINK $key-->";
3087                                 $title = $this->mLinkHolders['titles'][$key];
3088                                 if ( empty( $colours[$pdbk] ) ) {
3089                                         $wgLinkCache->addBadLinkObj( $title );
3090                                         $colours[$pdbk] = 0;
3091                                         $wgOutputReplace[$searchkey] = $sk->makeBrokenLinkObj( $title,
3092                                                                         $this->mLinkHolders['texts'][$key],
3093                                                                         $this->mLinkHolders['queries'][$key] );
3094                                 } elseif ( $colours[$pdbk] == 1 ) {
3095                                         $wgOutputReplace[$searchkey] = $sk->makeKnownLinkObj( $title,
3096                                                                         $this->mLinkHolders['texts'][$key],
3097                                                                         $this->mLinkHolders['queries'][$key] );
3098                                 } elseif ( $colours[$pdbk] == 2 ) {
3099                                         $wgOutputReplace[$searchkey] = $sk->makeStubLinkObj( $title,
3100                                                                         $this->mLinkHolders['texts'][$key],
3101                                                                         $this->mLinkHolders['queries'][$key] );
3102                                 }
3103                         }
3104                         wfProfileOut( $fname.'-construct' );
3105
3106                         # Do the thing
3107                         wfProfileIn( $fname.'-replace' );
3108
3109                         $text = preg_replace_callback(
3110                                 '/(<!--LINK .*?-->)/',
3111                                 "wfOutputReplaceMatches",
3112                                 $text);
3113
3114                         wfProfileOut( $fname.'-replace' );
3115                 }
3116
3117                 # Now process interwiki link holders
3118                 # This is quite a bit simpler than internal links
3119                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
3120                         wfProfileIn( $fname.'-interwiki' );
3121                         # Make interwiki link HTML
3122                         $wgOutputReplace = array();
3123                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
3124                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
3125                                 $wgOutputReplace[$key] = $sk->makeLinkObj( $title, $link );
3126                         }
3127
3128                         $text = preg_replace_callback(
3129                                 '/<!--IWLINK (.*?)-->/',
3130                                 "wfOutputReplaceMatches",
3131                                 $text );
3132                         wfProfileOut( $fname.'-interwiki' );
3133                 }
3134
3135                 wfProfileOut( $fname );
3136                 return $colours;
3137         }
3138
3139         /**
3140          * Replace <!--LINK--> link placeholders with plain text of links
3141          * (not HTML-formatted).
3142          * @param string $text
3143          * @return string
3144          */
3145         function replaceLinkHoldersText( $text ) {
3146                 global $wgUser, $wgLinkCache;
3147                 global $wgOutputReplace;
3148
3149                 $fname = 'Parser::replaceLinkHoldersText';
3150                 wfProfileIn( $fname );
3151
3152                 $text = preg_replace_callback(
3153                         '/<!--(LINK|IWLINK) (.*?)-->/',
3154                         array( &$this, 'replaceLinkHoldersTextCallback' ),
3155                         $text );
3156
3157                 wfProfileOut( $fname );
3158                 return $text;
3159         }
3160
3161         /**
3162          * @param array $matches
3163          * @return string
3164          * @access private
3165          */
3166         function replaceLinkHoldersTextCallback( $matches ) {
3167                 $type = $matches[1];
3168                 $key  = $matches[2];
3169                 if( $type == 'LINK' ) {
3170                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
3171                                 return $this->mLinkHolders['texts'][$key];
3172                         }
3173                 } elseif( $type == 'IWLINK' ) {
3174                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
3175                                 return $this->mInterwikiLinkHolders['texts'][$key];
3176                         }
3177                 }
3178                 return $matches[0];
3179         }
3180
3181         /**
3182          * Renders an image gallery from a text with one line per image.
3183          * text labels may be given by using |-style alternative text. E.g.
3184          *   Image:one.jpg|The number "1"
3185          *   Image:tree.jpg|A tree
3186          * given as text will return the HTML of a gallery with two images,
3187          * labeled 'The number "1"' and
3188          * 'A tree'.
