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', '1.6.0' );
  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 # Constants needed for external link processing
  47 define( 'HTTP_PROTOCOLS', 'http:\/\/|https:\/\/' );
  48 # Everything except bracket, space, or control characters
  49 define( 'EXT_LINK_URL_CLASS', '[^][<>"\\x00-\\x20\\x7F]' );
  50 # Including space
  51 define( 'EXT_LINK_TEXT_CLASS', '[^\]\\x00-\\x1F\\x7F]' );
  52 define( 'EXT_IMAGE_FNAME_CLASS', '[A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]' );
  53 define( 'EXT_IMAGE_EXTENSIONS', 'gif|png|jpg|jpeg' );
  54 define( 'EXT_LINK_BRACKETED',  '/\[(\b(' . wfUrlProtocols() . ')'.EXT_LINK_URL_CLASS.'+) *('.EXT_LINK_TEXT_CLASS.'*?)\]/S' );
  55 define( 'EXT_IMAGE_REGEX',
  56         '/^('.HTTP_PROTOCOLS.')'.  # Protocol
  57         '('.EXT_LINK_URL_CLASS.'+)\\/'.  # Hostname and path
  58         '('.EXT_IMAGE_FNAME_CLASS.'+)\\.((?i)'.EXT_IMAGE_EXTENSIONS.')$/S' # Filename
  59 );
  60
  61 /**
  62  * PHP Parser
  63  *
  64  * Processes wiki markup
  65  *
  66  * <pre>
  67  * There are three main entry points into the Parser class:
  68  * parse()
  69  *   produces HTML output
  70  * preSaveTransform().
  71  *   produces altered wiki markup.
  72  * transformMsg()
  73  *   performs brace substitution on MediaWiki messages
  74  *
  75  * Globals used:
  76  *    objects:   $wgLang
  77  *
  78  * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
  79  *
  80  * settings:
  81  *  $wgUseTex*, $wgUseDynamicDates*, $wgInterwikiMagic*,
  82  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  83  *  $wgLocaltimezone, $wgAllowSpecialInclusion*
  84  *
  85  *  * only within ParserOptions
  86  * </pre>
  87  *
  88  * @package MediaWiki
  89  */
  90 class Parser
  91 {
  92         /**#@+
  93          * @access private
  94          */
  95         # Persistent:
  96         var $mTagHooks;
  97
  98         # Cleared with clearState():
  99         var $mOutput, $mAutonumber, $mDTopen, $mStripState = array();
 100         var $mVariables, $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 101         var $mInterwikiLinkHolders, $mLinkHolders, $mUniqPrefix;
 102         var $mTemplates,        // cache of already loaded templates, avoids
 103                                 // multiple SQL queries for the same string
 104             $mTemplatePath;     // stores an unsorted hash of all the templates already loaded
 105                                 // in this path. Used for loop detection.
 106
 107         # Temporary
 108         # These are variables reset at least once per parse regardless of $clearState
 109         var $mOptions,      // ParserOptions object
 110                 $mTitle,        // Title context, used for self-link rendering and similar things
 111                 $mOutputType,   // Output type, one of the OT_xxx constants
 112                 $mRevisionId;   // ID to display in {{REVISIONID}} tags
 113
 114         /**#@-*/
 115
 116         /**
 117          * Constructor
 118          *
 119          * @access public
 120          */
 121         function Parser() {
 122                 $this->mTagHooks = array();
 123                 $this->clearState();
 124         }
 125
 126         /**
 127          * Clear Parser state
 128          *
 129          * @access private
 130          */
 131         function clearState() {
 132                 $this->mOutput = new ParserOutput;
 133                 $this->mAutonumber = 0;
 134                 $this->mLastSection = '';
 135                 $this->mDTopen = false;
 136                 $this->mVariables = false;
 137                 $this->mIncludeCount = array();
 138                 $this->mStripState = array();
 139                 $this->mArgStack = array();
 140                 $this->mInPre = false;
 141                 $this->mInterwikiLinkHolders = array(
 142                         'texts' => array(),
 143                         'titles' => array()
 144                 );
 145                 $this->mLinkHolders = array(
 146                         'namespaces' => array(),
 147                         'dbkeys' => array(),
 148                         'queries' => array(),
 149                         'texts' => array(),
 150                         'titles' => array()
 151                 );
 152                 $this->mRevisionId = null;
 153                 $this->mUniqPrefix = 'UNIQ' . Parser::getRandomString();
 154
 155                 # Clear these on every parse, bug 4549
 156                 $this->mTemplates = array();
 157                 $this->mTemplatePath = array();
 158
 159                 wfRunHooks( 'ParserClearState', array( &$this ) );
 160         }
 161
 162         /**
 163          * Accessor for mUniqPrefix.
 164          *
 165          * @access public
 166          */
 167         function UniqPrefix() {
 168                 return $this->mUniqPrefix;
 169         }
 170
 171         /**
 172          * Convert wikitext to HTML
 173          * Do not call this function recursively.
 174          *
 175          * @access private
 176          * @param string $text Text we want to parse
 177          * @param Title &$title A title object
 178          * @param array $options
 179          * @param boolean $linestart
 180          * @param boolean $clearState
 181          * @param int $revid number to pass in {{REVISIONID}}
 182          * @return ParserOutput a ParserOutput
 183          */
 184         function parse( $text, &$title, $options, $linestart = true, $clearState = true, $revid = null ) {
 185                 /**
 186                  * First pass--just handle <nowiki> sections, pass the rest off
 187                  * to internalParse() which does all the real work.
 188                  */
 189
 190                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
 191                 $fname = 'Parser::parse';
 192                 wfProfileIn( $fname );
 193
 194                 if ( $clearState ) {
 195                         $this->clearState();
 196                 }
 197
 198                 $this->mOptions = $options;
 199                 $this->mTitle =& $title;
 200                 $this->mRevisionId = $revid;
 201                 $this->mOutputType = OT_HTML;
 202
 203                 $this->mStripState = NULL;
 204
 205                 //$text = $this->strip( $text, $this->mStripState );
 206                 // VOODOO MAGIC FIX! Sometimes the above segfaults in PHP5.
 207                 $x =& $this->mStripState;
 208
 209                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$x ) );
 210                 $text = $this->strip( $text, $x );
 211                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$x ) );
 212
 213                 # Hook to suspend the parser in this state
 214                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$x ) ) ) {
 215                         wfProfileOut( $fname );
 216                         return $text ;
 217                 }
 218
 219                 $text = $this->internalParse( $text );
 220
 221                 $text = $this->unstrip( $text, $this->mStripState );
 222
 223                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 224                 $fixtags = array(
 225                         # french spaces, last one Guillemet-left
 226                         # only if there is something before the space
 227                         '/(.) (?=\\?|:|;|!|\\302\\273)/' => '\\1&nbsp;\\2',
 228                         # french spaces, Guillemet-right
 229                         '/(\\302\\253) /' => '\\1&nbsp;',
 230                         '/<center *>(.*)<\\/center *>/i' => '<div class="center">\\1</div>',
 231                 );
 232                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 233
 234                 # only once and last
 235                 $text = $this->doBlockLevels( $text, $linestart );
 236
 237                 $this->replaceLinkHolders( $text );
 238
 239                 # the position of the convert() call should not be changed. it
 240                 # assumes that the links are all replaces and the only thing left
 241                 # is the <nowiki> mark.
 242                 $text = $wgContLang->convert($text);
 243                 $this->mOutput->setTitleText($wgContLang->getParsedTitle());
 244
 245                 $text = $this->unstripNoWiki( $text, $this->mStripState );
 246
 247                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 248
 249                 $text = Sanitizer::normalizeCharReferences( $text );
 250
 251                 if (($wgUseTidy and $this->mOptions->mTidy) or $wgAlwaysUseTidy) {
 252                         $text = Parser::tidy($text);
 253                 }
 254
 255                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 256
 257                 $this->mOutput->setText( $text );
 258                 wfProfileOut( $fname );
 259
 260                 return $this->mOutput;
 261         }
 262
 263         /**
 264          * Get a random string
 265          *
 266          * @access private
 267          * @static
 268          */
 269         function getRandomString() {
 270                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 271         }
 272
 273         /**
 274          * Replaces all occurrences of <$tag>content</$tag> in the text
 275          * with a random marker and returns the new text. the output parameter
 276          * $content will be an associative array filled with data on the form
 277          * $unique_marker => content.
 278          *
 279          * If $content is already set, the additional entries will be appended
 280          * If $tag is set to STRIP_COMMENTS, the function will extract
 281          * <!-- HTML comments -->
 282          *
 283          * @access private
 284          * @static
 285          */
 286         function extractTagsAndParams($tag, $text, &$content, &$tags, &$params, $uniq_prefix = ''){
 287                 $rnd = $uniq_prefix . '-' . $tag . Parser::getRandomString();
 288                 if ( !$content ) {
 289                         $content = array( );
 290                 }
 291                 $n = 1;
 292                 $stripped = '';
 293
 294                 if ( !$tags ) {
 295                         $tags = array( );
 296                 }
 297
 298                 if ( !$params ) {
 299                         $params = array( );
 300                 }
 301
 302                 if( $tag == STRIP_COMMENTS ) {
 303                         $start = '/<!--()()/';
 304                         $end   = '/-->/';
 305                 } else {
 306                         $start = "/<$tag(\\s+[^\\/>]*|\\s*)(\\/?)>/i";
 307                         $end   = "/<\\/$tag\\s*>/i";
 308                 }
 309
 310                 while ( '' != $text ) {
 311                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 312                         $stripped .= $p[0];
 313                         if( count( $p ) < 4 ) {
 314                                 break;
 315                         }
 316                         $attributes = $p[1];
 317                         $empty      = $p[2];
 318                         $inside     = $p[3];
 319
 320                         $marker = $rnd . sprintf('%08X', $n++);
 321                         $stripped .= $marker;
 322
 323                         $tags[$marker] = "<$tag$attributes$empty>";
 324                         $params[$marker] = Sanitizer::decodeTagAttributes( $attributes );
 325
 326                         if ( $empty === '/' ) {
 327                                 // Empty element tag, <tag />
 328                                 $content[$marker] = null;
 329                                 $text = $inside;
 330                         } else {
 331                                 $q = preg_split( $end, $inside, 2 );
 332                                 $content[$marker] = $q[0];
 333                                 if( count( $q ) < 2 ) {
 334                                         # No end tag -- let it run out to the end of the text.
 335                                         break;
 336                                 } else {
 337                                         $text = $q[1];
 338                                 }
 339                         }
 340                 }
 341                 return $stripped;
 342         }
 343
 344         /**
 345          * Wrapper function for extractTagsAndParams
 346          * for cases where $tags and $params isn't needed
 347          * i.e. where tags will never have params, like <nowiki>
 348          *
 349          * @access private
 350          * @static
 351          */
 352         function extractTags( $tag, $text, &$content, $uniq_prefix = '' ) {
 353                 $dummy_tags = array();
 354                 $dummy_params = array();
 355
 356                 return Parser::extractTagsAndParams( $tag, $text, $content,
 357                         $dummy_tags, $dummy_params, $uniq_prefix );
 358         }
 359
 360         /**
 361          * Strips and renders nowiki, pre, math, hiero
 362          * If $render is set, performs necessary rendering operations on plugins
 363          * Returns the text, and fills an array with data needed in unstrip()
 364          * If the $state is already a valid strip state, it adds to the state
 365          *
 366          * @param bool $stripcomments when set, HTML comments <!-- like this -->
 367          *  will be stripped in addition to other tags. This is important
 368          *  for section editing, where these comments cause confusion when
 369          *  counting the sections in the wikisource
 370          *
 371          * @access private
 372          */
 373         function strip( $text, &$state, $stripcomments = false ) {
 374                 $render = ($this->mOutputType == OT_HTML);
 375                 $html_content = array();
 376                 $nowiki_content = array();
 377                 $math_content = array();
 378                 $pre_content = array();
 379                 $comment_content = array();
 380                 $ext_content = array();
 381                 $ext_tags = array();
 382                 $ext_params = array();
 383                 $gallery_content = array();
 384
 385                 # Replace any instances of the placeholders
 386                 $uniq_prefix = $this->mUniqPrefix;
 387                 #$text = str_replace( $uniq_prefix, wfHtmlEscapeFirst( $uniq_prefix ), $text );
 388
 389                 # html
 390                 global $wgRawHtml;
 391                 if( $wgRawHtml ) {
 392                         $text = Parser::extractTags('html', $text, $html_content, $uniq_prefix);
 393                         foreach( $html_content as $marker => $content ) {
 394                                 if ($render ) {
 395                                         # Raw and unchecked for validity.
 396                                         $html_content[$marker] = $content;
 397                                 } else {
 398                                         $html_content[$marker] = '<html>'.$content.'</html>';
 399                                 }
 400                         }
 401                 }
 402
 403                 # nowiki
 404                 $text = Parser::extractTags('nowiki', $text, $nowiki_content, $uniq_prefix);
 405                 foreach( $nowiki_content as $marker => $content ) {
 406                         if( $render ){
 407                                 $nowiki_content[$marker] = wfEscapeHTMLTagsOnly( $content );
 408                         } else {
 409                                 $nowiki_content[$marker] = '<nowiki>'.$content.'</nowiki>';
 410                         }
 411                 }
 412
 413                 # math
 414                 if( $this->mOptions->getUseTeX() ) {
 415                         $text = Parser::extractTags('math', $text, $math_content, $uniq_prefix);
 416                         foreach( $math_content as $marker => $content ){
 417                                 if( $render ) {
 418                                         $math_content[$marker] = renderMath( $content );
 419                                 } else {
 420                                         $math_content[$marker] = '<math>'.$content.'</math>';
 421                                 }
 422                         }
 423                 }
 424
 425                 # pre
 426                 $text = Parser::extractTags('pre', $text, $pre_content, $uniq_prefix);
 427                 foreach( $pre_content as $marker => $content ){
 428                         if( $render ){
 429                                 $pre_content[$marker] = '<pre>' . wfEscapeHTMLTagsOnly( $content ) . '</pre>';
 430                         } else {
 431                                 $pre_content[$marker] = '<pre>'.$content.'</pre>';
 432                         }
 433                 }
 434
 435                 # gallery
 436                 $text = Parser::extractTags('gallery', $text, $gallery_content, $uniq_prefix);
 437                 foreach( $gallery_content as $marker => $content ) {
 438                         require_once( 'ImageGallery.php' );
 439                         if ( $render ) {
 440                                 $gallery_content[$marker] = $this->renderImageGallery( $content );
 441                         } else {
 442                                 $gallery_content[$marker] = '<gallery>'.$content.'</gallery>';
 443                         }
 444                 }
 445
 446                 # Comments
 447                 if($stripcomments) {
 448                         $text = Parser::extractTags(STRIP_COMMENTS, $text, $comment_content, $uniq_prefix);
 449                         foreach( $comment_content as $marker => $content ){
 450                                 $comment_content[$marker] = '<!--'.$content.'-->';
 451                         }
 452                 }
 453
 454                 # Extensions
 455                 foreach ( $this->mTagHooks as $tag => $callback ) {
 456                         $ext_content[$tag] = array();
 457                         $text = Parser::extractTagsAndParams( $tag, $text, $ext_content[$tag],
 458                                 $ext_tags[$tag], $ext_params[$tag], $uniq_prefix );
 459                         foreach( $ext_content[$tag] as $marker => $content ) {
 460                                 $full_tag = $ext_tags[$tag][$marker];
 461                                 $params = $ext_params[$tag][$marker];
 462                                 if ( $render )
 463                                         $ext_content[$tag][$marker] = call_user_func_array( $callback, array( $content, $params, &$this ) );
 464                                 else {
 465                                         if ( is_null( $content ) ) {
 466                                                 // Empty element tag
 467                                                 $ext_content[$tag][$marker] = $full_tag;
 468                                         } else {
 469                                                 $ext_content[$tag][$marker] = "$full_tag$content</$tag>";
 470                                         }
 471                                 }
 472                         }
 473                 }
 474
 475                 # Merge state with the pre-existing state, if there is one
 476                 if ( $state ) {
 477                         $state['html'] = $state['html'] + $html_content;
 478                         $state['nowiki'] = $state['nowiki'] + $nowiki_content;
 479                         $state['math'] = $state['math'] + $math_content;
 480                         $state['pre'] = $state['pre'] + $pre_content;
 481                         $state['comment'] = $state['comment'] + $comment_content;
 482                         $state['gallery'] = $state['gallery'] + $gallery_content;
 483
 484                         foreach( $ext_content as $tag => $array ) {
 485                                 if ( array_key_exists( $tag, $state ) ) {
 486                                         $state[$tag] = $state[$tag] + $array;
 487                                 }
 488                         }
 489                 } else {
 490                         $state = array(
 491                           'html' => $html_content,
 492                           'nowiki' => $nowiki_content,
 493                           'math' => $math_content,
 494                           'pre' => $pre_content,
 495                           'comment' => $comment_content,
 496                           'gallery' => $gallery_content,
 497                         ) + $ext_content;
 498                 }
 499                 return $text;
 500         }
 501
 502         /**
 503          * restores pre, math, and hiero removed by strip()
 504          *
 505          * always call unstripNoWiki() after this one
 506          * @access private
 507          */
 508         function unstrip( $text, &$state ) {
 509                 if ( !is_array( $state ) ) {
 510                         return $text;
 511                 }
 512
 513                 # Must expand in reverse order, otherwise nested tags will be corrupted
 514                 foreach( array_reverse( $state, true ) as $tag => $contentDict ) {
 515                         if( $tag != 'nowiki' && $tag != 'html' ) {
 516                                 foreach( array_reverse( $contentDict, true ) as $uniq => $content ) {
 517                                         $text = str_replace( $uniq, $content, $text );
 518                                 }
 519                         }
 520                 }
 521
 522                 return $text;
 523         }
 524
 525         /**
 526          * always call this after unstrip() to preserve the order
 527          *
 528          * @access private
 529          */
 530         function unstripNoWiki( $text, &$state ) {
 531                 if ( !is_array( $state ) ) {
 532                         return $text;
 533                 }
 534
 535                 # Must expand in reverse order, otherwise nested tags will be corrupted
 536                 for ( $content = end($state['nowiki']); $content !== false; $content = prev( $state['nowiki'] ) ) {
 537                         $text = str_replace( key( $state['nowiki'] ), $content, $text );
 538                 }
 539
 540                 global $wgRawHtml;
 541                 if ($wgRawHtml) {
 542                         for ( $content = end($state['html']); $content !== false; $content = prev( $state['html'] ) ) {
 543                                 $text = str_replace( key( $state['html'] ), $content, $text );
 544                         }
 545                 }
 546
 547                 return $text;
 548         }
 549
 550         /**
 551          * Add an item to the strip state
 552          * Returns the unique tag which must be inserted into the stripped text
 553          * The tag will be replaced with the original text in unstrip()
 554          *
 555          * @access private
 556          */
 557         function insertStripItem( $text, &$state ) {
 558                 $rnd = $this->mUniqPrefix . '-item' . Parser::getRandomString();
 559                 if ( !$state ) {
 560                         $state = array(
 561                           'html' => array(),
 562                           'nowiki' => array(),
 563                           'math' => array(),
 564                           'pre' => array(),
 565                           'comment' => array(),
 566                           'gallery' => array(),
 567                         );
 568                 }
 569                 $state['item'][$rnd] = $text;
 570                 return $rnd;
 571         }
 572
 573         /**
 574          * Interface with html tidy, used if $wgUseTidy = true.