3189          *
3190          * @static
3191          */
3192         function renderImageGallery( $text ) {
3193                 # Setup the parser
3194                 global $wgUser, $wgTitle;
3195                 $parserOptions = ParserOptions::newFromUser( $wgUser );
3196                 $localParser = new Parser();
3197
3198                 global $wgLinkCache;
3199                 $ig = new ImageGallery();
3200                 $ig->setShowBytes( false );
3201                 $ig->setShowFilename( false );
3202                 $lines = explode( "\n", $text );
3203
3204                 foreach ( $lines as $line ) {
3205                         # match lines like these:
3206                         # Image:someimage.jpg|This is some image
3207                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
3208                         # Skip empty lines
3209                         if ( count( $matches ) == 0 ) {
3210                                 continue;
3211                         }
3212                         $nt = Title::newFromURL( $matches[1] );
3213                         if( is_null( $nt ) ) {
3214                                 # Bogus title. Ignore these so we don't bomb out later.
3215                                 continue;
3216                         }
3217                         if ( isset( $matches[3] ) ) {
3218                                 $label = $matches[3];
3219                         } else {
3220                                 $label = '';
3221                         }
3222
3223                         $html = $localParser->parse( $label , $wgTitle, $parserOptions );
3224                         $html = $html->mText;
3225
3226                         $ig->add( new Image( $nt ), $html );
3227                         $wgLinkCache->addImageLinkObj( $nt );
3228                 }
3229                 return $ig->toHTML();
3230         }
3231
3232         /**
3233          * Parse image options text and use it to make an image
3234          */
3235         function makeImage( &$nt, $options ) {
3236                 global $wgContLang, $wgUseImageResize;
3237                 global $wgUser, $wgThumbLimits;
3238
3239                 $align = '';
3240
3241                 # Check if the options text is of the form "options|alt text"
3242                 # Options are:
3243                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
3244                 #  * left               no resizing, just left align. label is used for alt= only
3245                 #  * right              same, but right aligned
3246                 #  * none               same, but not aligned
3247                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
3248                 #  * center             center the image
3249                 #  * framed             Keep original image size, no magnify-button.
3250
3251                 $part = explode( '|', $options);
3252
3253                 $mwThumb  =& MagicWord::get( MAG_IMG_THUMBNAIL );
3254                 $mwLeft   =& MagicWord::get( MAG_IMG_LEFT );
3255                 $mwRight  =& MagicWord::get( MAG_IMG_RIGHT );
3256                 $mwNone   =& MagicWord::get( MAG_IMG_NONE );
3257                 $mwWidth  =& MagicWord::get( MAG_IMG_WIDTH );
3258                 $mwCenter =& MagicWord::get( MAG_IMG_CENTER );
3259                 $mwFramed =& MagicWord::get( MAG_IMG_FRAMED );
3260                 $caption = '';
3261
3262                 $width = $height = $framed = $thumb = false;
3263                 $manual_thumb = '' ;
3264
3265                 foreach( $part as $key => $val ) {
3266                         $val_parts = explode ( '=' , $val , 2 ) ;
3267                         $left_part = array_shift ( $val_parts ) ;
3268                         if ( $wgUseImageResize && ! is_null( $mwThumb->matchVariableStartToEnd($val) ) ) {
3269                                 $thumb=true;
3270                         } elseif ( $wgUseImageResize && count ( $val_parts ) == 1 && ! is_null( $mwThumb->matchVariableStartToEnd($left_part) ) ) {
3271                                 # use manually specified thumbnail
3272                                 $thumb=true;
3273                                 $manual_thumb = array_shift ( $val_parts ) ;
3274                         } elseif ( ! is_null( $mwRight->matchVariableStartToEnd($val) ) ) {
3275                                 # remember to set an alignment, don't render immediately
3276                                 $align = 'right';
3277                         } elseif ( ! is_null( $mwLeft->matchVariableStartToEnd($val) ) ) {
3278                                 # remember to set an alignment, don't render immediately
3279                                 $align = 'left';
3280                         } elseif ( ! is_null( $mwCenter->matchVariableStartToEnd($val) ) ) {