 575          * If tidy isn't able to correct the markup, the original will be
 576          * returned in all its glory with a warning comment appended.
 577          *
 578          * Either the external tidy program or the in-process tidy extension
 579          * will be used depending on availability. Override the default
 580          * $wgTidyInternal setting to disable the internal if it's not working.
 581          *
 582          * @param string $text Hideous HTML input
 583          * @return string Corrected HTML output
 584          * @access public
 585          * @static
 586          */
 587         function tidy( $text ) {
 588                 global $wgTidyInternal;
 589                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 590 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 591 '<head><title>test</title></head><body>'.$text.'</body></html>';
 592                 if( $wgTidyInternal ) {
 593                         $correctedtext = Parser::internalTidy( $wrappedtext );
 594                 } else {
 595                         $correctedtext = Parser::externalTidy( $wrappedtext );
 596                 }
 597                 if( is_null( $correctedtext ) ) {
 598                         wfDebug( "Tidy error detected!\n" );
 599                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 600                 }
 601                 return $correctedtext;
 602         }
 603
 604         /**
 605          * Spawn an external HTML tidy process and get corrected markup back from it.
 606          *
 607          * @access private
 608          * @static
 609          */
 610         function externalTidy( $text ) {
 611                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 612                 $fname = 'Parser::externalTidy';
 613                 wfProfileIn( $fname );
 614
 615                 $cleansource = '';
 616                 $opts = ' -utf8';
 617
 618                 $descriptorspec = array(
 619                         0 => array('pipe', 'r'),
 620                         1 => array('pipe', 'w'),
 621                         2 => array('file', '/dev/null', 'a')
 622                 );
 623                 $pipes = array();
 624                 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 625                 if (is_resource($process)) {
 626                         fwrite($pipes[0], $text);
 627                         fclose($pipes[0]);
 628                         while (!feof($pipes[1])) {
 629                                 $cleansource .= fgets($pipes[1], 1024);
 630                         }
 631                         fclose($pipes[1]);
 632                         proc_close($process);
 633                 }
 634
 635                 wfProfileOut( $fname );
 636
 637                 if( $cleansource == '' && $text != '') {
 638                         // Some kind of error happened, so we couldn't get the corrected text.
 639                         // Just give up; we'll use the source text and append a warning.
 640                         return null;
 641                 } else {
 642                         return $cleansource;
 643                 }
 644         }
 645
 646         /**
 647          * Use the HTML tidy PECL extension to use the tidy library in-process,
 648          * saving the overhead of spawning a new process. Currently written to
 649          * the PHP 4.3.x version of the extension, may not work on PHP 5.
 650          *
 651          * 'pear install tidy' should be able to compile the extension module.
 652          *
 653          * @access private
 654          * @static
 655          */
 656         function internalTidy( $text ) {
 657                 global $wgTidyConf;
 658                 $fname = 'Parser::internalTidy';
 659                 wfProfileIn( $fname );
 660
 661                 tidy_load_config( $wgTidyConf );
 662                 tidy_set_encoding( 'utf8' );
 663                 tidy_parse_string( $text );
 664                 tidy_clean_repair();
 665                 if( tidy_get_status() == 2 ) {
 666                         // 2 is magic number for fatal error
 667                         // http://www.php.net/manual/en/function.tidy-get-status.php
 668                         $cleansource = null;
 669                 } else {
 670                         $cleansource = tidy_get_output();
 671                 }
 672                 wfProfileOut( $fname );
 673                 return $cleansource;
 674         }
 675
 676         /**
 677          * parse the wiki syntax used to render tables
 678          *
 679          * @access private
 680          */
 681         function doTableStuff ( $t ) {
 682                 $fname = 'Parser::doTableStuff';
 683                 wfProfileIn( $fname );
 684
 685                 $t = explode ( "\n" , $t ) ;
 686                 $td = array () ; # Is currently a td tag open?
 687                 $ltd = array () ; # Was it TD or TH?
 688                 $tr = array () ; # Is currently a tr tag open?
 689                 $ltr = array () ; # tr attributes
 690                 $indent_level = 0; # indent level of the table
 691                 foreach ( $t AS $k => $x )
 692                 {
 693                         $x = trim ( $x ) ;
 694                         $fc = substr ( $x , 0 , 1 ) ;
 695                         if ( preg_match( '/^(:*)\{\|(.*)$/', $x, $matches ) ) {
 696                                 $indent_level = strlen( $matches[1] );
 697
 698                                 $attributes = $this->unstripForHTML( $matches[2] );
 699
 700                                 $t[$k] = str_repeat( '<dl><dd>', $indent_level ) .
 701                                         '<table' . Sanitizer::fixTagAttributes ( $attributes, 'table' ) . '>' ;
 702                                 array_push ( $td , false ) ;
 703                                 array_push ( $ltd , '' ) ;
 704                                 array_push ( $tr , false ) ;
 705                                 array_push ( $ltr , '' ) ;
 706                         }
 707                         else if ( count ( $td ) == 0 ) { } # Don't do any of the following
 708                         else if ( '|}' == substr ( $x , 0 , 2 ) ) {
 709                                 $z = "</table>" . substr ( $x , 2);
 710                                 $l = array_pop ( $ltd ) ;
 711                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 712                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 713                                 if ( $l == "" ) $z = "<tr><td/></tr>" . $z ;
 714                                 array_pop ( $ltr ) ;
 715                                 $t[$k] = $z . str_repeat( '</dd></dl>', $indent_level );
 716                         }
 717                         else if ( '|-' == substr ( $x , 0 , 2 ) ) { # Allows for |---------------
 718                                 $x = substr ( $x , 1 ) ;
 719                                 while ( $x != '' && substr ( $x , 0 , 1 ) == '-' ) $x = substr ( $x , 1 ) ;
 720                                 $z = '' ;
 721                                 $l = array_pop ( $ltd ) ;
 722                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 723                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 724                                 array_pop ( $ltr ) ;
 725                                 $t[$k] = $z ;
 726                                 array_push ( $tr , false ) ;
 727                                 array_push ( $td , false ) ;
 728                                 array_push ( $ltd , '' ) ;
 729                                 $attributes = $this->unstripForHTML( $x );
 730                                 array_push ( $ltr , Sanitizer::fixTagAttributes ( $attributes, 'tr' ) ) ;
 731                         }
 732                         else if ( '|' == $fc || '!' == $fc || '|+' == substr ( $x , 0 , 2 ) ) { # Caption
 733                                 # $x is a table row
 734                                 if ( '|+' == substr ( $x , 0 , 2 ) ) {
 735                                         $fc = '+' ;
 736                                         $x = substr ( $x , 1 ) ;
 737                                 }
 738                                 $after = substr ( $x , 1 ) ;
 739                                 if ( $fc == '!' ) $after = str_replace ( '!!' , '||' , $after ) ;
 740                                 $after = explode ( '||' , $after ) ;
 741                                 $t[$k] = '' ;
 742
 743                                 # Loop through each table cell
 744                                 foreach ( $after AS $theline )
 745                                 {
 746                                         $z = '' ;
 747                                         if ( $fc != '+' )
 748                                         {
 749                                                 $tra = array_pop ( $ltr ) ;
 750                                                 if ( !array_pop ( $tr ) ) $z = '<tr'.$tra.">\n" ;
 751                                                 array_push ( $tr , true ) ;
 752                                                 array_push ( $ltr , '' ) ;
 753                                         }
 754
 755                                         $l = array_pop ( $ltd ) ;
 756                                         if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 757                                         if ( $fc == '|' ) $l = 'td' ;
 758                                         else if ( $fc == '!' ) $l = 'th' ;
 759                                         else if ( $fc == '+' ) $l = 'caption' ;
 760                                         else $l = '' ;
 761                                         array_push ( $ltd , $l ) ;
 762
 763                                         # Cell parameters
 764                                         $y = explode ( '|' , $theline , 2 ) ;
 765                                         # Note that a '|' inside an invalid link should not
 766                                         # be mistaken as delimiting cell parameters
 767                                         if ( strpos( $y[0], '[[' ) !== false ) {
 768                                                 $y = array ($theline);
 769                                         }
 770                                         if ( count ( $y ) == 1 )
 771                                                 $y = "{$z}<{$l}>{$y[0]}" ;
 772                                         else {
 773                                                 $attributes = $this->unstripForHTML( $y[0] );
 774                                                 $y = "{$z}<{$l}".Sanitizer::fixTagAttributes($attributes, $l).">{$y[1]}" ;
 775                                         }
 776                                         $t[$k] .= $y ;
 777                                         array_push ( $td , true ) ;
 778                                 }
 779                         }
 780                 }
 781
 782                 # Closing open td, tr && table
 783                 while ( count ( $td ) > 0 )
 784                 {
 785                         $l = array_pop ( $ltd ) ;
 786                         if ( $l == "" ) $t[] = "<tr><td/></tr>" ;
 787                         if ( array_pop ( $td ) ) $t[] = '</td>' ;
 788                         if ( array_pop ( $tr ) ) $t[] = '</tr>' ;
 789                         $t[] = '</table>' ;
 790                 }
 791
 792                 $t = implode ( "\n" , $t ) ;
 793                 wfProfileOut( $fname );
 794                 return $t ;
 795         }
 796
 797         /**
 798          * Helper function for parse() that transforms wiki markup into
 799          * HTML. Only called for $mOutputType == OT_HTML.
 800          *
 801          * @access private
 802          */
 803         function internalParse( $text ) {
 804                 global $wgContLang;
 805                 $args = array();
 806                 $isMain = true;
 807                 $fname = 'Parser::internalParse';
 808                 wfProfileIn( $fname );
 809
 810                 # Remove <noinclude> tags and <includeonly> sections
 811                 $text = strtr( $text, array( '<onlyinclude>' => '' , '</onlyinclude>' => '' ) );
 812                 $text = strtr( $text, array( '<noinclude>' => '', '</noinclude>' => '') );
 813                 $text = preg_replace( '/<includeonly>.*?<\/includeonly>/s', '', $text );
 814
 815                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ) );
 816                 $text = $this->replaceVariables( $text, $args );
 817
 818                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 819
 820                 $text = $this->doHeadings( $text );
 821                 if($this->mOptions->getUseDynamicDates()) {
 822                         $df =& DateFormatter::getInstance();
 823                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 824                 }
 825                 $text = $this->doAllQuotes( $text );
 826                 $text = $this->replaceInternalLinks( $text );
 827                 $text = $this->replaceExternalLinks( $text );
 828
 829                 # replaceInternalLinks may sometimes leave behind
 830                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 831                 $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
 832
 833                 $text = $this->doMagicLinks( $text );
 834                 $text = $this->doTableStuff( $text );
 835                 $text = $this->formatHeadings( $text, $isMain );
 836
 837                 wfProfileOut( $fname );
 838                 return $text;
 839         }
 840
 841         /**
 842          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 843          * magic external links.
 844          *
 845          * @access private
 846          */
 847         function &doMagicLinks( &$text ) {
 848                 $text = $this->magicISBN( $text );
 849                 $text = $this->magicRFC( $text, 'RFC ', 'rfcurl' );
 850                 $text = $this->magicRFC( $text, 'PMID ', 'pubmedurl' );
 851                 return $text;
 852         }
 853
 854         /**
 855          * Parse headers and return html
 856          *
 857          * @access private
 858          */
 859         function doHeadings( $text ) {
 860                 $fname = 'Parser::doHeadings';
 861                 wfProfileIn( $fname );
 862                 for ( $i = 6; $i >= 1; --$i ) {
 863                         $h = str_repeat( '=', $i );
 864                         $text = preg_replace( "/^{$h}(.+){$h}(\\s|$)/m",
 865                           "<h{$i}>\\1</h{$i}>\\2", $text );
 866                 }
 867                 wfProfileOut( $fname );
 868                 return $text;
 869         }
 870
 871         /**
 872          * Replace single quotes with HTML markup
 873          * @access private
 874          * @return string the altered text
 875          */
 876         function doAllQuotes( $text ) {
 877                 $fname = 'Parser::doAllQuotes';
 878                 wfProfileIn( $fname );
 879                 $outtext = '';
 880                 $lines = explode( "\n", $text );
 881                 foreach ( $lines as $line ) {
 882                         $outtext .= $this->doQuotes ( $line ) . "\n";
 883                 }
 884                 $outtext = substr($outtext, 0,-1);
 885                 wfProfileOut( $fname );
 886                 return $outtext;
 887         }
 888
 889         /**
 890          * Helper function for doAllQuotes()
 891          * @access private
 892          */
 893         function doQuotes( $text ) {
 894                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
 895                 if ( count( $arr ) == 1 )
 896                         return $text;
 897                 else
 898                 {
 899                         # First, do some preliminary work. This may shift some apostrophes from
 900                         # being mark-up to being text. It also counts the number of occurrences
 901                         # of bold and italics mark-ups.
 902                         $i = 0;
 903                         $numbold = 0;
 904                         $numitalics = 0;
 905                         foreach ( $arr as $r )
 906                         {
 907                                 if ( ( $i % 2 ) == 1 )
 908                                 {
 909                                         # If there are ever four apostrophes, assume the first is supposed to
 910                                         # be text, and the remaining three constitute mark-up for bold text.
 911                                         if ( strlen( $arr[$i] ) == 4 )
 912                                         {
 913                                                 $arr[$i-1] .= "'";
 914                                                 $arr[$i] = "'''";
 915                                         }
 916                                         # If there are more than 5 apostrophes in a row, assume they're all
 917                                         # text except for the last 5.
 918                                         else if ( strlen( $arr[$i] ) > 5 )
 919                                         {
 920                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
 921                                                 $arr[$i] = "'''''";
 922                                         }
 923                                         # Count the number of occurrences of bold and italics mark-ups.
 924                                         # We are not counting sequences of five apostrophes.
 925                                         if ( strlen( $arr[$i] ) == 2 ) $numitalics++;  else
 926                                         if ( strlen( $arr[$i] ) == 3 ) $numbold++;     else
 927                                         if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
 928                                 }
 929                                 $i++;
 930                         }
 931
 932                         # If there is an odd number of both bold and italics, it is likely
 933                         # that one of the bold ones was meant to be an apostrophe followed
 934                         # by italics. Which one we cannot know for certain, but it is more
 935                         # likely to be one that has a single-letter word before it.
 936                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
 937                         {
 938                                 $i = 0;
 939                                 $firstsingleletterword = -1;
 940                                 $firstmultiletterword = -1;
 941                                 $firstspace = -1;
 942                                 foreach ( $arr as $r )
 943                                 {
 944                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
 945                                         {
 946                                                 $x1 = substr ($arr[$i-1], -1);
 947                                                 $x2 = substr ($arr[$i-1], -2, 1);
 948                                                 if ($x1 == ' ') {
 949                                                         if ($firstspace == -1) $firstspace = $i;
 950                                                 } else if ($x2 == ' ') {
 951                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
 952                                                 } else {
 953                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
 954                                                 }
 955                                         }
 956                                         $i++;
 957                                 }
 958
 959                                 # If there is a single-letter word, use it!
 960                                 if ($firstsingleletterword > -1)
 961                                 {
 962                                         $arr [ $firstsingleletterword ] = "''";
 963                                         $arr [ $firstsingleletterword-1 ] .= "'";
 964                                 }
 965                                 # If not, but there's a multi-letter word, use that one.
 966                                 else if ($firstmultiletterword > -1)
 967                                 {
 968                                         $arr [ $firstmultiletterword ] = "''";
 969                                         $arr [ $firstmultiletterword-1 ] .= "'";
 970                                 }
 971                                 # ... otherwise use the first one that has neither.
 972                                 # (notice that it is possible for all three to be -1 if, for example,
 973                                 # there is only one pentuple-apostrophe in the line)
 974                                 else if ($firstspace > -1)
 975                                 {
 976                                         $arr [ $firstspace ] = "''";
 977                                         $arr [ $firstspace-1 ] .= "'";
 978                                 }
 979                         }
 980
 981                         # Now let's actually convert our apostrophic mush to HTML!
 982                         $output = '';
 983                         $buffer = '';
 984                         $state = '';
 985                         $i = 0;
 986                         foreach ($arr as $r)
 987                         {
 988                                 if (($i % 2) == 0)
 989                                 {
 990                                         if ($state == 'both')
 991                                                 $buffer .= $r;
 992                                         else
 993                                                 $output .= $r;
 994                                 }
 995                                 else
 996                                 {
 997                                         if (strlen ($r) == 2)
 998                                         {
 999                                                 if ($state == 'i')
1000                                                 { $output .= '</i>'; $state = ''; }
1001                                                 else if ($state == 'bi')
1002                                                 { $output .= '</i>'; $state = 'b'; }
1003                                                 else if ($state == 'ib')
1004                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1005                                                 else if ($state == 'both')
1006                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1007                                                 else # $state can be 'b' or ''
1008                                                 { $output .= '<i>'; $state .= 'i'; }
1009                                         }
1010                                         else if (strlen ($r) == 3)
1011                                         {
1012                                                 if ($state == 'b')
1013                                                 { $output .= '</b>'; $state = ''; }
1014                                                 else if ($state == 'bi')
1015                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1016                                                 else if ($state == 'ib')
1017                                                 { $output .= '</b>'; $state = 'i'; }
1018                                                 else if ($state == 'both')
1019                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1020                                                 else # $state can be 'i' or ''
1021                                                 { $output .= '<b>'; $state .= 'b'; }
1022                                         }
1023                                         else if (strlen ($r) == 5)
1024                                         {
1025                                                 if ($state == 'b')
1026                                                 { $output .= '</b><i>'; $state = 'i'; }
1027                                                 else if ($state == 'i')
1028                                                 { $output .= '</i><b>'; $state = 'b'; }
1029                                                 else if ($state == 'bi')
1030                                                 { $output .= '</i></b>'; $state = ''; }
1031                                                 else if ($state == 'ib')
1032                                                 { $output .= '</b></i>'; $state = ''; }
1033                                                 else if ($state == 'both')
1034                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1035                                                 else # ($state == '')
1036                                                 { $buffer = ''; $state = 'both'; }
1037                                         }
1038                                 }
1039                                 $i++;
1040                         }
1041                         # Now close all remaining tags.  Notice that the order is important.
1042                         if ($state == 'b' || $state == 'ib')
1043                                 $output .= '</b>';
1044                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
1045                                 $output .= '</i>';
1046                         if ($state == 'bi')
1047                                 $output .= '</b>';
1048                         if ($state == 'both')
1049                                 $output .= '<b><i>'.$buffer.'</i></b>';
1050                         return $output;
1051                 }
1052         }
1053
1054         /**
1055          * Replace external links
1056          *
1057          * Note: this is all very hackish and the order of execution matters a lot.
1058          * Make sure to run maintenance/parserTests.php if you change this code.
1059          *
1060          * @access private
1061          */
1062         function replaceExternalLinks( $text ) {
1063                 global $wgContLang;
1064                 $fname = 'Parser::replaceExternalLinks';
1065                 wfProfileIn( $fname );
1066
1067                 $sk =& $this->mOptions->getSkin();
1068
1069                 $bits = preg_split( EXT_LINK_BRACKETED, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1070
1071                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
1072
1073                 $i = 0;
1074                 while ( $i<count( $bits ) ) {
1075                         $url = $bits[$i++];
1076                         $protocol = $bits[$i++];
1077                         $text = $bits[$i++];
1078                         $trail = $bits[$i++];
1079
1080                         # The characters '<' and '>' (which were escaped by
1081                         # removeHTMLtags()) should not be included in
1082                         # URLs, per RFC 2396.
1083                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1084                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1085                                 $url = substr($url, 0, $m2[0][1]);
1086                         }
1087
1088                         # If the link text is an image URL, replace it with an <img> tag
1089                         # This happened by accident in the original parser, but some people used it extensively
1090                         $img = $this->maybeMakeExternalImage( $text );
1091                         if ( $img !== false ) {
1092                                 $text = $img;
1093                         }
1094
1095                         $dtrail = '';
1096
1097                         # Set linktype for CSS - if URL==text, link is essentially free
1098                         $linktype = ($text == $url) ? 'free' : 'text';
1099
1100                         # No link text, e.g. [http://domain.tld/some.link]
1101                         if ( $text == '' ) {
1102                                 # Autonumber if allowed
1103                                 if ( strpos( HTTP_PROTOCOLS, str_replace('/','\/', $protocol) ) !== false ) {
1104                                         $text = '[' . ++$this->mAutonumber . ']';
1105                                         $linktype = 'autonumber';
1106                                 } else {
1107                                         # Otherwise just use the URL
1108                                         $text = htmlspecialchars( $url );
1109                                         $linktype = 'free';
1110                                 }
1111                         } else {
1112                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1113                                 # Check for trail
1114                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1115                         }
1116
1117                         $text = $wgContLang->markNoConversion($text);
1118
1119                         # Replace &amp; from obsolete syntax with &.
1120                         # All HTML entities will be escaped by makeExternalLink()
1121                         $url = str_replace( '&amp;', '&', $url );
1122                         # Replace unnecessary URL escape codes with the referenced character
1123                         # This prevents spammers from hiding links from the filters
1124                         $url = Parser::replaceUnusualEscapes( $url );
1125
1126                         # Process the trail (i.e. everything after this link up until start of the next link),
1127                         # replacing any non-bracketed links
1128                         $trail = $this->replaceFreeExternalLinks( $trail );
1129
1130                         # Use the encoded URL
1131                         # This means that users can paste URLs directly into the text
1132                         # Funny characters like &ouml; aren't valid in URLs anyway
1133                         # This was changed in August 2004
1134                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype ) . $dtrail . $trail;
1135
1136                         # Register link in the output object
1137                         $this->mOutput->addExternalLink( $url );
1138                 }
1139
1140                 wfProfileOut( $fname );
1141                 return $s;
1142         }
1143
1144         /**
1145          * Replace anything that looks like a URL with a link
1146          * @access private
1147          */
1148         function replaceFreeExternalLinks( $text ) {
1149                 global $wgContLang;
1150                 $fname = 'Parser::replaceFreeExternalLinks';
1151                 wfProfileIn( $fname );
1152
1153                 $bits = preg_split( '/(\b(?:' . wfUrlProtocols() . '))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1154                 $s = array_shift( $bits );
1155                 $i = 0;
1156
1157                 $sk =& $this->mOptions->getSkin();
1158
1159                 while ( $i < count( $bits ) ){
1160                         $protocol = $bits[$i++];
1161                         $remainder = $bits[$i++];
1162
1163                         if ( preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1164                                 # Found some characters after the protocol that look promising
1165                                 $url = $protocol . $m[1];
1166                                 $trail = $m[2];
1167
1168                                 # The characters '<' and '>' (which were escaped by
1169                                 # removeHTMLtags()) should not be included in
1170                                 # URLs, per RFC 2396.
1171                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1172                                         $trail = substr($url, $m2[0][1]) . $trail;
1173                                         $url = substr($url, 0, $m2[0][1]);
1174                                 }
1175
1176                                 # Move trailing punctuation to $trail
1177                                 $sep = ',;\.:!?';
1178                                 # If there is no left bracket, then consider right brackets fair game too
1179                                 if ( strpos( $url, '(' ) === false ) {
1180                                         $sep .= ')';
1181                                 }
1182
1183                                 $numSepChars = strspn( strrev( $url ), $sep );
1184                                 if ( $numSepChars ) {
1185                                         $trail = substr( $url, -$numSepChars ) . $trail;
1186                                         $url = substr( $url, 0, -$numSepChars );
1187                                 }
1188
1189                                 # Replace &amp; from obsolete syntax with &.
1190                                 # All HTML entities will be escaped by makeExternalLink()
1191                                 # or maybeMakeExternalImage()
1192                                 $url = str_replace( '&amp;', '&', $url );
1193                                 # Replace unnecessary URL escape codes with their equivalent characters
1194                                 $url = Parser::replaceUnusualEscapes( $url );
1195
1196                                 # Is this an external image?
1197                                 $text = $this->maybeMakeExternalImage( $url );
1198                                 if ( $text === false ) {
1199                                         # Not an image, make a link
1200                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free' );
1201                                         # Register it in the output object
1202                                         $this->mOutput->addExternalLink( $url );
1203                                 }
1204                                 $s .= $text . $trail;
1205                         } else {
1206                                 $s .= $protocol . $remainder;
1207                         }
1208                 }
1209                 wfProfileOut( $fname );
1210                 return $s;
1211         }
1212
1213         /**
1214          * Replace unusual URL escape codes with their equivalent characters
1215          * @param string
1216          * @return string
1217          * @static
1218          */
1219         function replaceUnusualEscapes( $url ) {
1220                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1221                         array( 'Parser', 'replaceUnusualEscapesCallback' ), $url );
1222         }
1223
1224         /**
1225          * Callback function used in replaceUnusualEscapes().
1226          * Replaces unusual URL escape codes with their equivalent character
1227          * @static
1228          * @access private
1229          */
1230         function replaceUnusualEscapesCallback( $matches ) {
1231                 $char = urldecode( $matches[0] );
1232                 $ord = ord( $char );
1233                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1234                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1235                         // No, shouldn't be escaped
1236                         return $char;
1237                 } else {
1238                         // Yes, leave it escaped
1239                         return $matches[0];
1240                 }
1241         }
1242
1243         /**
1244          * make an image if it's allowed, either through the global
1245          * option or through the exception
1246          * @access private
1247          */
1248         function maybeMakeExternalImage( $url ) {
1249                 $sk =& $this->mOptions->getSkin();
1250                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1251                 $imagesexception = !empty($imagesfrom);
1252                 $text = false;
1253                 if ( $this->mOptions->getAllowExternalImages()
1254                      || ( $imagesexception && strpos( $url, $imagesfrom ) === 0 ) ) {
1255                         if ( preg_match( EXT_IMAGE_REGEX, $url ) ) {
1256                                 # Image found
1257                                 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
1258                         }
1259                 }
1260                 return $text;
1261         }
1262
1263         /**
1264          * Process [[ ]] wikilinks
1265          *
1266          * @access private
1267          */
1268         function replaceInternalLinks( $s ) {
1269                 global $wgContLang;
1270                 static $fname = 'Parser::replaceInternalLinks' ;
1271
1272                 wfProfileIn( $fname );
1273
1274                 wfProfileIn( $fname.'-setup' );
1275                 static $tc = FALSE;
1276                 # the % is needed to support urlencoded titles as well
1277                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1278
1279                 $sk =& $this->mOptions->getSkin();
1280
1281                 #split the entire text string on occurences of [[
1282                 $a = explode( '[[', ' ' . $s );
1283                 #get the first element (all text up to first [[), and remove the space we added
1284                 $s = array_shift( $a );
1285                 $s = substr( $s, 1 );
1286
1287                 # Match a link having the form [[namespace:link|alternate]]trail
1288                 static $e1 = FALSE;
1289                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1290                 # Match cases where there is no "]]", which might still be images
1291                 static $e1_img = FALSE;
1292                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1293                 # Match the end of a line for a word that's not followed by whitespace,
1294                 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1295                 $e2 = wfMsgForContent( 'linkprefix' );
1296
1297                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1298
1299                 if( is_null( $this->mTitle ) ) {
1300                         wfDebugDieBacktrace( 'nooo' );
1301                 }
1302                 $nottalk = !$this->mTitle->isTalkPage();
1303
1304                 if ( $useLinkPrefixExtension ) {
1305                         if ( preg_match( $e2, $s, $m ) ) {
1306                                 $first_prefix = $m[2];
1307                         } else {
1308                                 $first_prefix = false;
1309                         }
1310                 } else {
1311                         $prefix = '';
1312                 }
1313
1314                 $selflink = $this->mTitle->getPrefixedText();
1315                 wfProfileOut( $fname.'-setup' );
1316
1317                 $checkVariantLink = sizeof($wgContLang->getVariants())>1;
1318                 $useSubpages = $this->areSubpagesAllowed();
1319
1320                 # Loop for each link
1321                 for ($k = 0; isset( $a[$k] ); $k++) {
1322                         $line = $a[$k];
1323                         if ( $useLinkPrefixExtension ) {
1324                                 wfProfileIn( $fname.'-prefixhandling' );
1325                                 if ( preg_match( $e2, $s, $m ) ) {
1326                                         $prefix = $m[2];
1327                                         $s = $m[1];
1328                                 } else {
1329                                         $prefix='';
1330                                 }
1331                                 # first link
1332                                 if($first_prefix) {
1333                                         $prefix = $first_prefix;
1334                                         $first_prefix = false;
1335                                 }
1336                                 wfProfileOut( $fname.'-prefixhandling' );
1337                         }
1338
1339                         $might_be_img = false;
1340
1341                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1342                                 $text = $m[2];
1343                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1344                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1345                                 # the real problem is with the $e1 regex
1346                                 # See bug 1300.
1347                                 #
1348                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1349                                 # and no image is in sight. See bug 2095.
1350                                 #
1351                                 if( $text !== '' && preg_match( "/^\](.*)/s", $m[3], $n ) ) {
1352                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1353                                         $m[3] = $n[1];
1354                                 }
1355                                 # fix up urlencoded title texts
1356                                 if(preg_match('/%/', $m[1] )) $m[1] = urldecode($m[1]);
1357                                 $trail = $m[3];
1358                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1359                                 $might_be_img = true;
1360                                 $text = $m[2];
1361                                 if(preg_match('/%/', $m[1] )) $m[1] = urldecode($m[1]);
1362                                 $trail = "";
1363                         } else { # Invalid form; output directly
1364                                 $s .= $prefix . '[[' . $line ;
1365                                 continue;
1366                         }
1367
1368                         # Don't allow internal links to pages containing
1369                         # PROTO: where PROTO is a valid URL protocol; these
1370                         # should be external links.
1371                         if (preg_match('/^(\b(?:' . wfUrlProtocols() . '))/', $m[1])) {
1372                                 $s .= $prefix . '[[' . $line ;
1373                                 continue;
1374                         }
1375
1376                         # Make subpage if necessary
1377                         if( $useSubpages ) {
1378                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1379                         } else {
1380                                 $link = $m[1];
1381                         }
1382
1383                         $noforce = (substr($m[1], 0, 1) != ':');
1384                         if (!$noforce) {
1385                                 # Strip off leading ':'
1386                                 $link = substr($link, 1);
1387                         }
1388
1389                         $nt = Title::newFromText( $this->unstripNoWiki($link, $this->mStripState) );
1390                         if( !$nt ) {
1391                                 $s .= $prefix . '[[' . $line;
1392                                 continue;
1393                         }
1394
1395                         #check other language variants of the link
1396                         #if the article does not exist
1397                         if( $checkVariantLink
1398                             && $nt->getArticleID() == 0 ) {
1399                                 $wgContLang->findVariantLink($link, $nt);
1400                         }
1401
1402                         $ns = $nt->getNamespace();
1403                         $iw = $nt->getInterWiki();
1404
1405                         if ($might_be_img) { # if this is actually an invalid link
1406                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1407                                         $found = false;
1408                                         while (isset ($a[$k+1]) ) {
1409                                                 #look at the next 'line' to see if we can close it there
1410                                                 $spliced = array_splice( $a, $k + 1, 1 );
1411                                                 $next_line = array_shift( $spliced );
1412                                                 if( preg_match("/^(.*?]].*?)]](.*)$/sD", $next_line, $m) ) {
1413                                                 # the first ]] closes the inner link, the second the image
1414                                                         $found = true;
1415                                                         $text .= '[[' . $m[1];
1416                                                         $trail = $m[2];
1417                                                         break;
1418                                                 } elseif( preg_match("/^.*?]].*$/sD", $next_line, $m) ) {
1419                                                         #if there's exactly one ]] that's fine, we'll keep looking
1420                                                         $text .= '[[' . $m[0];
1421                                                 } else {
1422                                                         #if $next_line is invalid too, we need look no further
1423                                                         $text .= '[[' . $next_line;
1424                                                         break;
1425                                                 }
1426                                         }
1427                                         if ( !$found ) {
1428                                                 # we couldn't find the end of this imageLink, so output it raw
1429                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1430                                                 $text = $this->replaceInternalLinks($text);
1431                                                 $s .= $prefix . '[[' . $link . '|' . $text;
1432                                                 # note: no $trail, because without an end, there *is* no trail
1433                                                 continue;
1434                                         }
1435                                 } else { #it's not an image, so output it raw
1436                                         $s .= $prefix . '[[' . $link . '|' . $text;
1437                                         # note: no $trail, because without an end, there *is* no trail
1438                                         continue;
1439                                 }
1440                         }
1441
1442                         $wasblank = ( '' == $text );
1443                         if( $wasblank ) $text = $link;
1444
1445
1446                         # Link not escaped by : , create the various objects
1447                         if( $noforce ) {
1448
1449                                 # Interwikis
1450                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1451                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1452                                         $s = rtrim($s . "\n");
1453                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1454                                         continue;
1455                                 }
1456
1457                                 if ( $ns == NS_IMAGE ) {
1458                                         wfProfileIn( "$fname-image" );
1459                                         if ( !wfIsBadImage( $nt->getDBkey() ) ) {
1460                                                 # recursively parse links inside the image caption
1461                                                 # actually, this will parse them in any other parameters, too,
1462                                                 # but it might be hard to fix that, and it doesn't matter ATM
1463                                                 $text = $this->replaceExternalLinks($text);
1464                                                 $text = $this->replaceInternalLinks($text);
1465
1466                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1467                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text ) ) . $trail;
1468                                                 $this->mOutput->addImage( $nt->getDBkey() );
1469
1470                                                 wfProfileOut( "$fname-image" );
1471                                                 continue;
1472                                         }
1473                                         wfProfileOut( "$fname-image" );
1474
1475                                 }
1476
1477                                 if ( $ns == NS_CATEGORY ) {
1478                                         wfProfileIn( "$fname-category" );
1479                                         $s = rtrim($s . "\n"); # bug 87
1480
1481                                         if ( $wasblank ) {
1482                                                 if ( $this->mTitle->getNamespace() == NS_CATEGORY ) {
1483                                                         $sortkey = $this->mTitle->getText();
1484                                                 } else {
1485                                                         $sortkey = $this->mTitle->getPrefixedText();
1486                                                 }
1487                                         } else {
1488                                                 $sortkey = $text;
1489                                         }
1490                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1491                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1492                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1493
1494                                         /**
1495                                          * Strip the whitespace Category links produce, see bug 87
1496                                          * @todo We might want to use trim($tmp, "\n") here.
1497                                          */
1498                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1499
1500                                         wfProfileOut( "$fname-category" );
1501                                         continue;
1502                                 }
1503                         }
1504
1505                         if( ( $nt->getPrefixedText() === $selflink ) &&
1506                             ( $nt->getFragment() === '' ) ) {
1507                                 # Self-links are handled specially; generally de-link and change to bold.
1508                                 $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1509                                 continue;
1510                         }
1511
1512                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1513                         if( $ns == NS_MEDIA ) {
1514                                 $link = $sk->makeMediaLinkObj( $nt, $text );
1515                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1516                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1517                                 $this->mOutput->addImage( $nt->getDBkey() );
1518                                 continue;
1519                         } elseif( $ns == NS_SPECIAL ) {
1520                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1521                                 continue;
1522                         } elseif( $ns == NS_IMAGE ) {
1523                                 $img = Image::newFromTitle( $nt );
1524                                 if( $img->exists() ) {
1525                                         // Force a blue link if the file exists; may be a remote
1526                                         // upload on the shared repository, and we want to see its
1527                                         // auto-generated page.
1528                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1529                                         continue;
1530                                 }
1531                         }
1532                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1533                 }
1534                 wfProfileOut( $fname );
1535                 return $s;
1536         }
1537
1538         /**
1539          * Make a link placeholder. The text returned can be later resolved to a real link with
1540          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1541          * parsing of interwiki links, and secondly to allow all extistence checks and
1542          * article length checks (for stub links) to be bundled into a single query.
1543          *
1544          */
1545         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1546                 if ( ! is_object($nt) ) {
1547                         # Fail gracefully
1548                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1549                 } else {
1550                         # Separate the link trail from the rest of the link
1551                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1552
1553                         if ( $nt->isExternal() ) {
1554                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1555                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1556                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1557                         } else {
1558                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1559                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1560                                 $this->mLinkHolders['queries'][] = $query;
1561                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1562                                 $this->mLinkHolders['titles'][] = $nt;
1563
1564                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1565                         }
1566                 }
1567                 return $retVal;
1568         }
1569
1570         /**
1571          * Render a forced-blue link inline; protect against double expansion of
1572          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1573          * Since this little disaster has to split off the trail text to avoid
1574          * breaking URLs in the following text without breaking trails on the
1575          * wiki links, it's been made into a horrible function.
1576          *
1577          * @param Title $nt
1578          * @param string $text
1579          * @param string $query
1580          * @param string $trail
1581          * @param string $prefix
1582          * @return string HTML-wikitext mix oh yuck
1583          */
1584         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1585                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1586                 $sk =& $this->mOptions->getSkin();
1587                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1588                 return $this->armorLinks( $link ) . $trail;
1589         }
1590
1591         /**
1592          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1593          * going to go through further parsing steps before inline URL expansion.
1594          *
1595          * In particular this is important when using action=render, which causes
1596          * full URLs to be included.
1597          *
1598          * Oh man I hate our multi-layer parser!
1599          *
1600          * @param string more-or-less HTML
1601          * @return string less-or-more HTML with NOPARSE bits
1602          */
1603         function armorLinks( $text ) {
1604                 return preg_replace( "/\b(" . wfUrlProtocols() . ')/',
1605                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1606         }
1607
1608         /**
1609          * Return true if subpage links should be expanded on this page.