3281                                 # remember to set an alignment, don't render immediately
3282                                 $align = 'center';
3283                         } elseif ( ! is_null( $mwNone->matchVariableStartToEnd($val) ) ) {
3284                                 # remember to set an alignment, don't render immediately
3285                                 $align = 'none';
3286                         } elseif ( $wgUseImageResize && ! is_null( $match = $mwWidth->matchVariableStartToEnd($val) ) ) {
3287                                 wfDebug( "MAG_IMG_WIDTH match: $match\n" );
3288                                 # $match is the image width in pixels
3289                                 if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $match, $m ) ) {
3290                                         $width = intval( $m[1] );
3291                                         $height = intval( $m[2] );
3292                                 } else {
3293                                         $width = intval($match);
3294                                 }
3295                         } elseif ( ! is_null( $mwFramed->matchVariableStartToEnd($val) ) ) {
3296                                 $framed=true;
3297                         } else {
3298                                 $caption = $val;
3299                         }
3300                 }
3301                 # Strip bad stuff out of the alt text
3302                 $alt = $this->replaceLinkHoldersText( $caption );
3303                 $alt = Sanitizer::stripAllTags( $alt );
3304
3305                 # Linker does the rest
3306                 $sk =& $this->mOptions->getSkin();
3307                 return $sk->makeImageLinkObj( $nt, $caption, $alt, $align, $width, $height, $framed, $thumb, $manual_thumb );
3308         }
3309
3310         /**
3311          * Set a flag in the output object indicating that the content is dynamic and
3312          * shouldn't be cached.
3313          */
3314         function disableCache() {
3315                 $this->mOutput->mCacheTime = -1;
3316         }
3317 }
3318
3319 /**
3320  * @todo document
3321  * @package MediaWiki
3322  */
3323 class ParserOutput
3324 {
3325         var $mText, $mLanguageLinks, $mCategoryLinks, $mContainsOldMagic;
3326         var $mCacheTime; # Timestamp on this article, or -1 for uncacheable. Used in ParserCache.
3327         var $mVersion;   # Compatibility check
3328         var $mTitleText; # title text of the chosen language variant
3329
3330         function ParserOutput( $text = '', $languageLinks = array(), $categoryLinks = array(),
3331                 $containsOldMagic = false, $titletext = '' )
3332         {
3333                 $this->mText = $text;
3334                 $this->mLanguageLinks = $languageLinks;
3335                 $this->mCategoryLinks = $categoryLinks;
3336                 $this->mContainsOldMagic = $containsOldMagic;
3337                 $this->mCacheTime = '';
3338                 $this->mVersion = MW_PARSER_VERSION;
3339                 $this->mTitleText = $titletext;
3340         }
3341
3342         function getText()                   { return $this->mText; }
3343         function getLanguageLinks()          { return $this->mLanguageLinks; }
3344         function getCategoryLinks()          { return array_keys( $this->mCategoryLinks ); }
3345         function getCacheTime()              { return $this->mCacheTime; }
3346         function getTitleText()              { return $this->mTitleText; }
3347         function containsOldMagic()          { return $this->mContainsOldMagic; }
3348         function setText( $text )            { return wfSetVar( $this->mText, $text ); }
3349         function setLanguageLinks( $ll )     { return wfSetVar( $this->mLanguageLinks, $ll ); }
3350         function setCategoryLinks( $cl )     { return wfSetVar( $this->mCategoryLinks, $cl ); }
3351         function setContainsOldMagic( $com ) { return wfSetVar( $this->mContainsOldMagic, $com ); }
3352         function setCacheTime( $t )          { return wfSetVar( $this->mCacheTime, $t ); }
3353         function setTitleText( $t )          { return wfSetVar ($this->mTitleText, $t); }
3354
3355         function addCategoryLink( $c )       { $this->mCategoryLinks[$c] = 1; }
3356
3357         function merge( $other ) {
3358                 $this->mLanguageLinks = array_merge( $this->mLanguageLinks, $other->mLanguageLinks );
3359                 $this->mCategoryLinks = array_merge( $this->mCategoryLinks, $this->mLanguageLinks );
3360                 $this->mContainsOldMagic = $this->mContainsOldMagic || $other->mContainsOldMagic;
3361         }
3362
3363         /**
3364          * Return true if this cached output object predates the global or
3365          * per-article cache invalidation timestamps, or if it comes from
3366          * an incompatible older version.