1610          * @return bool
1611          */
1612         function areSubpagesAllowed() {
1613                 # Some namespaces don't allow subpages
1614                 global $wgNamespacesWithSubpages;
1615                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1616         }
1617
1618         /**
1619          * Handle link to subpage if necessary
1620          * @param string $target the source of the link
1621          * @param string &$text the link text, modified as necessary
1622          * @return string the full name of the link
1623          * @access private
1624          */
1625         function maybeDoSubpageLink($target, &$text) {
1626                 # Valid link forms:
1627                 # Foobar -- normal
1628                 # :Foobar -- override special treatment of prefix (images, language links)
1629                 # /Foobar -- convert to CurrentPage/Foobar
1630                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1631                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1632                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1633
1634                 $fname = 'Parser::maybeDoSubpageLink';
1635                 wfProfileIn( $fname );
1636                 $ret = $target; # default return value is no change
1637
1638                 # Some namespaces don't allow subpages,
1639                 # so only perform processing if subpages are allowed
1640                 if( $this->areSubpagesAllowed() ) {
1641                         # Look at the first character
1642                         if( $target != '' && $target{0} == '/' ) {
1643                                 # / at end means we don't want the slash to be shown
1644                                 if( substr( $target, -1, 1 ) == '/' ) {
1645                                         $target = substr( $target, 1, -1 );
1646                                         $noslash = $target;
1647                                 } else {
1648                                         $noslash = substr( $target, 1 );
1649                                 }
1650
1651                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash);
1652                                 if( '' === $text ) {
1653                                         $text = $target;
1654                                 } # this might be changed for ugliness reasons
1655                         } else {
1656                                 # check for .. subpage backlinks
1657                                 $dotdotcount = 0;
1658                                 $nodotdot = $target;
1659                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1660                                         ++$dotdotcount;
1661                                         $nodotdot = substr( $nodotdot, 3 );
1662                                 }
1663                                 if($dotdotcount > 0) {
1664                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1665                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1666                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1667                                                 # / at the end means don't show full path
1668                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1669                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1670                                                         if( '' === $text ) {
1671                                                                 $text = $nodotdot;
1672                                                         }
1673                                                 }
1674                                                 $nodotdot = trim( $nodotdot );
1675                                                 if( $nodotdot != '' ) {
1676                                                         $ret .= '/' . $nodotdot;
1677                                                 }
1678                                         }
1679                                 }
1680                         }
1681                 }
1682
1683                 wfProfileOut( $fname );
1684                 return $ret;
1685         }
1686
1687         /**#@+
1688          * Used by doBlockLevels()
1689          * @access private
1690          */
1691         /* private */ function closeParagraph() {
1692                 $result = '';
1693                 if ( '' != $this->mLastSection ) {
1694                         $result = '</' . $this->mLastSection  . ">\n";
1695                 }
1696                 $this->mInPre = false;
1697                 $this->mLastSection = '';
1698                 return $result;
1699         }
1700         # getCommon() returns the length of the longest common substring
1701         # of both arguments, starting at the beginning of both.
1702         #
1703         /* private */ function getCommon( $st1, $st2 ) {
1704                 $fl = strlen( $st1 );
1705                 $shorter = strlen( $st2 );
1706                 if ( $fl < $shorter ) { $shorter = $fl; }
1707
1708                 for ( $i = 0; $i < $shorter; ++$i ) {
1709                         if ( $st1{$i} != $st2{$i} ) { break; }
1710                 }
1711                 return $i;
1712         }
1713         # These next three functions open, continue, and close the list
1714         # element appropriate to the prefix character passed into them.
1715         #
1716         /* private */ function openList( $char ) {
1717                 $result = $this->closeParagraph();
1718
1719                 if ( '*' == $char ) { $result .= '<ul><li>'; }
1720                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
1721                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
1722                 else if ( ';' == $char ) {
1723                         $result .= '<dl><dt>';
1724                         $this->mDTopen = true;
1725                 }
1726                 else { $result = '<!-- ERR 1 -->'; }
1727
1728                 return $result;
1729         }
1730
1731         /* private */ function nextItem( $char ) {
1732                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
1733                 else if ( ':' == $char || ';' == $char ) {
1734                         $close = '</dd>';
1735                         if ( $this->mDTopen ) { $close = '</dt>'; }
1736                         if ( ';' == $char ) {
1737                                 $this->mDTopen = true;
1738                                 return $close . '<dt>';
1739                         } else {
1740                                 $this->mDTopen = false;
1741                                 return $close . '<dd>';
1742                         }
1743                 }
1744                 return '<!-- ERR 2 -->';
1745         }
1746
1747         /* private */ function closeList( $char ) {
1748                 if ( '*' == $char ) { $text = '</li></ul>'; }
1749                 else if ( '#' == $char ) { $text = '</li></ol>'; }
1750                 else if ( ':' == $char ) {
1751                         if ( $this->mDTopen ) {
1752                                 $this->mDTopen = false;
1753                                 $text = '</dt></dl>';
1754                         } else {
1755                                 $text = '</dd></dl>';
1756                         }
1757                 }
1758                 else {  return '<!-- ERR 3 -->'; }
1759                 return $text."\n";
1760         }
1761         /**#@-*/
1762
1763         /**
1764          * Make lists from lines starting with ':', '*', '#', etc.
1765          *
1766          * @access private
1767          * @return string the lists rendered as HTML
1768          */
1769         function doBlockLevels( $text, $linestart ) {
1770                 $fname = 'Parser::doBlockLevels';
1771                 wfProfileIn( $fname );
1772
1773                 # Parsing through the text line by line.  The main thing
1774                 # happening here is handling of block-level elements p, pre,
1775                 # and making lists from lines starting with * # : etc.
1776                 #
1777                 $textLines = explode( "\n", $text );
1778
1779                 $lastPrefix = $output = '';
1780                 $this->mDTopen = $inBlockElem = false;
1781                 $prefixLength = 0;
1782                 $paragraphStack = false;
1783
1784                 if ( !$linestart ) {
1785                         $output .= array_shift( $textLines );
1786                 }
1787                 foreach ( $textLines as $oLine ) {
1788                         $lastPrefixLength = strlen( $lastPrefix );
1789                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
1790                         $preOpenMatch = preg_match('/<pre/i', $oLine );
1791                         if ( !$this->mInPre ) {
1792                                 # Multiple prefixes may abut each other for nested lists.
1793                                 $prefixLength = strspn( $oLine, '*#:;' );
1794                                 $pref = substr( $oLine, 0, $prefixLength );
1795
1796                                 # eh?
1797                                 $pref2 = str_replace( ';', ':', $pref );
1798                                 $t = substr( $oLine, $prefixLength );
1799                                 $this->mInPre = !empty($preOpenMatch);
1800                         } else {
1801                                 # Don't interpret any other prefixes in preformatted text
1802                                 $prefixLength = 0;
1803                                 $pref = $pref2 = '';
1804                                 $t = $oLine;
1805                         }
1806
1807                         # List generation
1808                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
1809                                 # Same as the last item, so no need to deal with nesting or opening stuff
1810                                 $output .= $this->nextItem( substr( $pref, -1 ) );
1811                                 $paragraphStack = false;
1812
1813                                 if ( substr( $pref, -1 ) == ';') {
1814                                         # The one nasty exception: definition lists work like this:
1815                                         # ; title : definition text
1816                                         # So we check for : in the remainder text to split up the
1817                                         # title and definition, without b0rking links.
1818                                         $term = $t2 = '';
1819                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
1820                                                 $t = $t2;
1821                                                 $output .= $term . $this->nextItem( ':' );
1822                                         }
1823                                 }
1824                         } elseif( $prefixLength || $lastPrefixLength ) {
1825                                 # Either open or close a level...
1826                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
1827                                 $paragraphStack = false;
1828
1829                                 while( $commonPrefixLength < $lastPrefixLength ) {
1830                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
1831                                         --$lastPrefixLength;
1832                                 }
1833                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
1834                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
1835                                 }
1836                                 while ( $prefixLength > $commonPrefixLength ) {
1837                                         $char = substr( $pref, $commonPrefixLength, 1 );
1838                                         $output .= $this->openList( $char );
1839
1840                                         if ( ';' == $char ) {
1841                                                 # FIXME: This is dupe of code above
1842                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
1843                                                         $t = $t2;
1844                                                         $output .= $term . $this->nextItem( ':' );
1845                                                 }
1846                                         }
1847                                         ++$commonPrefixLength;
1848                                 }
1849                                 $lastPrefix = $pref2;
1850                         }
1851                         if( 0 == $prefixLength ) {
1852                                 wfProfileIn( "$fname-paragraph" );
1853                                 # No prefix (not in list)--go to paragraph mode
1854                                 // XXX: use a stack for nestable elements like span, table and div
1855                                 $openmatch = preg_match('/(<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
1856                                 $closematch = preg_match(
1857                                         '/(<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
1858                                         '<td|<th|<div|<\\/div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul)/iS', $t );
1859                                 if ( $openmatch or $closematch ) {
1860                                         $paragraphStack = false;
1861                                         $output .= $this->closeParagraph();
1862                                         if ( $preOpenMatch and !$preCloseMatch ) {
1863                                                 $this->mInPre = true;
1864                                         }
1865                                         if ( $closematch ) {
1866                                                 $inBlockElem = false;
1867                                         } else {
1868                                                 $inBlockElem = true;
1869                                         }
1870                                 } else if ( !$inBlockElem && !$this->mInPre ) {
1871                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
1872                                                 // pre
1873                                                 if ($this->mLastSection != 'pre') {
1874                                                         $paragraphStack = false;
1875                                                         $output .= $this->closeParagraph().'<pre>';
1876                                                         $this->mLastSection = 'pre';
1877                                                 }
1878                                                 $t = substr( $t, 1 );
1879                                         } else {
1880                                                 // paragraph
1881                                                 if ( '' == trim($t) ) {
1882                                                         if ( $paragraphStack ) {
1883                                                                 $output .= $paragraphStack.'<br />';
1884                                                                 $paragraphStack = false;
1885                                                                 $this->mLastSection = 'p';
1886                                                         } else {
1887                                                                 if ($this->mLastSection != 'p' ) {
1888                                                                         $output .= $this->closeParagraph();
1889                                                                         $this->mLastSection = '';
1890                                                                         $paragraphStack = '<p>';
1891                                                                 } else {
1892                                                                         $paragraphStack = '</p><p>';
1893                                                                 }
1894                                                         }
1895                                                 } else {
1896                                                         if ( $paragraphStack ) {
1897                                                                 $output .= $paragraphStack;
1898                                                                 $paragraphStack = false;
1899                                                                 $this->mLastSection = 'p';
1900                                                         } else if ($this->mLastSection != 'p') {
1901                                                                 $output .= $this->closeParagraph().'<p>';
1902                                                                 $this->mLastSection = 'p';
1903                                                         }
1904                                                 }
1905                                         }
1906                                 }
1907                                 wfProfileOut( "$fname-paragraph" );
1908                         }
1909                         // somewhere above we forget to get out of pre block (bug 785)
1910                         if($preCloseMatch && $this->mInPre) {
1911                                 $this->mInPre = false;
1912                         }
1913                         if ($paragraphStack === false) {
1914                                 $output .= $t."\n";
1915                         }
1916                 }
1917                 while ( $prefixLength ) {
1918                         $output .= $this->closeList( $pref2{$prefixLength-1} );
1919                         --$prefixLength;
1920                 }
1921                 if ( '' != $this->mLastSection ) {
1922                         $output .= '</' . $this->mLastSection . '>';
1923                         $this->mLastSection = '';
1924                 }
1925
1926                 wfProfileOut( $fname );
1927                 return $output;
1928         }
1929
1930         /**
1931          * Split up a string on ':', ignoring any occurences inside
1932          * <a>..</a> or <span>...</span>
1933          * @param string $str the string to split
1934          * @param string &$before set to everything before the ':'
1935          * @param string &$after set to everything after the ':'
1936          * return string the position of the ':', or false if none found
1937          */
1938         function findColonNoLinks($str, &$before, &$after) {
1939                 # I wonder if we should make this count all tags, not just <a>
1940                 # and <span>. That would prevent us from matching a ':' that
1941                 # comes in the middle of italics other such formatting....
1942                 # -- Wil
1943                 $fname = 'Parser::findColonNoLinks';
1944                 wfProfileIn( $fname );
1945                 $pos = 0;
1946                 do {
1947                         $colon = strpos($str, ':', $pos);
1948
1949                         if ($colon !== false) {
1950                                 $before = substr($str, 0, $colon);
1951                                 $after = substr($str, $colon + 1);
1952
1953                                 # Skip any ':' within <a> or <span> pairs
1954                                 $a = substr_count($before, '<a');
1955                                 $s = substr_count($before, '<span');
1956                                 $ca = substr_count($before, '</a>');
1957                                 $cs = substr_count($before, '</span>');
1958
1959                                 if ($a <= $ca and $s <= $cs) {
1960                                         # Tags are balanced before ':'; ok
1961                                         break;
1962                                 }
1963                                 $pos = $colon + 1;
1964                         }
1965                 } while ($colon !== false);
1966                 wfProfileOut( $fname );
1967                 return $colon;
1968         }
1969
1970         /**
1971          * Return value of a magic variable (like PAGENAME)
1972          *
1973          * @access private
1974          */
1975         function getVariableValue( $index ) {
1976                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
1977
1978                 /**
1979                  * Some of these require message or data lookups and can be
1980                  * expensive to check many times.
1981                  */
1982                 static $varCache = array();
1983                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$varCache ) ) )
1984                         if ( isset( $varCache[$index] ) )
1985                                 return $varCache[$index];
1986
1987                 $ts = time();
1988                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
1989
1990                 switch ( $index ) {
1991                         case MAG_CURRENTMONTH:
1992                                 return $varCache[$index] = $wgContLang->formatNum( date( 'm', $ts ) );
1993                         case MAG_CURRENTMONTHNAME:
1994                                 return $varCache[$index] = $wgContLang->getMonthName( date( 'n', $ts ) );
1995                         case MAG_CURRENTMONTHNAMEGEN:
1996                                 return $varCache[$index] = $wgContLang->getMonthNameGen( date( 'n', $ts ) );
1997                         case MAG_CURRENTMONTHABBREV:
1998                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( date( 'n', $ts ) );
1999                         case MAG_CURRENTDAY:
2000                                 return $varCache[$index] = $wgContLang->formatNum( date( 'j', $ts ) );
2001                         case MAG_CURRENTDAY2:
2002                                 return $varCache[$index] = $wgContLang->formatNum( date( 'd', $ts ) );
2003                         case MAG_PAGENAME:
2004                                 return $this->mTitle->getText();
2005                         case MAG_PAGENAMEE:
2006                                 return $this->mTitle->getPartialURL();
2007                         case MAG_FULLPAGENAME:
2008                                 return $this->mTitle->getPrefixedText();
2009                         case MAG_FULLPAGENAMEE:
2010                                 return $this->mTitle->getPrefixedURL();
2011                         case MAG_REVISIONID:
2012                                 return $this->mRevisionId;
2013                         case MAG_NAMESPACE:
2014                                 return $wgContLang->getNsText( $this->mTitle->getNamespace() );
2015                         case MAG_NAMESPACEE:
2016                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2017                         case MAG_CURRENTDAYNAME:
2018                                 return $varCache[$index] = $wgContLang->getWeekdayName( date( 'w', $ts ) + 1 );
2019                         case MAG_CURRENTYEAR:
2020                                 return $varCache[$index] = $wgContLang->formatNum( date( 'Y', $ts ), true );
2021                         case MAG_CURRENTTIME:
2022                                 return $varCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2023                         case MAG_CURRENTWEEK:
2024                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2025                                 // int to remove the padding
2026                                 return $varCache[$index] = $wgContLang->formatNum( (int)date( 'W', $ts ) );
2027                         case MAG_CURRENTDOW:
2028                                 return $varCache[$index] = $wgContLang->formatNum( date( 'w', $ts ) );
2029                         case MAG_NUMBEROFARTICLES:
2030                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfArticles() );
2031                         case MAG_NUMBEROFFILES:
2032                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfFiles() );
2033                         case MAG_SITENAME:
2034                                 return $wgSitename;
2035                         case MAG_SERVER:
2036                                 return $wgServer;
2037                         case MAG_SERVERNAME:
2038                                 return $wgServerName;
2039                         case MAG_SCRIPTPATH:
2040                                 return $wgScriptPath;
2041                         default:
2042                                 $ret = null;
2043                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$varCache, &$index, &$ret ) ) )
2044                                         return $ret;
2045                                 else
2046                                         return null;
2047                 }
2048         }
2049
2050         /**
2051          * initialise the magic variables (like CURRENTMONTHNAME)
2052          *
2053          * @access private
2054          */
2055         function initialiseVariables() {
2056                 $fname = 'Parser::initialiseVariables';
2057                 wfProfileIn( $fname );
2058                 global $wgVariableIDs;
2059                 $this->mVariables = array();
2060                 foreach ( $wgVariableIDs as $id ) {
2061                         $mw =& MagicWord::get( $id );
2062                         $mw->addToArray( $this->mVariables, $id );
2063                 }
2064                 wfProfileOut( $fname );
2065         }
2066
2067         /**
2068          * parse any parentheses in format ((title|part|part))
2069          * and call callbacks to get a replacement text for any found piece
2070          *
2071          * @param string $text The text to parse
2072          * @param array $callbacks rules in form:
2073          *     '{' => array(                            # opening parentheses
2074          *                                      'end' => '}',   # closing parentheses
2075          *                                      'cb' => array(2 => callback,    # replacement callback to call if {{..}} is found
2076          *                                                                4 => callback         # replacement callback to call if {{{{..}}}} is found
2077          *                                                                )
2078          *                                      )
2079          * @access private
2080          */
2081         function replace_callback ($text, $callbacks) {
2082                 $openingBraceStack = array();   # this array will hold a stack of parentheses which are not closed yet
2083                 $lastOpeningBrace = -1;                 # last not closed parentheses
2084
2085                 for ($i = 0; $i < strlen($text); $i++) {
2086                         # check for any opening brace
2087                         $rule = null;
2088                         $nextPos = -1;
2089                         foreach ($callbacks as $key => $value) {
2090                                 $pos = strpos ($text, $key, $i);
2091                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)) {
2092                                         $rule = $value;
2093                                         $nextPos = $pos;
2094                                 }
2095                         }
2096
2097                         if ($lastOpeningBrace >= 0) {
2098                                 $pos = strpos ($text, $openingBraceStack[$lastOpeningBrace]['braceEnd'], $i);
2099
2100                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)){
2101                                         $rule = null;
2102                                         $nextPos = $pos;
2103                                 }
2104
2105                                 $pos = strpos ($text, '|', $i);
2106
2107                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)){
2108                                         $rule = null;
2109                                         $nextPos = $pos;
2110                                 }
2111                         }
2112
2113                         if ($nextPos == -1)
2114                                 break;
2115
2116                         $i = $nextPos;
2117
2118                         # found openning brace, lets add it to parentheses stack
2119                         if (null != $rule) {
2120                                 $piece = array('brace' => $text[$i],
2121                                                            'braceEnd' => $rule['end'],
2122                                                            'count' => 1,
2123                                                            'title' => '',
2124                                                            'parts' => null);
2125
2126                                 # count openning brace characters
2127                                 while ($i+1 < strlen($text) && $text[$i+1] == $piece['brace']) {
2128                                         $piece['count']++;
2129                                         $i++;
2130                                 }
2131
2132                                 $piece['startAt'] = $i+1;
2133                                 $piece['partStart'] = $i+1;
2134
2135                                 # we need to add to stack only if openning brace count is enough for any given rule
2136                                 foreach ($rule['cb'] as $cnt => $fn) {
2137                                         if ($piece['count'] >= $cnt) {
2138                                                 $lastOpeningBrace ++;
2139                                                 $openingBraceStack[$lastOpeningBrace] = $piece;
2140                                                 break;
2141                                         }
2142                                 }
2143
2144                                 continue;
2145                         }
2146                         else if ($lastOpeningBrace >= 0) {
2147                                 # first check if it is a closing brace
2148                                 if ($openingBraceStack[$lastOpeningBrace]['braceEnd'] == $text[$i]) {
2149                                         # lets check if it is enough characters for closing brace
2150                                         $count = 1;
2151                                         while ($i+$count < strlen($text) && $text[$i+$count] == $text[$i])
2152                                                 $count++;
2153
2154                                         # if there are more closing parentheses than opening ones, we parse less
2155                                         if ($openingBraceStack[$lastOpeningBrace]['count'] < $count)
2156                                                 $count = $openingBraceStack[$lastOpeningBrace]['count'];
2157
2158                                         # check for maximum matching characters (if there are 5 closing characters, we will probably need only 3 - depending on the rules)
2159                                         $matchingCount = 0;
2160                                         $matchingCallback = null;
2161                                         foreach ($callbacks[$openingBraceStack[$lastOpeningBrace]['brace']]['cb'] as $cnt => $fn) {
2162                                                 if ($count >= $cnt && $matchingCount < $cnt) {
2163                                                         $matchingCount = $cnt;
2164                                                         $matchingCallback = $fn;
2165                                                 }
2166                                         }
2167
2168                                         if ($matchingCount == 0) {
2169                                                 $i += $count - 1;
2170                                                 continue;
2171                                         }
2172
2173                                         # lets set a title or last part (if '|' was found)
2174                                         if (null === $openingBraceStack[$lastOpeningBrace]['parts'])
2175                                                 $openingBraceStack[$lastOpeningBrace]['title'] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2176                                         else
2177                                                 $openingBraceStack[$lastOpeningBrace]['parts'][] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2178
2179                                         $pieceStart = $openingBraceStack[$lastOpeningBrace]['startAt'] - $matchingCount;
2180                                         $pieceEnd = $i + $matchingCount;
2181
2182                                         if( is_callable( $matchingCallback ) ) {
2183                                                 $cbArgs = array (
2184                                                                                  'text' => substr($text, $pieceStart, $pieceEnd - $pieceStart),
2185                                                                                  'title' => trim($openingBraceStack[$lastOpeningBrace]['title']),
2186                                                                                  'parts' => $openingBraceStack[$lastOpeningBrace]['parts'],
2187                                                                                  'lineStart' => (($pieceStart > 0) && ($text[$pieceStart-1] == '\n')),
2188                                                                                  );
2189                                                 # finally we can call a user callback and replace piece of text
2190                                                 $replaceWith = call_user_func( $matchingCallback, $cbArgs );
2191                                                 $text = substr($text, 0, $pieceStart) . $replaceWith . substr($text, $pieceEnd);
2192                                                 $i = $pieceStart + strlen($replaceWith) - 1;
2193                                         }
2194                                         else {
2195                                                 # null value for callback means that parentheses should be parsed, but not replaced
2196                                                 $i += $matchingCount - 1;
2197                                         }
2198
2199                                         # reset last openning parentheses, but keep it in case there are unused characters
2200                                         $piece = array('brace' => $openingBraceStack[$lastOpeningBrace]['brace'],
2201                                                                    'braceEnd' => $openingBraceStack[$lastOpeningBrace]['braceEnd'],
2202                                                                    'count' => $openingBraceStack[$lastOpeningBrace]['count'],
2203                                                                    'title' => '',
2204                                                                    'parts' => null,
2205                                                                    'startAt' => $openingBraceStack[$lastOpeningBrace]['startAt']);
2206                                         $openingBraceStack[$lastOpeningBrace--] = null;
2207
2208                                         if ($matchingCount < $piece['count']) {
2209                                                 $piece['count'] -= $matchingCount;
2210                                                 $piece['startAt'] -= $matchingCount;
2211                                                 $piece['partStart'] = $piece['startAt'];
2212                                                 # do we still qualify for any callback with remaining count?
2213                                                 foreach ($callbacks[$piece['brace']]['cb'] as $cnt => $fn) {
2214                                                         if ($piece['count'] >= $cnt) {
2215                                                                 $lastOpeningBrace ++;
2216                                                                 $openingBraceStack[$lastOpeningBrace] = $piece;
2217                                                                 break;
2218                                                         }
2219                                                 }
2220                                         }
2221                                         continue;
2222                                 }
2223
2224                                 # lets set a title if it is a first separator, or next part otherwise
2225                                 if ($text[$i] == '|') {
2226                                         if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2227                                                 $openingBraceStack[$lastOpeningBrace]['title'] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2228                                                 $openingBraceStack[$lastOpeningBrace]['parts'] = array();
2229                                         }
2230                                         else
2231                                                 $openingBraceStack[$lastOpeningBrace]['parts'][] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2232
2233                                         $openingBraceStack[$lastOpeningBrace]['partStart'] = $i + 1;
2234                                 }
2235                         }
2236                 }
2237
2238                 return $text;
2239         }
2240
2241         /**
2242          * Replace magic variables, templates, and template arguments
2243          * with the appropriate text. Templates are substituted recursively,
2244          * taking care to avoid infinite loops.
2245          *
2246          * Note that the substitution depends on value of $mOutputType:
2247          *  OT_WIKI: only {{subst:}} templates
2248          *  OT_MSG: only magic variables
2249          *  OT_HTML: all templates and magic variables
2250          *
2251          * @param string $tex The text to transform
2252          * @param array $args Key-value pairs representing template parameters to substitute
2253          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2254          * @access private
2255          */
2256         function replaceVariables( $text, $args = array(), $argsOnly = false ) {
2257                 # Prevent too big inclusions
2258                 if( strlen( $text ) > MAX_INCLUDE_SIZE ) {
2259                         return $text;
2260                 }
2261
2262                 $fname = 'Parser::replaceVariables';
2263                 wfProfileIn( $fname );
2264
2265                 # This function is called recursively. To keep track of arguments we need a stack:
2266                 array_push( $this->mArgStack, $args );
2267
2268                 $braceCallbacks = array();
2269                 if ( !$argsOnly ) {
2270                         $braceCallbacks[2] = array( &$this, 'braceSubstitution' );
2271                 }
2272                 if ( $this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI ) {
2273                         $braceCallbacks[3] = array( &$this, 'argSubstitution' );
2274                 }
2275                 $callbacks = array();
2276                 $callbacks['{'] = array('end' => '}', 'cb' => $braceCallbacks);
2277                 $callbacks['['] = array('end' => ']', 'cb' => array(2=>null));
2278                 $text = $this->replace_callback ($text, $callbacks);
2279
2280                 array_pop( $this->mArgStack );
2281
2282                 wfProfileOut( $fname );
2283                 return $text;
2284         }
2285
2286         /**
2287          * Replace magic variables
2288          * @access private
2289          */
2290         function variableSubstitution( $matches ) {
2291                 $fname = 'Parser::variableSubstitution';
2292                 $varname = $matches[1];
2293                 wfProfileIn( $fname );
2294                 if ( !$this->mVariables ) {
2295                         $this->initialiseVariables();
2296                 }
2297                 $skip = false;
2298                 if ( $this->mOutputType == OT_WIKI ) {
2299                         # Do only magic variables prefixed by SUBST
2300                         $mwSubst =& MagicWord::get( MAG_SUBST );
2301                         if (!$mwSubst->matchStartAndRemove( $varname ))
2302                                 $skip = true;
2303                         # Note that if we don't substitute the variable below,
2304                         # we don't remove the {{subst:}} magic word, in case
2305                         # it is a template rather than a magic variable.
2306                 }
2307                 if ( !$skip && array_key_exists( $varname, $this->mVariables ) ) {
2308                         $id = $this->mVariables[$varname];
2309                         $text = $this->getVariableValue( $id );
2310                         $this->mOutput->mContainsOldMagic = true;
2311                 } else {
2312                         $text = $matches[0];
2313                 }
2314                 wfProfileOut( $fname );
2315                 return $text;
2316         }
2317
2318         # Split template arguments
2319         function getTemplateArgs( $argsString ) {
2320                 if ( $argsString === '' ) {
2321                         return array();
2322                 }
2323
2324                 $args = explode( '|', substr( $argsString, 1 ) );
2325
2326                 # If any of the arguments contains a '[[' but no ']]', it needs to be
2327                 # merged with the next arg because the '|' character between belongs
2328                 # to the link syntax and not the template parameter syntax.
2329                 $argc = count($args);
2330
2331                 for ( $i = 0; $i < $argc-1; $i++ ) {
2332                         if ( substr_count ( $args[$i], '[[' ) != substr_count ( $args[$i], ']]' ) ) {
2333                                 $args[$i] .= '|'.$args[$i+1];
2334                                 array_splice($args, $i+1, 1);
2335                                 $i--;
2336                                 $argc--;
2337                         }
2338                 }
2339
2340                 return $args;
2341         }
2342
2343         /**
2344          * Return the text of a template, after recursively
2345          * replacing any variables or templates within the template.
2346          *
2347          * @param array $piece The parts of the template
2348          *  $piece['text']: matched text
2349          *  $piece['title']: the title, i.e. the part before the |
2350          *  $piece['parts']: the parameter array
2351          * @return string the text of the template
2352          * @access private
2353          */
2354         function braceSubstitution( $piece ) {
2355                 global $wgContLang;
2356                 $fname = 'Parser::braceSubstitution';
2357                 wfProfileIn( $fname );
2358
2359                 # Flags
2360                 $found = false;             # $text has been filled
2361                 $nowiki = false;            # wiki markup in $text should be escaped
2362                 $noparse = false;           # Unsafe HTML tags should not be stripped, etc.
2363                 $noargs = false;            # Don't replace triple-brace arguments in $text
2364                 $replaceHeadings = false;   # Make the edit section links go to the template not the article
2365                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2366                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2367
2368                 # Title object, where $text came from
2369                 $title = NULL;
2370
2371                 $linestart = '';
2372
2373                 # $part1 is the bit before the first |, and must contain only title characters
2374                 # $args is a list of arguments, starting from index 0, not including $part1
2375
2376                 $part1 = $piece['title'];
2377                 # If the third subpattern matched anything, it will start with |
2378
2379                 if (null == $piece['parts']) {
2380                         $replaceWith = $this->variableSubstitution (array ($piece['text'], $piece['title']));
2381                         if ($replaceWith != $piece['text']) {
2382                                 $text = $replaceWith;
2383                                 $found = true;
2384                                 $noparse = true;
2385                                 $noargs = true;
2386                         }
2387                 }
2388
2389                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2390                 $argc = count( $args );
2391
2392                 # SUBST
2393                 if ( !$found ) {
2394                         $mwSubst =& MagicWord::get( MAG_SUBST );
2395                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor ($this->mOutputType == OT_WIKI) ) {
2396                                 # One of two possibilities is true:
2397                                 # 1) Found SUBST but not in the PST phase
2398                                 # 2) Didn't find SUBST and in the PST phase
2399                                 # In either case, return without further processing
2400                                 $text = $piece['text'];
2401                                 $found = true;
2402                                 $noparse = true;
2403                                 $noargs = true;
2404                         }
2405                 }
2406
2407                 # MSG, MSGNW, INT and RAW
2408                 if ( !$found ) {
2409                         # Check for MSGNW:
2410                         $mwMsgnw =& MagicWord::get( MAG_MSGNW );
2411                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2412                                 $nowiki = true;
2413                         } else {
2414                                 # Remove obsolete MSG:
2415                                 $mwMsg =& MagicWord::get( MAG_MSG );
2416                                 $mwMsg->matchStartAndRemove( $part1 );
2417                         }
2418
2419                         # Check for RAW:
2420                         $mwRaw =& MagicWord::get( MAG_RAW );
2421                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2422                                 $forceRawInterwiki = true;
2423                         }
2424
2425                         # Check if it is an internal message
2426                         $mwInt =& MagicWord::get( MAG_INT );
2427                         if ( $mwInt->matchStartAndRemove( $part1 ) ) {
2428                                 if ( $this->incrementIncludeCount( 'int:'.$part1 ) ) {
2429                                         $text = $linestart . wfMsgReal( $part1, $args, true );
2430                                         $found = true;
2431                                 }
2432                         }
2433                 }
2434
2435                 # NS
2436                 if ( !$found ) {
2437                         # Check for NS: (namespace expansion)
2438                         $mwNs = MagicWord::get( MAG_NS );
2439                         if ( $mwNs->matchStartAndRemove( $part1 ) ) {
2440                                 if ( intval( $part1 ) || $part1 == "0" ) {
2441                                         $text = $linestart . $wgContLang->getNsText( intval( $part1 ) );
2442                                         $found = true;
2443                                 } else {
2444                                         $index = Namespace::getCanonicalIndex( strtolower( $part1 ) );
2445                                         if ( !is_null( $index ) ) {
2446                                                 $text = $linestart . $wgContLang->getNsText( $index );
2447                                                 $found = true;
2448                                         }
2449                                 }
2450                         }
2451                 }
2452
2453                 # LCFIRST, UCFIRST, LC and UC
2454                 if ( !$found ) {
2455                         $lcfirst =& MagicWord::get( MAG_LCFIRST );
2456                         $ucfirst =& MagicWord::get( MAG_UCFIRST );
2457                         $lc =& MagicWord::get( MAG_LC );
2458                         $uc =& MagicWord::get( MAG_UC );
2459                         if ( $lcfirst->matchStartAndRemove( $part1 ) ) {
2460                                 $text = $linestart . $wgContLang->lcfirst( $part1 );
2461                                 $found = true;
2462                         } else if ( $ucfirst->matchStartAndRemove( $part1 ) ) {
2463                                 $text = $linestart . $wgContLang->ucfirst( $part1 );
2464                                 $found = true;
2465                         } else if ( $lc->matchStartAndRemove( $part1 ) ) {
2466                                 $text = $linestart . $wgContLang->lc( $part1 );
2467                                 $found = true;
2468                         } else if ( $uc->matchStartAndRemove( $part1 ) ) {
2469                                  $text = $linestart . $wgContLang->uc( $part1 );
2470                                  $found = true;
2471                         }
2472                 }
2473
2474                 # LOCALURL and FULLURL
2475                 if ( !$found ) {
2476                         $mwLocal =& MagicWord::get( MAG_LOCALURL );
2477                         $mwLocalE =& MagicWord::get( MAG_LOCALURLE );
2478                         $mwFull =& MagicWord::get( MAG_FULLURL );
2479                         $mwFullE =& MagicWord::get( MAG_FULLURLE );
2480
2481
2482                         if ( $mwLocal->matchStartAndRemove( $part1 ) ) {
2483                                 $func = 'getLocalURL';
2484                         } elseif ( $mwLocalE->matchStartAndRemove( $part1 ) ) {
2485                                 $func = 'escapeLocalURL';
2486                         } elseif ( $mwFull->matchStartAndRemove( $part1 ) ) {
2487                                 $func = 'getFullURL';
2488                         } elseif ( $mwFullE->matchStartAndRemove( $part1 ) ) {
2489                                 $func = 'escapeFullURL';
2490                         } else {
2491                                 $func = false;
2492                         }
2493
2494                         if ( $func !== false ) {
2495                                 $title = Title::newFromText( $part1 );
2496                                 if ( !is_null( $title ) ) {
2497                                         if ( $argc > 0 ) {
2498                                                 $text = $linestart . $title->$func( $args[0] );
2499                                         } else {
2500                                                 $text = $linestart . $title->$func();
2501                                         }
2502                                         $found = true;
2503                                 }
2504                         }
2505                 }
2506
2507                 # GRAMMAR
2508                 if ( !$found && $argc == 1 ) {
2509                         $mwGrammar =& MagicWord::get( MAG_GRAMMAR );
2510                         if ( $mwGrammar->matchStartAndRemove( $part1 ) ) {
2511                                 $text = $linestart . $wgContLang->convertGrammar( $args[0], $part1 );
2512                                 $found = true;
2513                         }
2514                 }
2515
2516                 # PLURAL
2517                 if ( !$found && $argc >= 2 ) {
2518                         $mwPluralForm =& MagicWord::get( MAG_PLURAL );
2519                         if ( $mwPluralForm->matchStartAndRemove( $part1 ) ) {
2520                                 if ($argc==2) {$args[2]=$args[1];}
2521                                 $text = $linestart . $wgContLang->convertPlural( $part1, $args[0], $args[1], $args[2]);
2522                                 $found = true;
2523                         }
2524                 }
2525
2526                 # Template table test
2527
2528                 # Did we encounter this template already? If yes, it is in the cache
2529                 # and we need to check for loops.
2530                 if ( !$found && isset( $this->mTemplates[$piece['title']] ) ) {
2531                         $found = true;
2532
2533                         # Infinite loop test
2534                         if ( isset( $this->mTemplatePath[$part1] ) ) {
2535                                 $noparse = true;
2536                                 $noargs = true;
2537                                 $found = true;
2538                                 $text = $linestart .
2539                                         '{{' . $part1 . '}}' .
2540                                         '<!-- WARNING: template loop detected -->';
2541                                 wfDebug( "$fname: template loop broken at '$part1'\n" );
2542                         } else {
2543                                 # set $text to cached message.
2544                                 $text = $linestart . $this->mTemplates[$piece['title']];
2545                         }
2546                 }
2547
2548                 # Load from database
2549                 $lastPathLevel = $this->mTemplatePath;
2550                 if ( !$found ) {
2551                         $ns = NS_TEMPLATE;
2552                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage='' );
2553                         if ($subpage !== '') {
2554                                 $ns = $this->mTitle->getNamespace();
2555                         }
2556                         $title = Title::newFromText( $part1, $ns );
2557
2558                         if ( !is_null( $title ) ) {
2559                                 if ( !$title->isExternal() ) {
2560                                         # Check for excessive inclusion
2561                                         $dbk = $title->getPrefixedDBkey();
2562                                         if ( $this->incrementIncludeCount( $dbk ) ) {
2563                                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() ) {
2564                                                         # Capture special page output
2565                                                         $text = SpecialPage::capturePath( $title );
2566                                                         if ( is_string( $text ) ) {
2567                                                                 $found = true;
2568                                                                 $noparse = true;
2569                                                                 $noargs = true;
2570                                                                 $isHTML = true;
2571                                                                 $this->disableCache();
2572                                                         }
2573                                                 } else {
2574                                                         $articleContent = $this->fetchTemplate( $title );
2575                                                         if ( $articleContent !== false ) {
2576                                                                 $found = true;
2577                                                                 $text = $articleContent;
2578                                                                 $replaceHeadings = true;
2579                                                         }
2580                                                 }
2581                                         }
2582
2583                                         # If the title is valid but undisplayable, make a link to it
2584                                         if ( $this->mOutputType == OT_HTML && !$found ) {
2585                                                 $text = '[['.$title->getPrefixedText().']]';
2586                                                 $found = true;
2587                                         }
2588                                 } elseif ( $title->isTrans() ) {
2589                                         // Interwiki transclusion
2590                                         if ( $this->mOutputType == OT_HTML && !$forceRawInterwiki ) {
2591                                                 $text = $this->interwikiTransclude( $title, 'render' );
2592                                                 $isHTML = true;
2593                                                 $noparse = true;
2594                                         } else {
2595                                                 $text = $this->interwikiTransclude( $title, 'raw' );
2596                                                 $replaceHeadings = true;
2597                                         }
2598                                         $found = true;
2599                                 }
2600
2601                                 # Template cache array insertion
2602                                 # Use the original $piece['title'] not the mangled $part1, so that
2603                                 # modifiers such as RAW: produce separate cache entries
2604                                 if( $found ) {
2605                                         $this->mTemplates[$piece['title']] = $text;
2606                                         $text = $linestart . $text;
2607                                 }
2608                         }
2609                 }
2610
2611                 # Recursive parsing, escaping and link table handling
2612                 # Only for HTML output
2613                 if ( $nowiki && $found && $this->mOutputType == OT_HTML ) {
2614                         $text = wfEscapeWikiText( $text );
2615                 } elseif ( ($this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI) && $found ) {
2616                         if ( !$noargs ) {
2617                                 # Clean up argument array
2618                                 $assocArgs = array();
2619                                 $index = 1;
2620                                 foreach( $args as $arg ) {
2621                                         $eqpos = strpos( $arg, '=' );
2622                                         if ( $eqpos === false ) {
2623                                                 $assocArgs[$index++] = $arg;
2624                                         } else {
2625                                                 $name = trim( substr( $arg, 0, $eqpos ) );
2626                                                 $value = trim( substr( $arg, $eqpos+1 ) );
2627                                                 if ( $value === false ) {
2628                                                         $value = '';
2629                                                 }
2630                                                 if ( $name !== false ) {
2631                                                         $assocArgs[$name] = $value;
2632                                                 }
2633                                         }
2634                                 }
2635
2636                                 # Add a new element to the templace recursion path
2637                                 $this->mTemplatePath[$part1] = 1;
2638                         }
2639
2640                         if ( !$noparse ) {
2641                                 # If there are any <onlyinclude> tags, only include them
2642                                 if ( in_string( '<onlyinclude>', $text ) && in_string( '</onlyinclude>', $text ) ) {
2643                                         preg_match_all( '/<onlyinclude>(.*?)\n?<\/onlyinclude>/s', $text, $m );
2644                                         $text = '';
2645                                         foreach ($m[1] as $piece)
2646                                                 $text .= $piece;
2647                                 }
2648                                 # Remove <noinclude> sections and <includeonly> tags
2649                                 $text = preg_replace( '/<noinclude>.*?<\/noinclude>/s', '', $text );
2650                                 $text = strtr( $text, array( '<includeonly>' => '' , '</includeonly>' => '' ) );
2651
2652                                 if( $this->mOutputType == OT_HTML ) {
2653                                         # Strip <nowiki>, <pre>, etc.