3367          *
3368          * @param string $touched the affected article's last touched timestamp
3369          * @return bool
3370          * @access public
3371          */
3372         function expired( $touched ) {
3373                 global $wgCacheEpoch;
3374                 return $this->getCacheTime() == -1 || // parser says it's uncacheable
3375                        $this->getCacheTime() <= $touched ||
3376                        $this->getCacheTime() <= $wgCacheEpoch ||
3377                        !isset( $this->mVersion ) ||
3378                        version_compare( $this->mVersion, MW_PARSER_VERSION, "lt" );
3379         }
3380 }
3381
3382 /**
3383  * Set options of the Parser
3384  * @todo document
3385  * @package MediaWiki
3386  */
3387 class ParserOptions
3388 {
3389         # All variables are private
3390         var $mUseTeX;                    # Use texvc to expand <math> tags
3391         var $mUseDynamicDates;           # Use DateFormatter to format dates
3392         var $mInterwikiMagic;            # Interlanguage links are removed and returned in an array
3393         var $mAllowExternalImages;       # Allow external images inline
3394         var $mSkin;                      # Reference to the preferred skin
3395         var $mDateFormat;                # Date format index
3396         var $mEditSection;               # Create "edit section" links
3397         var $mNumberHeadings;            # Automatically number headings
3398         var $mAllowSpecialInclusion;     # Allow inclusion of special pages
3399
3400         function getUseTeX()                        { return $this->mUseTeX; }
3401         function getUseDynamicDates()               { return $this->mUseDynamicDates; }
3402         function getInterwikiMagic()                { return $this->mInterwikiMagic; }
3403         function getAllowExternalImages()           { return $this->mAllowExternalImages; }
3404         function &getSkin()                         { return $this->mSkin; }
3405         function getDateFormat()                    { return $this->mDateFormat; }
3406         function getEditSection()                   { return $this->mEditSection; }
3407         function getNumberHeadings()                { return $this->mNumberHeadings; }
3408         function getAllowSpecialInclusion()         { return $this->mAllowSpecialInclusion; }
3409
3410
3411         function setUseTeX( $x )                    { return wfSetVar( $this->mUseTeX, $x ); }
3412         function setUseDynamicDates( $x )           { return wfSetVar( $this->mUseDynamicDates, $x ); }
3413         function setInterwikiMagic( $x )            { return wfSetVar( $this->mInterwikiMagic, $x ); }
3414         function setAllowExternalImages( $x )       { return wfSetVar( $this->mAllowExternalImages, $x ); }
3415         function setDateFormat( $x )                { return wfSetVar( $this->mDateFormat, $x ); }
3416         function setEditSection( $x )               { return wfSetVar( $this->mEditSection, $x ); }
3417         function setNumberHeadings( $x )            { return wfSetVar( $this->mNumberHeadings, $x ); }
3418         function setAllowSpecialInclusion( $x )     { return wfSetVar( $this->mAllowSpecialInclusion, $x ); }
3419
3420         function setSkin( &$x ) { $this->mSkin =& $x; }
3421
3422         function ParserOptions() {
3423                 global $wgUser;
3424                 $this->initialiseFromUser( $wgUser );
3425         }
3426
3427         /**
3428          * Get parser options
3429          * @static
3430          */
3431         function newFromUser( &$user ) {
3432                 $popts = new ParserOptions;