2654                                         $text = $this->strip( $text, $this->mStripState );
2655                                         $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'replaceVariables' ), $assocArgs );
2656                                 }
2657                                 $text = $this->replaceVariables( $text, $assocArgs );
2658
2659                                 # If the template begins with a table or block-level
2660                                 # element, it should be treated as beginning a new line.
2661                                 if (!$piece['lineStart'] && preg_match('/^({\\||:|;|#|\*)/', $text)) {
2662                                         $text = "\n" . $text;
2663                                 }
2664                         } elseif ( !$noargs ) {
2665                                 # $noparse and !$noargs
2666                                 # Just replace the arguments, not any double-brace items
2667                                 # This is used for rendered interwiki transclusion
2668                                 $text = $this->replaceVariables( $text, $assocArgs, true );
2669                         }
2670                 }
2671                 # Prune lower levels off the recursion check path
2672                 $this->mTemplatePath = $lastPathLevel;
2673
2674                 if ( !$found ) {
2675                         wfProfileOut( $fname );
2676                         return $piece['text'];
2677                 } else {
2678                         if ( $isHTML ) {
2679                                 # Replace raw HTML by a placeholder
2680                                 # Add a blank line preceding, to prevent it from mucking up
2681                                 # immediately preceding headings
2682                                 $text = "\n\n" . $this->insertStripItem( $text, $this->mStripState );
2683                         } else {
2684                                 # replace ==section headers==
2685                                 # XXX this needs to go away once we have a better parser.
2686                                 if ( $this->mOutputType != OT_WIKI && $replaceHeadings ) {
2687                                         if( !is_null( $title ) )
2688                                                 $encodedname = base64_encode($title->getPrefixedDBkey());
2689                                         else
2690                                                 $encodedname = base64_encode("");
2691                                         $m = preg_split('/(^={1,6}.*?={1,6}\s*?$)/m', $text, -1,
2692                                                 PREG_SPLIT_DELIM_CAPTURE);
2693                                         $text = '';
2694                                         $nsec = 0;
2695                                         for( $i = 0; $i < count($m); $i += 2 ) {
2696                                                 $text .= $m[$i];
2697                                                 if (!isset($m[$i + 1]) || $m[$i + 1] == "") continue;
2698                                                 $hl = $m[$i + 1];
2699                                                 if( strstr($hl, "<!--MWTEMPLATESECTION") ) {
2700                                                         $text .= $hl;
2701                                                         continue;
2702                                                 }
2703                                                 preg_match('/^(={1,6})(.*?)(={1,6})\s*?$/m', $hl, $m2);
2704                                                 $text .= $m2[1] . $m2[2] . "<!--MWTEMPLATESECTION="
2705                                                         . $encodedname . "&" . base64_encode("$nsec") . "-->" . $m2[3];
2706
2707                                                 $nsec++;
2708                                         }
2709                                 }
2710                         }
2711                 }
2712
2713                 # Prune lower levels off the recursion check path
2714                 $this->mTemplatePath = $lastPathLevel;
2715
2716                 if ( !$found ) {
2717                         wfProfileOut( $fname );
2718                         return $piece['text'];
2719                 } else {
2720                         wfProfileOut( $fname );
2721                         return $text;
2722                 }
2723         }
2724
2725         /**
2726          * Fetch the unparsed text of a template and register a reference to it.
2727          */
2728         function fetchTemplate( $title ) {
2729                 $text = false;
2730                 // Loop to fetch the article, with up to 1 redirect
2731                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
2732                         $rev = Revision::newFromTitle( $title );
2733                         $this->mOutput->addTemplate( $title, $title->getArticleID() );
2734                         if ( !$rev ) {
2735                                 break;
2736                         }
2737                         $text = $rev->getText();
2738                         if ( $text === false ) {
2739                                 break;
2740                         }
2741                         // Redirect?
2742                         $title = Title::newFromRedirect( $text );
2743                 }
2744                 return $text;
2745         }
2746
2747         /**
2748          * Transclude an interwiki link.
2749          */
2750         function interwikiTransclude( $title, $action ) {
2751                 global $wgEnableScaryTranscluding, $wgCanonicalNamespaceNames;
2752
2753                 if (!$wgEnableScaryTranscluding)
2754                         return wfMsg('scarytranscludedisabled');
2755
2756                 // The namespace will actually only be 0 or 10, depending on whether there was a leading :
2757                 // But we'll handle it generally anyway
2758                 if ( $title->getNamespace() ) {
2759                         // Use the canonical namespace, which should work anywhere
2760                         $articleName = $wgCanonicalNamespaceNames[$title->getNamespace()] . ':' . $title->getDBkey();
2761                 } else {
2762                         $articleName = $title->getDBkey();
2763                 }
2764
2765                 $url = str_replace('$1', urlencode($articleName), Title::getInterwikiLink($title->getInterwiki()));
2766                 $url .= "?action=$action";
2767                 if (strlen($url) > 255)
2768                         return wfMsg('scarytranscludetoolong');
2769                 return $this->fetchScaryTemplateMaybeFromCache($url);
2770         }
2771
2772         function fetchScaryTemplateMaybeFromCache($url) {
2773                 global $wgTranscludeCacheExpiry;
2774                 $dbr =& wfGetDB(DB_SLAVE);
2775                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
2776                                 array('tc_url' => $url));
2777                 if ($obj) {
2778                         $time = $obj->tc_time;
2779                         $text = $obj->tc_contents;
2780                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
2781                                 return $text;
2782                         }
2783                 }
2784
2785                 $text = wfGetHTTP($url);
2786                 if (!$text)
2787                         return wfMsg('scarytranscludefailed', $url);
2788
2789                 $dbw =& wfGetDB(DB_MASTER);
2790                 $dbw->replace('transcache', array('tc_url'), array(
2791                         'tc_url' => $url,
2792                         'tc_time' => time(),
2793                         'tc_contents' => $text));
2794                 return $text;
2795         }
2796
2797
2798         /**
2799          * Triple brace replacement -- used for template arguments
2800          * @access private
2801          */
2802         function argSubstitution( $matches ) {
2803                 $arg = trim( $matches['title'] );
2804                 $text = $matches['text'];
2805                 $inputArgs = end( $this->mArgStack );
2806
2807                 if ( array_key_exists( $arg, $inputArgs ) ) {
2808                         $text = $inputArgs[$arg];
2809                 } else if ($this->mOutputType == OT_HTML && null != $matches['parts'] && count($matches['parts']) > 0) {
2810                         $text = $matches['parts'][0];
2811                 }
2812
2813                 return $text;
2814         }
2815
2816         /**
2817          * Returns true if the function is allowed to include this entity
2818          * @access private
2819          */
2820         function incrementIncludeCount( $dbk ) {
2821                 if ( !array_key_exists( $dbk, $this->mIncludeCount ) ) {
2822                         $this->mIncludeCount[$dbk] = 0;
2823                 }
2824                 if ( ++$this->mIncludeCount[$dbk] <= MAX_INCLUDE_REPEAT ) {
2825                         return true;
2826                 } else {
2827                         return false;
2828                 }
2829         }
2830
2831         /**
2832          * This function accomplishes several tasks:
2833          * 1) Auto-number headings if that option is enabled
2834          * 2) Add an [edit] link to sections for logged in users who have enabled the option
2835          * 3) Add a Table of contents on the top for users who have enabled the option
2836          * 4) Auto-anchor headings
2837          *
2838          * It loops through all headlines, collects the necessary data, then splits up the
2839          * string and re-inserts the newly formatted headlines.
2840          *
2841          * @param string $text
2842          * @param boolean $isMain
2843          * @access private
2844          */
2845         function formatHeadings( $text, $isMain=true ) {
2846                 global $wgMaxTocLevel, $wgContLang;
2847
2848                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
2849                 $doShowToc = true;
2850                 $forceTocHere = false;
2851                 if( !$this->mTitle->userCanEdit() ) {
2852                         $showEditLink = 0;
2853                 } else {
2854                         $showEditLink = $this->mOptions->getEditSection();
2855                 }
2856
2857                 # Inhibit editsection links if requested in the page
2858                 $esw =& MagicWord::get( MAG_NOEDITSECTION );
2859                 if( $esw->matchAndRemove( $text ) ) {
2860                         $showEditLink = 0;
2861                 }
2862                 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
2863                 # do not add TOC
2864                 $mw =& MagicWord::get( MAG_NOTOC );
2865                 if( $mw->matchAndRemove( $text ) ) {
2866                         $doShowToc = false;
2867                 }
2868
2869                 # Get all headlines for numbering them and adding funky stuff like [edit]
2870                 # links - this is for later, but we need the number of headlines right now
2871                 $numMatches = preg_match_all( '/<H([1-6])(.*?'.'>)(.*?)<\/H[1-6] *>/i', $text, $matches );
2872
2873                 # if there are fewer than 4 headlines in the article, do not show TOC
2874                 if( $numMatches < 4 ) {
2875                         $doShowToc = false;
2876                 }
2877
2878                 # if the string __TOC__ (not case-sensitive) occurs in the HTML,
2879                 # override above conditions and always show TOC at that place
2880
2881                 $mw =& MagicWord::get( MAG_TOC );
2882                 if($mw->match( $text ) ) {
2883                         $doShowToc = true;
2884                         $forceTocHere = true;
2885                 } else {
2886                         # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
2887                         # override above conditions and always show TOC above first header
2888                         $mw =& MagicWord::get( MAG_FORCETOC );
2889                         if ($mw->matchAndRemove( $text ) ) {
2890                                 $doShowToc = true;
2891                         }
2892                 }
2893
2894                 # Never ever show TOC if no headers
2895                 if( $numMatches < 1 ) {
2896                         $doShowToc = false;
2897                 }
2898
2899                 # We need this to perform operations on the HTML
2900                 $sk =& $this->mOptions->getSkin();
2901
2902                 # headline counter
2903                 $headlineCount = 0;
2904                 $sectionCount = 0; # headlineCount excluding template sections
2905
2906                 # Ugh .. the TOC should have neat indentation levels which can be
2907                 # passed to the skin functions. These are determined here
2908                 $toc = '';
2909                 $full = '';
2910                 $head = array();
2911                 $sublevelCount = array();
2912                 $levelCount = array();
2913                 $toclevel = 0;
2914                 $level = 0;
2915                 $prevlevel = 0;
2916                 $toclevel = 0;
2917                 $prevtoclevel = 0;
2918
2919                 foreach( $matches[3] as $headline ) {
2920                         $istemplate = 0;
2921                         $templatetitle = '';
2922                         $templatesection = 0;
2923                         $numbering = '';
2924
2925                         if (preg_match("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", $headline, $mat)) {
2926                                 $istemplate = 1;
2927                                 $templatetitle = base64_decode($mat[1]);
2928                                 $templatesection = 1 + (int)base64_decode($mat[2]);
2929                                 $headline = preg_replace("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", "", $headline);
2930                         }
2931
2932                         if( $toclevel ) {
2933                                 $prevlevel = $level;
2934                                 $prevtoclevel = $toclevel;
2935                         }
2936                         $level = $matches[1][$headlineCount];
2937
2938                         if( $doNumberHeadings || $doShowToc ) {
2939
2940                                 if ( $level > $prevlevel ) {
2941                                         # Increase TOC level
2942                                         $toclevel++;
2943                                         $sublevelCount[$toclevel] = 0;
2944                                         $toc .= $sk->tocIndent();
2945                                 }
2946                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
2947                                         # Decrease TOC level, find level to jump to
2948
2949                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
2950                                                 # Can only go down to level 1
2951                                                 $toclevel = 1;
2952                                         } else {
2953                                                 for ($i = $toclevel; $i > 0; $i--) {
2954                                                         if ( $levelCount[$i] == $level ) {
2955                                                                 # Found last matching level
2956                                                                 $toclevel = $i;
2957                                                                 break;
2958                                                         }
2959                                                         elseif ( $levelCount[$i] < $level ) {
2960                                                                 # Found first matching level below current level
2961                                                                 $toclevel = $i + 1;
2962                                                                 break;
2963                                                         }
2964                                                 }
2965                                         }
2966
2967                                         $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
2968                                 }
2969                                 else {
2970                                         # No change in level, end TOC line
2971                                         $toc .= $sk->tocLineEnd();
2972                                 }
2973
2974                                 $levelCount[$toclevel] = $level;
2975
2976                                 # count number of headlines for each level
2977                                 @$sublevelCount[$toclevel]++;
2978                                 $dot = 0;
2979                                 for( $i = 1; $i <= $toclevel; $i++ ) {
2980                                         if( !empty( $sublevelCount[$i] ) ) {
2981                                                 if( $dot ) {
2982                                                         $numbering .= '.';
2983                                                 }
2984                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
2985                                                 $dot = 1;
2986                                         }
2987                                 }
2988                         }
2989
2990                         # The canonized header is a version of the header text safe to use for links
2991                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
2992                         $canonized_headline = $this->unstrip( $headline, $this->mStripState );
2993                         $canonized_headline = $this->unstripNoWiki( $canonized_headline, $this->mStripState );
2994
2995                         # Remove link placeholders by the link text.
2996                         #     <!--LINK number-->
2997                         # turns into
2998                         #     link text with suffix
2999                         $canonized_headline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
3000                                                             "\$this->mLinkHolders['texts'][\$1]",
3001                                                             $canonized_headline );
3002                         $canonized_headline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
3003                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
3004                                                             $canonized_headline );
3005
3006                         # strip out HTML
3007                         $canonized_headline = preg_replace( '/<.*?' . '>/','',$canonized_headline );
3008                         $tocline = trim( $canonized_headline );
3009                         $canonized_headline = Sanitizer::escapeId( $tocline );
3010                         $refers[$headlineCount] = $canonized_headline;
3011
3012                         # count how many in assoc. array so we can track dupes in anchors
3013                         @$refers[$canonized_headline]++;
3014                         $refcount[$headlineCount]=$refers[$canonized_headline];
3015
3016                         # Don't number the heading if it is the only one (looks silly)
3017                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3018                                 # the two are different if the line contains a link
3019                                 $headline=$numbering . ' ' . $headline;
3020                         }
3021
3022                         # Create the anchor for linking from the TOC to the section
3023                         $anchor = $canonized_headline;
3024                         if($refcount[$headlineCount] > 1 ) {
3025                                 $anchor .= '_' . $refcount[$headlineCount];
3026                         }
3027                         if( $doShowToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3028                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3029                         }
3030                         if( $showEditLink && ( !$istemplate || $templatetitle !== "" ) ) {
3031                                 if ( empty( $head[$headlineCount] ) ) {
3032                                         $head[$headlineCount] = '';
3033                                 }
3034                                 if( $istemplate )
3035                                         $head[$headlineCount] .= $sk->editSectionLinkForOther($templatetitle, $templatesection);
3036                                 else
3037                                         $head[$headlineCount] .= $sk->editSectionLink($this->mTitle, $sectionCount+1);
3038                         }
3039
3040                         # give headline the correct <h#> tag
3041                         @$head[$headlineCount] .= "<a name=\"$anchor\"></a><h".$level.$matches[2][$headlineCount] .$headline.'</h'.$level.'>';
3042
3043                         $headlineCount++;
3044                         if( !$istemplate )
3045                                 $sectionCount++;
3046                 }
3047
3048                 if( $doShowToc ) {
3049                         $toc .= $sk->tocUnindent( $toclevel - 1 );
3050                         $toc = $sk->tocList( $toc );
3051                 }
3052
3053                 # split up and insert constructed headlines
3054
3055                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3056                 $i = 0;
3057
3058                 foreach( $blocks as $block ) {
3059                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
3060                                 # This is the [edit] link that appears for the top block of text when
3061                                 # section editing is enabled
3062
3063                                 # Disabled because it broke block formatting
3064                                 # For example, a bullet point in the top line
3065                                 # $full .= $sk->editSectionLink(0);
3066                         }
3067                         $full .= $block;
3068                         if( $doShowToc && !$i && $isMain && !$forceTocHere) {
3069                         # Top anchor now in skin
3070                                 $full = $full.$toc;
3071                         }
3072
3073                         if( !empty( $head[$i] ) ) {
3074                                 $full .= $head[$i];
3075                         }
3076                         $i++;
3077                 }
3078                 if($forceTocHere) {
3079                         $mw =& MagicWord::get( MAG_TOC );
3080                         return $mw->replace( $toc, $full );
3081                 } else {
3082                         return $full;
3083                 }
3084         }
3085
3086         /**
3087          * Return an HTML link for the "ISBN 123456" text
3088          * @access private
3089          */
3090         function magicISBN( $text ) {
3091                 $fname = 'Parser::magicISBN';
3092                 wfProfileIn( $fname );
3093
3094                 $a = split( 'ISBN ', ' '.$text );
3095                 if ( count ( $a ) < 2 ) {
3096                         wfProfileOut( $fname );
3097                         return $text;
3098                 }
3099                 $text = substr( array_shift( $a ), 1);
3100                 $valid = '0123456789-Xx';
3101
3102                 foreach ( $a as $x ) {
3103                         $isbn = $blank = '' ;
3104                         while ( ' ' == $x{0} ) {
3105                                 $blank .= ' ';
3106                                 $x = substr( $x, 1 );
3107                         }
3108                         if ( $x == '' ) { # blank isbn
3109                                 $text .= "ISBN $blank";
3110                                 continue;
3111                         }
3112                         while ( strstr( $valid, $x{0} ) != false ) {
3113                                 $isbn .= $x{0};
3114                                 $x = substr( $x, 1 );
3115                         }
3116                         $num = str_replace( '-', '', $isbn );
3117                         $num = str_replace( ' ', '', $num );
3118                         $num = str_replace( 'x', 'X', $num );
3119
3120                         if ( '' == $num ) {
3121                                 $text .= "ISBN $blank$x";
3122                         } else {
3123                                 $titleObj = Title::makeTitle( NS_SPECIAL, 'Booksources' );
3124                                 $text .= '<a href="' .
3125                                 $titleObj->escapeLocalUrl( 'isbn='.$num ) .
3126                                         "\" class=\"internal\">ISBN $isbn</a>";
3127                                 $text .= $x;
3128                         }
3129                 }
3130                 wfProfileOut( $fname );
3131                 return $text;
3132         }
3133
3134         /**
3135          * Return an HTML link for the "RFC 1234" text
3136          *
3137          * @access private
3138          * @param string $text     Text to be processed
3139          * @param string $keyword  Magic keyword to use (default RFC)
3140          * @param string $urlmsg   Interface message to use (default rfcurl)
3141          * @return string
3142          */
3143         function magicRFC( $text, $keyword='RFC ', $urlmsg='rfcurl'  ) {
3144
3145                 $valid = '0123456789';
3146                 $internal = false;
3147
3148                 $a = split( $keyword, ' '.$text );
3149                 if ( count ( $a ) < 2 ) {
3150                         return $text;
3151                 }
3152                 $text = substr( array_shift( $a ), 1);
3153
3154                 /* Check if keyword is preceed by [[.
3155                  * This test is made here cause of the array_shift above
3156                  * that prevent the test to be done in the foreach.
3157                  */
3158                 if ( substr( $text, -2 ) == '[[' ) {
3159                         $internal = true;
3160                 }
3161
3162                 foreach ( $a as $x ) {
3163                         /* token might be empty if we have RFC RFC 1234 */
3164                         if ( $x=='' ) {
3165                                 $text.=$keyword;
3166                                 continue;
3167                                 }
3168
3169                         $id = $blank = '' ;
3170
3171                         /** remove and save whitespaces in $blank */
3172                         while ( $x{0} == ' ' ) {
3173                                 $blank .= ' ';
3174                                 $x = substr( $x, 1 );
3175                         }
3176
3177                         /** remove and save the rfc number in $id */
3178                         while ( strstr( $valid, $x{0} ) != false ) {
3179                                 $id .= $x{0};
3180                                 $x = substr( $x, 1 );
3181                         }
3182
3183                         if ( $id == '' ) {
3184                                 /* call back stripped spaces*/
3185                                 $text .= $keyword.$blank.$x;
3186                         } elseif( $internal ) {
3187                                 /* normal link */
3188                                 $text .= $keyword.$id.$x;
3189                         } else {
3190                                 /* build the external link*/
3191                                 $url = wfMsg( $urlmsg, $id);
3192                                 $sk =& $this->mOptions->getSkin();
3193                                 $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
3194                                 $text .= "<a href='{$url}'{$la}>{$keyword}{$id}</a>{$x}";
3195                         }
3196
3197                         /* Check if the next RFC keyword is preceed by [[ */
3198                         $internal = ( substr($x,-2) == '[[' );
3199                 }
3200                 return $text;
3201         }
3202
3203         /**
3204          * Transform wiki markup when saving a page by doing \r\n -> \n
3205          * conversion, substitting signatures, {{subst:}} templates, etc.
3206          *
3207          * @param string $text the text to transform
3208          * @param Title &$title the Title object for the current article
3209          * @param User &$user the User object describing the current user
3210          * @param ParserOptions $options parsing options
3211          * @param bool $clearState whether to clear the parser state first
3212          * @return string the altered wiki markup
3213          * @access public
3214          */
3215         function preSaveTransform( $text, &$title, &$user, $options, $clearState = true ) {
3216                 $this->mOptions = $options;
3217                 $this->mTitle =& $title;
3218                 $this->mOutputType = OT_WIKI;
3219
3220                 if ( $clearState ) {
3221                         $this->clearState();
3222                 }
3223
3224                 $stripState = false;
3225                 $pairs = array(
3226                         "\r\n" => "\n",
3227                 );
3228                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3229                 $text = $this->strip( $text, $stripState, true );
3230                 $text = $this->pstPass2( $text, $user );
3231                 $text = $this->unstrip( $text, $stripState );
3232                 $text = $this->unstripNoWiki( $text, $stripState );
3233                 return $text;
3234         }
3235
3236         /**
3237          * Pre-save transform helper function
3238          * @access private
3239          */
3240         function pstPass2( $text, &$user ) {
3241                 global $wgContLang, $wgLocaltimezone;
3242
3243                 /* Note: This is the timestamp saved as hardcoded wikitext to
3244                  * the database, we use $wgContLang here in order to give
3245                  * everyone the same signiture and use the default one rather
3246                  * than the one selected in each users preferences.
3247                  */
3248                 if ( isset( $wgLocaltimezone ) ) {
3249                         $oldtz = getenv( 'TZ' );
3250                         putenv( 'TZ='.$wgLocaltimezone );
3251                 }
3252                 $d = $wgContLang->timeanddate( date( 'YmdHis' ), false, false) .
3253                   ' (' . date( 'T' ) . ')';
3254                 if ( isset( $wgLocaltimezone ) ) {
3255                         putenv( 'TZ='.$oldtz );
3256                 }
3257
3258                 # Variable replacement
3259                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3260                 $text = $this->replaceVariables( $text );
3261
3262                 # Signatures
3263                 $sigText = $this->getUserSig( $user );
3264                 $text = strtr( $text, array(
3265                         '~~~~~' => $d,
3266                         '~~~~' => "$sigText $d",
3267                         '~~~' => $sigText
3268                 ) );
3269
3270                 # Context links: [[|name]] and [[name (context)|]]
3271                 #
3272                 global $wgLegalTitleChars;
3273                 $tc = "[$wgLegalTitleChars]";
3274                 $np = str_replace( array( '(', ')' ), array( '', '' ), $tc ); # No parens
3275
3276                 $namespacechar = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
3277                 $conpat = "/^({$np}+) \\(({$tc}+)\\)$/";
3278
3279                 $p1 = "/\[\[({$np}+) \\(({$np}+)\\)\\|]]/";             # [[page (context)|]]
3280                 $p2 = "/\[\[\\|({$tc}+)]]/";                                    # [[|page]]
3281                 $p3 = "/\[\[(:*$namespacechar+):({$np}+)\\|]]/";                # [[namespace:page|]] and [[:namespace:page|]]
3282                 $p4 = "/\[\[(:*$namespacechar+):({$np}+) \\(({$np}+)\\)\\|]]/"; # [[ns:page (cont)|]] and [[:ns:page (cont)|]]
3283                 $context = '';
3284                 $t = $this->mTitle->getText();
3285                 if ( preg_match( $conpat, $t, $m ) ) {
3286                         $context = $m[2];
3287                 }
3288                 $text = preg_replace( $p4, '[[\\1:\\2 (\\3)|\\2]]', $text );
3289                 $text = preg_replace( $p1, '[[\\1 (\\2)|\\1]]', $text );
3290                 $text = preg_replace( $p3, '[[\\1:\\2|\\2]]', $text );
3291
3292                 if ( '' == $context ) {
3293                         $text = preg_replace( $p2, '[[\\1]]', $text );
3294                 } else {
3295                         $text = preg_replace( $p2, "[[\\1 ({$context})|\\1]]", $text );
3296                 }
3297
3298                 # Trim trailing whitespace
3299                 # MAG_END (__END__) tag allows for trailing
3300                 # whitespace to be deliberately included
3301                 $text = rtrim( $text );
3302                 $mw =& MagicWord::get( MAG_END );
3303                 $mw->matchAndRemove( $text );
3304
3305                 return $text;
3306         }
3307
3308         /**
3309          * Fetch the user's signature text, if any, and normalize to
3310          * validated, ready-to-insert wikitext.
3311          *
3312          * @param User $user
3313          * @return string
3314          * @access private
3315          */
3316         function getUserSig( &$user ) {
3317                 global $wgContLang;
3318
3319                 $username = $user->getName();
3320                 $nickname = $user->getOption( 'nickname' );
3321                 $nickname = $nickname === '' ? $username : $nickname;
3322
3323                 if( $user->getBoolOption( 'fancysig' ) !== false ) {
3324                         # Sig. might contain markup; validate this
3325                         if( $this->validateSig( $nickname ) !== false ) {
3326                                 # Validated; clean up (if needed) and return it
3327                                 return( $this->cleanSig( $nickname ) );
3328                         } else {
3329                                 # Failed to validate; fall back to the default
3330                                 $nickname = $username;
3331                                 wfDebug( "Parser::getUserSig: $username has bad XML tags in signature.\n" );
3332                         }
3333                 }
3334
3335                 # If we're still here, make it a link to the user page
3336                 $userpage = $user->getUserPage();
3337                 return( '[[' . $userpage->getPrefixedText() . '|' . wfEscapeWikiText( $nickname ) . ']]' );
3338         }
3339
3340         /**
3341          * Check that the user's signature contains no bad XML
3342          *
3343          * @param string $text
3344          * @return mixed An expanded string, or false if invalid.
3345          */
3346         function validateSig( $text ) {
3347                 return( wfIsWellFormedXmlFragment( $text ) ? $text : false );
3348         }
3349
3350         /**
3351          * Clean up signature text
3352          *
3353          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures
3354          * 2) Substitute all transclusions
3355          *
3356          * @param string $text
3357          * @return string Signature text
3358          */
3359         function cleanSig( $text ) {
3360                 $substWord = MagicWord::get( MAG_SUBST );
3361                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3362                 $substText = '{{' . $substWord->getSynonym( 0 );
3363
3364                 $text = preg_replace( $substRegex, $substText, $text );
3365                 $text = preg_replace( '/~{3,5}/', '', $text );
3366                 $text = $this->replaceVariables( $text );
3367
3368                 return $text;
3369         }
3370
3371         /**
3372          * Set up some variables which are usually set up in parse()
3373          * so that an external function can call some class members with confidence
3374          * @access public
3375          */
3376         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
3377                 $this->mTitle =& $title;
3378                 $this->mOptions = $options;
3379                 $this->mOutputType = $outputType;
3380                 if ( $clearState ) {
3381                         $this->clearState();
3382                 }
3383         }
3384
3385         /**
3386          * Transform a MediaWiki message by replacing magic variables.
3387          *
3388          * @param string $text the text to transform
3389          * @param ParserOptions $options  options
3390          * @return string the text with variables substituted
3391          * @access public
3392          */
3393         function transformMsg( $text, $options ) {
3394                 global $wgTitle;
3395                 static $executing = false;
3396
3397                 $fname = "Parser::transformMsg";
3398
3399                 # Guard against infinite recursion
3400                 if ( $executing ) {
3401                         return $text;
3402                 }
3403                 $executing = true;
3404
3405                 wfProfileIn($fname);
3406
3407                 $this->mTitle = $wgTitle;
3408                 $this->mOptions = $options;
3409                 $this->mOutputType = OT_MSG;
3410                 $this->clearState();
3411                 $text = $this->replaceVariables( $text );
3412
3413                 $executing = false;
3414                 wfProfileOut($fname);
3415                 return $text;
3416         }
3417
3418         /**
3419          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
3420          * Callback will be called with the text within
3421          * Transform and return the text within
3422          *
3423          * @access public
3424          *
3425          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
3426          * @param mixed $callback The callback function (and object) to use for the tag
3427          *
3428          * @return The old value of the mTagHooks array associated with the hook
3429          */
3430         function setHook( $tag, $callback ) {
3431                 $oldVal = @$this->mTagHooks[$tag];
3432                 $this->mTagHooks[$tag] = $callback;
3433
3434                 return $oldVal;
3435         }
3436
3437         /**
3438          * Replace <!--LINK--> link placeholders with actual links, in the buffer
3439          * Placeholders created in Skin::makeLinkObj()
3440          * Returns an array of links found, indexed by PDBK:
3441          *  0 - broken
3442          *  1 - normal link
3443          *  2 - stub
3444          * $options is a bit field, RLH_FOR_UPDATE to select for update
3445          */
3446         function replaceLinkHolders( &$text, $options = 0 ) {
3447                 global $wgUser;
3448                 global $wgOutputReplace;
3449
3450                 $fname = 'Parser::replaceLinkHolders';
3451                 wfProfileIn( $fname );
3452
3453                 $pdbks = array();
3454                 $colours = array();
3455                 $sk =& $this->mOptions->getSkin();
3456                 $linkCache =& LinkCache::singleton();
3457
3458                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
3459                         wfProfileIn( $fname.'-check' );
3460                         $dbr =& wfGetDB( DB_SLAVE );
3461                         $page = $dbr->tableName( 'page' );
3462                         $threshold = $wgUser->getOption('stubthreshold');
3463
3464                         # Sort by namespace
3465                         asort( $this->mLinkHolders['namespaces'] );
3466
3467                         # Generate query
3468                         $query = false;
3469                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3470                                 # Make title object
3471                                 $title = $this->mLinkHolders['titles'][$key];
3472
3473                                 # Skip invalid entries.
3474                                 # Result will be ugly, but prevents crash.
3475                                 if ( is_null( $title ) ) {
3476                                         continue;
3477                                 }
3478                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
3479
3480                                 # Check if it's a static known link, e.g. interwiki
3481                                 if ( $title->isAlwaysKnown() ) {
3482                                         $colours[$pdbk] = 1;
3483                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
3484                                         $colours[$pdbk] = 1;
3485                                         $this->mOutput->addLink( $title, $id );
3486                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
3487                                         $colours[$pdbk] = 0;
3488                                 } else {
3489                                         # Not in the link cache, add it to the query
3490                                         if ( !isset( $current ) ) {
3491                                                 $current = $ns;
3492                                                 $query =  "SELECT page_id, page_namespace, page_title";
3493                                                 if ( $threshold > 0 ) {
3494                                                         $query .= ', page_len, page_is_redirect';
3495                                                 }
3496                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
3497                                         } elseif ( $current != $ns ) {
3498                                                 $current = $ns;
3499                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
3500                                         } else {
3501                                                 $query .= ', ';
3502                                         }
3503
3504                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
3505                                 }
3506                         }
3507                         if ( $query ) {
3508                                 $query .= '))';
3509                                 if ( $options & RLH_FOR_UPDATE ) {
3510                                         $query .= ' FOR UPDATE';
3511                                 }
3512
3513                                 $res = $dbr->query( $query, $fname );
3514
3515                                 # Fetch data and form into an associative array
3516                                 # non-existent = broken
3517                                 # 1 = known
3518                                 # 2 = stub
3519                                 while ( $s = $dbr->fetchObject($res) ) {
3520                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
3521                                         $pdbk = $title->getPrefixedDBkey();
3522                                         $linkCache->addGoodLinkObj( $s->page_id, $title );
3523                                         $this->mOutput->addLink( $title, $s->page_id );
3524
3525                                         if ( $threshold >  0 ) {
3526                                                 $size = $s->page_len;
3527                                                 if ( $s->page_is_redirect || $s->page_namespace != 0 || $size >= $threshold ) {
3528                                                         $colours[$pdbk] = 1;
3529                                                 } else {
3530                                                         $colours[$pdbk] = 2;
3531                                                 }
3532                                         } else {
3533                                                 $colours[$pdbk] = 1;
3534                                         }
3535                                 }
3536                         }
3537                         wfProfileOut( $fname.'-check' );
3538
3539                         # Construct search and replace arrays
3540                         wfProfileIn( $fname.'-construct' );
3541                         $wgOutputReplace = array();
3542                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3543                                 $pdbk = $pdbks[$key];
3544                                 $searchkey = "<!--LINK $key-->";
3545                                 $title = $this->mLinkHolders['titles'][$key];
3546                                 if ( empty( $colours[$pdbk] ) ) {
3547                                         $linkCache->addBadLinkObj( $title );
3548                                         $colours[$pdbk] = 0;
3549                                         $this->mOutput->addLink( $title, 0 );
3550                                         $wgOutputReplace[$searchkey] = $sk->makeBrokenLinkObj( $title,
3551                                                                         $this->mLinkHolders['texts'][$key],
3552                                                                         $this->mLinkHolders['queries'][$key] );
3553                                 } elseif ( $colours[$pdbk] == 1 ) {
3554                                         $wgOutputReplace[$searchkey] = $sk->makeKnownLinkObj( $title,
3555                                                                         $this->mLinkHolders['texts'][$key],
3556                                                                         $this->mLinkHolders['queries'][$key] );
3557                                 } elseif ( $colours[$pdbk] == 2 ) {
3558                                         $wgOutputReplace[$searchkey] = $sk->makeStubLinkObj( $title,
3559                                                                         $this->mLinkHolders['texts'][$key],
3560                                                                         $this->mLinkHolders['queries'][$key] );
3561                                 }
3562                         }
3563                         wfProfileOut( $fname.'-construct' );
3564
3565                         # Do the thing
3566                         wfProfileIn( $fname.'-replace' );
3567
3568                         $text = preg_replace_callback(
3569                                 '/(<!--LINK .*?-->)/',
3570                                 "wfOutputReplaceMatches",
3571                                 $text);
3572
3573                         wfProfileOut( $fname.'-replace' );
3574                 }
3575
3576                 # Now process interwiki link holders
3577                 # This is quite a bit simpler than internal links
3578                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
3579                         wfProfileIn( $fname.'-interwiki' );
3580                         # Make interwiki link HTML
3581                         $wgOutputReplace = array();
3582                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
3583                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
3584                                 $wgOutputReplace[$key] = $sk->makeLinkObj( $title, $link );
3585                         }
3586
3587                         $text = preg_replace_callback(
3588                                 '/<!--IWLINK (.*?)-->/',
3589                                 "wfOutputReplaceMatches",
3590                                 $text );
3591                         wfProfileOut( $fname.'-interwiki' );
3592                 }
3593
3594                 wfProfileOut( $fname );
3595                 return $colours;
3596         }
3597
3598         /**
3599          * Replace <!--LINK--> link placeholders with plain text of links
3600          * (not HTML-formatted).
3601          * @param string $text
3602          * @return string
3603          */
3604         function replaceLinkHoldersText( $text ) {
3605                 global $wgUser;
3606                 global $wgOutputReplace;
3607
3608                 $fname = 'Parser::replaceLinkHoldersText';
3609                 wfProfileIn( $fname );
3610
3611                 $text = preg_replace_callback(
3612                         '/<!--(LINK|IWLINK) (.*?)-->/',
3613                         array( &$this, 'replaceLinkHoldersTextCallback' ),
3614                         $text );
3615
3616                 wfProfileOut( $fname );
3617                 return $text;
3618         }
3619
3620         /**
3621          * @param array $matches
3622          * @return string
3623          * @access private
3624          */
3625         function replaceLinkHoldersTextCallback( $matches ) {
3626                 $type = $matches[1];
3627                 $key  = $matches[2];
3628                 if( $type == 'LINK' ) {
3629                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
3630                                 return $this->mLinkHolders['texts'][$key];
3631                         }
3632                 } elseif( $type == 'IWLINK' ) {
3633                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
3634                                 return $this->mInterwikiLinkHolders['texts'][$key];
3635                         }
3636                 }
3637                 return $matches[0];
3638         }
3639
3640         /**
3641          * Renders an image gallery from a text with one line per image.