3433                 $popts->initialiseFromUser( $user );
3434                 return $popts;
3435         }
3436
3437         /** Get user options */
3438         function initialiseFromUser( &$userInput ) {
3439                 global $wgUseTeX, $wgUseDynamicDates, $wgInterwikiMagic, $wgAllowExternalImages,
3440                        $wgAllowSpecialInclusion;
3441                 $fname = 'ParserOptions::initialiseFromUser';
3442                 wfProfileIn( $fname );
3443                 if ( !$userInput ) {
3444                         $user = new User;
3445                         $user->setLoaded( true );
3446                 } else {
3447                         $user =& $userInput;
3448                 }
3449
3450                 $this->mUseTeX = $wgUseTeX;
3451                 $this->mUseDynamicDates = $wgUseDynamicDates;
3452                 $this->mInterwikiMagic = $wgInterwikiMagic;
3453                 $this->mAllowExternalImages = $wgAllowExternalImages;
3454                 wfProfileIn( $fname.'-skin' );
3455                 $this->mSkin =& $user->getSkin();
3456                 wfProfileOut( $fname.'-skin' );
3457                 $this->mDateFormat = $user->getOption( 'date' );
3458                 $this->mEditSection = true;
3459                 $this->mNumberHeadings = $user->getOption( 'numberheadings' );
3460                 $this->mAllowSpecialInclusion = $wgAllowSpecialInclusion;
3461                 wfProfileOut( $fname );
3462         }
3463 }
3464
3465 /**
3466  * Callback function used by Parser::replaceLinkHolders()
3467  * to substitute link placeholders.
3468  */
3469 function &wfOutputReplaceMatches( $matches ) {
3470         global $wgOutputReplace;
3471         return $wgOutputReplace[$matches[1]];
3472 }
3473
3474 /**
3475  * Return the total number of articles
3476  */
3477 function wfNumberOfArticles() {
3478         global $wgNumberOfArticles;
3479
3480         wfLoadSiteStats();
3481         return $wgNumberOfArticles;
3482 }
3483
3484 /**
3485  * Return the number of files
3486  */
3487 function wfNumberOfFiles() {
3488         $fname = 'Parser::wfNumberOfFiles';
3489
3490         wfProfileIn( $fname );
3491         $dbr =& wfGetDB( DB_SLAVE );
3492         $res = $dbr->selectField('image', 'COUNT(*)', array(), $fname );
3493         wfProfileOut( $fname );
3494
3495         return $res;
3496 }
3497
3498 /**
3499  * Get various statistics from the database
3500  * @private
3501  */
3502 function wfLoadSiteStats() {
3503         global $wgNumberOfArticles, $wgTotalViews, $wgTotalEdits;
3504         $fname = 'wfLoadSiteStats';
3505
3506         if ( -1 != $wgNumberOfArticles ) return;
3507         $dbr =& wfGetDB( DB_SLAVE );
3508         $s = $dbr->selectRow( 'site_stats',
3509                 array( 'ss_total_views', 'ss_total_edits', 'ss_good_articles' ),
3510                 array( 'ss_row_id' => 1 ), $fname
3511         );
3512
3513         if ( $s === false ) {
3514                 return;
3515         } else {
3516                 $wgTotalViews = $s->ss_total_views;
3517                 $wgTotalEdits = $s->ss_total_edits;
3518                 $wgNumberOfArticles = $s->ss_good_articles;
3519         }
3520 }
3521
3522 /**
3523  * Escape html tags
3524  * Basicly replacing " > and < with HTML entities ( &quot;, &gt;, &lt;)
3525  *
3526  * @param string $in Text that might contain HTML tags
3527  * @return string Escaped string
3528  */
3529 function wfEscapeHTMLTagsOnly( $in ) {
3530         return str_replace(
3531                 array( '"', '>', '<' ),
3532                 array( '&quot;', '&gt;', '&lt;' ),
3533                 $in );
3534 }
3535
3536 ?>