3642          * text labels may be given by using |-style alternative text. E.g.
3643          *   Image:one.jpg|The number "1"
3644          *   Image:tree.jpg|A tree
3645          * given as text will return the HTML of a gallery with two images,
3646          * labeled 'The number "1"' and
3647          * 'A tree'.
3648          */
3649         function renderImageGallery( $text ) {
3650                 # Setup the parser
3651                 $parserOptions = new ParserOptions;
3652                 $localParser = new Parser();
3653
3654                 $ig = new ImageGallery();
3655                 $ig->setShowBytes( false );
3656                 $ig->setShowFilename( false );
3657                 $lines = explode( "\n", $text );
3658
3659                 foreach ( $lines as $line ) {
3660                         # match lines like these:
3661                         # Image:someimage.jpg|This is some image
3662                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
3663                         # Skip empty lines
3664                         if ( count( $matches ) == 0 ) {
3665                                 continue;
3666                         }
3667                         $nt =& Title::newFromText( $matches[1] );
3668                         if( is_null( $nt ) ) {
3669                                 # Bogus title. Ignore these so we don't bomb out later.
3670                                 continue;
3671                         }
3672                         if ( isset( $matches[3] ) ) {
3673                                 $label = $matches[3];
3674                         } else {
3675                                 $label = '';
3676                         }
3677
3678                         $pout = $localParser->parse( $label , $this->mTitle, $parserOptions );
3679                         $html = $pout->getText();
3680
3681                         $ig->add( new Image( $nt ), $html );
3682                         $this->mOutput->addImage( $nt->getDBkey() );
3683                 }
3684                 return $ig->toHTML();
3685         }
3686
3687         /**
3688          * Parse image options text and use it to make an image
3689          */
3690         function makeImage( &$nt, $options ) {
3691                 global $wgContLang, $wgUseImageResize, $wgUser;
3692
3693                 $align = '';
3694
3695                 # Check if the options text is of the form "options|alt text"
3696                 # Options are:
3697                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
3698                 #  * left               no resizing, just left align. label is used for alt= only
3699                 #  * right              same, but right aligned
3700                 #  * none               same, but not aligned
3701                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
3702                 #  * center             center the image
3703                 #  * framed             Keep original image size, no magnify-button.
3704
3705                 $part = explode( '|', $options);
3706
3707                 $mwThumb  =& MagicWord::get( MAG_IMG_THUMBNAIL );
3708                 $mwManualThumb =& MagicWord::get( MAG_IMG_MANUALTHUMB );
3709                 $mwLeft   =& MagicWord::get( MAG_IMG_LEFT );
3710                 $mwRight  =& MagicWord::get( MAG_IMG_RIGHT );
3711                 $mwNone   =& MagicWord::get( MAG_IMG_NONE );
3712                 $mwWidth  =& MagicWord::get( MAG_IMG_WIDTH );
3713                 $mwCenter =& MagicWord::get( MAG_IMG_CENTER );
3714                 $mwFramed =& MagicWord::get( MAG_IMG_FRAMED );
3715                 $caption = '';
3716
3717                 $width = $height = $framed = $thumb = false;
3718                 $manual_thumb = '' ;
3719
3720                 foreach( $part as $key => $val ) {
3721                         if ( $wgUseImageResize && ! is_null( $mwThumb->matchVariableStartToEnd($val) ) ) {
3722                                 $thumb=true;
3723                         } elseif ( ! is_null( $match = $mwManualThumb->matchVariableStartToEnd($val) ) ) {
3724                                 # use manually specified thumbnail
3725                                 $thumb=true;
3726                                 $manual_thumb = $match;
3727                         } elseif ( ! is_null( $mwRight->matchVariableStartToEnd($val) ) ) {
3728                                 # remember to set an alignment, don't render immediately
3729                                 $align = 'right';
3730                         } elseif ( ! is_null( $mwLeft->matchVariableStartToEnd($val) ) ) {
3731                                 # remember to set an alignment, don't render immediately
3732                                 $align = 'left';
3733                         } elseif ( ! is_null( $mwCenter->matchVariableStartToEnd($val) ) ) {
3734                                 # remember to set an alignment, don't render immediately
3735                                 $align = 'center';
3736                         } elseif ( ! is_null( $mwNone->matchVariableStartToEnd($val) ) ) {
3737                                 # remember to set an alignment, don't render immediately
3738                                 $align = 'none';
3739                         } elseif ( $wgUseImageResize && ! is_null( $match = $mwWidth->matchVariableStartToEnd($val) ) ) {
3740                                 wfDebug( "MAG_IMG_WIDTH match: $match\n" );
3741                                 # $match is the image width in pixels
3742                                 if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $match, $m ) ) {
3743                                         $width = intval( $m[1] );
3744                                         $height = intval( $m[2] );
3745                                 } else {
3746                                         $width = intval($match);
3747                                 }
3748                         } elseif ( ! is_null( $mwFramed->matchVariableStartToEnd($val) ) ) {
3749                                 $framed=true;
3750                         } else {
3751                                 $caption = $val;
3752                         }
3753                 }
3754                 # Strip bad stuff out of the alt text
3755                 $alt = $this->replaceLinkHoldersText( $caption );
3756                 $alt = Sanitizer::stripAllTags( $alt );
3757
3758                 # Linker does the rest
3759                 $sk =& $this->mOptions->getSkin();
3760                 return $sk->makeImageLinkObj( $nt, $caption, $alt, $align, $width, $height, $framed, $thumb, $manual_thumb );
3761         }
3762
3763         /**
3764          * Set a flag in the output object indicating that the content is dynamic and
3765          * shouldn't be cached.
3766          */
3767         function disableCache() {
3768                 $this->mOutput->mCacheTime = -1;
3769         }
3770
3771         /**#@+
3772          * Callback from the Sanitizer for expanding items found in HTML attribute
3773          * values, so they can be safely tested and escaped.
3774          * @param string $text
3775          * @param array $args
3776          * @return string
3777          * @access private
3778          */
3779         function attributeStripCallback( &$text, $args ) {
3780                 $text = $this->replaceVariables( $text, $args );
3781                 $text = $this->unstripForHTML( $text );
3782                 return $text;
3783         }
3784
3785         function unstripForHTML( $text ) {
3786                 $text = $this->unstrip( $text, $this->mStripState );
3787                 $text = $this->unstripNoWiki( $text, $this->mStripState );
3788                 return $text;
3789         }
3790         /**#@-*/
3791
3792         /**#@+
3793          * Accessor/mutator
3794          */
3795         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
3796         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
3797         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
3798         /**#@-*/
3799
3800         /**#@+
3801          * Accessor
3802          */
3803         function getTags() { return array_keys( $this->mTagHooks ); }
3804         /**#@-*/
3805 }
3806
3807 /**
3808  * @todo document
3809  * @package MediaWiki
3810  */
3811 class ParserOutput
3812 {
3813         var $mText,             # The output text
3814                 $mLanguageLinks,    # List of the full text of language links, in the order they appear
3815                 $mCategories,       # Map of category names to sort keys
3816                 $mContainsOldMagic, # Boolean variable indicating if the input contained variables like {{CURRENTDAY}}
3817                 $mCacheTime,        # Timestamp on this article, or -1 for uncacheable. Used in ParserCache.
3818                 $mVersion,          # Compatibility check
3819                 $mTitleText,        # title text of the chosen language variant
3820                 $mLinks,            # 2-D map of NS/DBK to ID for the links in the document. ID=zero for broken.
3821                 $mTemplates,        # 2-D map of NS/DBK to ID for the template references. ID=zero for broken.
3822                 $mImages,           # DB keys of the images used, in the array key only
3823                 $mExternalLinks;    # External link URLs, in the key only
3824
3825         function ParserOutput( $text = '', $languageLinks = array(), $categoryLinks = array(),
3826                 $containsOldMagic = false, $titletext = '' )
3827         {
3828                 $this->mText = $text;
3829                 $this->mLanguageLinks = $languageLinks;
3830                 $this->mCategories = $categoryLinks;
3831                 $this->mContainsOldMagic = $containsOldMagic;
3832                 $this->mCacheTime = '';
3833                 $this->mVersion = MW_PARSER_VERSION;
3834                 $this->mTitleText = $titletext;
3835                 $this->mLinks = array();
3836                 $this->mTemplates = array();
3837                 $this->mImages = array();
3838                 $this->mExternalLinks = array();
3839         }
3840
3841         function getText()                   { return $this->mText; }
3842         function getLanguageLinks()          { return $this->mLanguageLinks; }
3843         function getCategoryLinks()          { return array_keys( $this->mCategories ); }
3844         function &getCategories()            { return $this->mCategories; }
3845         function getCacheTime()              { return $this->mCacheTime; }
3846         function getTitleText()              { return $this->mTitleText; }
3847         function &getLinks()                 { return $this->mLinks; }
3848         function &getTemplates()             { return $this->mTemplates; }
3849         function &getImages()                { return $this->mImages; }
3850         function &getExternalLinks()         { return $this->mExternalLinks; }
3851
3852         function containsOldMagic()          { return $this->mContainsOldMagic; }
3853         function setText( $text )            { return wfSetVar( $this->mText, $text ); }
3854         function setLanguageLinks( $ll )     { return wfSetVar( $this->mLanguageLinks, $ll ); }
3855         function setCategoryLinks( $cl )     { return wfSetVar( $this->mCategories, $cl ); }
3856         function setContainsOldMagic( $com ) { return wfSetVar( $this->mContainsOldMagic, $com ); }
3857         function setCacheTime( $t )          { return wfSetVar( $this->mCacheTime, $t ); }
3858         function setTitleText( $t )          { return wfSetVar ($this->mTitleText, $t); }
3859
3860         function addCategory( $c, $sort )    { $this->mCategories[$c] = $sort; }
3861         function addImage( $name )           { $this->mImages[$name] = 1; }
3862         function addLanguageLink( $t )       { $this->mLanguageLinks[] = $t; }
3863         function addExternalLink( $url )     { $this->mExternalLinks[$url] = 1; }
3864
3865         function addLink( $title, $id ) {
3866                 $ns = $title->getNamespace();
3867                 $dbk = $title->getDBkey();
3868                 if ( !isset( $this->mLinks[$ns] ) ) {
3869                         $this->mLinks[$ns] = array();
3870                 }
3871                 $this->mLinks[$ns][$dbk] = $id;
3872         }
3873
3874         function addTemplate( $title, $id ) {
3875                 $ns = $title->getNamespace();
3876                 $dbk = $title->getDBkey();
3877                 if ( !isset( $this->mTemplates[$ns] ) ) {
3878                         $this->mTemplates[$ns] = array();
3879                 }
3880                 $this->mTemplates[$ns][$dbk] = $id;
3881         }
3882
3883         /**
3884          * @deprecated
3885          */
3886         /*
3887         function merge( $other ) {
3888                 $this->mLanguageLinks = array_merge( $this->mLanguageLinks, $other->mLanguageLinks );
3889                 $this->mCategories = array_merge( $this->mCategories, $this->mLanguageLinks );
3890                 $this->mContainsOldMagic = $this->mContainsOldMagic || $other->mContainsOldMagic;
3891         }*/
3892
3893         /**
3894          * Return true if this cached output object predates the global or
3895          * per-article cache invalidation timestamps, or if it comes from
3896          * an incompatible older version.
3897          *
3898          * @param string $touched the affected article's last touched timestamp
3899          * @return bool
3900          * @access public
3901          */
3902         function expired( $touched ) {
3903                 global $wgCacheEpoch;
3904                 return $this->getCacheTime() == -1 || // parser says it's uncacheable
3905                        $this->getCacheTime() < $touched ||
3906                        $this->getCacheTime() <= $wgCacheEpoch ||
3907                        !isset( $this->mVersion ) ||
3908                        version_compare( $this->mVersion, MW_PARSER_VERSION, "lt" );
3909         }
3910 }
3911
3912 /**
3913  * Set options of the Parser
3914  * @todo document
3915  * @package MediaWiki
3916  */
3917 class ParserOptions
3918 {
3919         # All variables are private
3920         var $mUseTeX;                    # Use texvc to expand <math> tags
3921         var $mUseDynamicDates;           # Use DateFormatter to format dates
3922         var $mInterwikiMagic;            # Interlanguage links are removed and returned in an array
3923         var $mAllowExternalImages;       # Allow external images inline
3924         var $mAllowExternalImagesFrom;   # If not, any exception?
3925         var $mSkin;                      # Reference to the preferred skin
3926         var $mDateFormat;                # Date format index
3927         var $mEditSection;               # Create "edit section" links
3928         var $mNumberHeadings;            # Automatically number headings
3929         var $mAllowSpecialInclusion;     # Allow inclusion of special pages
3930         var $mTidy;                      # Ask for tidy cleanup
3931
3932         function getUseTeX()                        { return $this->mUseTeX; }
3933         function getUseDynamicDates()               { return $this->mUseDynamicDates; }
3934         function getInterwikiMagic()                { return $this->mInterwikiMagic; }
3935         function getAllowExternalImages()           { return $this->mAllowExternalImages; }
3936         function getAllowExternalImagesFrom()       { return $this->mAllowExternalImagesFrom; }
3937         function &getSkin()                         { return $this->mSkin; }
3938         function getDateFormat()                    { return $this->mDateFormat; }
3939         function getEditSection()                   { return $this->mEditSection; }
3940         function getNumberHeadings()                { return $this->mNumberHeadings; }
3941         function getAllowSpecialInclusion()         { return $this->mAllowSpecialInclusion; }
3942         function getTidy()                          { return $this->mTidy; }
3943
3944         function setUseTeX( $x )                    { return wfSetVar( $this->mUseTeX, $x ); }
3945         function setUseDynamicDates( $x )           { return wfSetVar( $this->mUseDynamicDates, $x ); }
3946         function setInterwikiMagic( $x )            { return wfSetVar( $this->mInterwikiMagic, $x ); }
3947         function setAllowExternalImages( $x )       { return wfSetVar( $this->mAllowExternalImages, $x ); }
3948         function setAllowExternalImagesFrom( $x )   { return wfSetVar( $this->mAllowExternalImagesFrom, $x ); }
3949         function setDateFormat( $x )                { return wfSetVar( $this->mDateFormat, $x ); }
3950         function setEditSection( $x )               { return wfSetVar( $this->mEditSection, $x ); }
3951         function setNumberHeadings( $x )            { return wfSetVar( $this->mNumberHeadings, $x ); }
3952         function setAllowSpecialInclusion( $x )     { return wfSetVar( $this->mAllowSpecialInclusion, $x ); }
3953         function setTidy( $x )                      { return wfSetVar( $this->mTidy, $x); }
3954         function setSkin( &$x ) { $this->mSkin =& $x; }
3955
3956         function ParserOptions() {
3957                 global $wgUser;
3958                 $this->initialiseFromUser( $wgUser );
3959         }
3960
3961         /**
3962          * Get parser options
3963          * @static
3964          */
3965         function newFromUser( &$user ) {
3966                 $popts = new ParserOptions;
3967                 $popts->initialiseFromUser( $user );
3968                 return $popts;
3969         }
3970
3971         /** Get user options */
3972         function initialiseFromUser( &$userInput ) {
3973                 global $wgUseTeX, $wgUseDynamicDates, $wgInterwikiMagic, $wgAllowExternalImages,
3974                        $wgAllowExternalImagesFrom, $wgAllowSpecialInclusion;
3975                 $fname = 'ParserOptions::initialiseFromUser';
3976                 wfProfileIn( $fname );
3977                 if ( !$userInput ) {
3978                         $user = new User;
3979                         $user->setLoaded( true );
3980                 } else {
3981                         $user =& $userInput;
3982                 }
3983
3984                 $this->mUseTeX = $wgUseTeX;
3985                 $this->mUseDynamicDates = $wgUseDynamicDates;
3986                 $this->mInterwikiMagic = $wgInterwikiMagic;
3987                 $this->mAllowExternalImages = $wgAllowExternalImages;
3988                 $this->mAllowExternalImagesFrom = $wgAllowExternalImagesFrom;
3989                 wfProfileIn( $fname.'-skin' );
3990                 $this->mSkin =& $user->getSkin();
3991                 wfProfileOut( $fname.'-skin' );
3992                 $this->mDateFormat = $user->getOption( 'date' );
3993                 $this->mEditSection = true;
3994                 $this->mNumberHeadings = $user->getOption( 'numberheadings' );
3995                 $this->mAllowSpecialInclusion = $wgAllowSpecialInclusion;
3996                 $this->mTidy = false;
3997                 wfProfileOut( $fname );
3998         }
3999 }
4000
4001 /**
4002  * Callback function used by Parser::replaceLinkHolders()
4003  * to substitute link placeholders.
4004  */
4005 function &wfOutputReplaceMatches( $matches ) {
4006         global $wgOutputReplace;
4007         return $wgOutputReplace[$matches[1]];
4008 }
4009
4010 /**
4011  * Return the total number of articles
4012  */
4013 function wfNumberOfArticles() {
4014         global $wgNumberOfArticles;
4015
4016         wfLoadSiteStats();
4017         return $wgNumberOfArticles;
4018 }
4019
4020 /**
4021  * Return the number of files
4022  */
4023 function wfNumberOfFiles() {
4024         $fname = 'Parser::wfNumberOfFiles';
4025
4026         wfProfileIn( $fname );
4027         $dbr =& wfGetDB( DB_SLAVE );
4028         $res = $dbr->selectField('image', 'COUNT(*)', array(), $fname );
4029         wfProfileOut( $fname );
4030
4031         return $res;
4032 }
4033
4034 /**
4035  * Get various statistics from the database
4036  * @access private
4037  */
4038 function wfLoadSiteStats() {
4039         global $wgNumberOfArticles, $wgTotalViews, $wgTotalEdits;
4040         $fname = 'wfLoadSiteStats';
4041
4042         if ( -1 != $wgNumberOfArticles ) return;
4043         $dbr =& wfGetDB( DB_SLAVE );
4044         $s = $dbr->selectRow( 'site_stats',
4045                 array( 'ss_total_views', 'ss_total_edits', 'ss_good_articles' ),
4046                 array( 'ss_row_id' => 1 ), $fname
4047         );
4048
4049         if ( $s === false ) {
4050                 return;
4051         } else {
4052                 $wgTotalViews = $s->ss_total_views;
4053                 $wgTotalEdits = $s->ss_total_edits;
4054                 $wgNumberOfArticles = $s->ss_good_articles;
4055         }
4056 }
4057
4058 /**
4059  * Escape html tags
4060  * Basically replacing " > and < with HTML entities ( &quot;, &gt;, &lt;)
4061  *
4062  * @param string $in Text that might contain HTML tags
4063  * @return string Escaped string
4064  */
4065 function wfEscapeHTMLTagsOnly( $in ) {
4066         return str_replace(
4067                 array( '"', '>', '<' ),
4068                 array( '&quot;', '&gt;', '&lt;' ),
4069                 $in );
4070 }
4071
4072 ?>