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, $wgContLang
  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 parserConvert() call should not be changed. it
 240                 # assumes that the links are all replaced and the only thing left
 241                 # is the <nowiki> mark.
 242                 # Side-effects: this calls $this->mOutput->setTitleText()
 243                 $text = $wgContLang->parserConvert( $text, $this );
 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         function &getTitle() { return $this->mTitle; }
 274         function getOptions() { return $this->mOptions; }
 275
 276         /**
 277          * Replaces all occurrences of <$tag>content</$tag> in the text
 278          * with a random marker and returns the new text. the output parameter
 279          * $content will be an associative array filled with data on the form
 280          * $unique_marker => content.
 281          *
 282          * If $content is already set, the additional entries will be appended
 283          * If $tag is set to STRIP_COMMENTS, the function will extract
 284          * <!-- HTML comments -->
 285          *
 286          * @access private
 287          * @static
 288          */
 289         function extractTagsAndParams($tag, $text, &$content, &$tags, &$params, $uniq_prefix = ''){
 290                 $rnd = $uniq_prefix . '-' . $tag . Parser::getRandomString();
 291                 if ( !$content ) {
 292                         $content = array( );
 293                 }
 294                 $n = 1;
 295                 $stripped = '';
 296
 297                 if ( !$tags ) {
 298                         $tags = array( );
 299                 }
 300
 301                 if ( !$params ) {
 302                         $params = array( );
 303                 }
 304
 305                 if( $tag == STRIP_COMMENTS ) {
 306                         $start = '/<!--()()/';
 307                         $end   = '/-->/';
 308                 } else {
 309                         $start = "/<$tag(\\s+[^\\/>]*|\\s*)(\\/?)>/i";
 310                         $end   = "/<\\/$tag\\s*>/i";
 311                 }
 312
 313                 while ( '' != $text ) {
 314                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 315                         $stripped .= $p[0];
 316                         if( count( $p ) < 4 ) {
 317                                 break;
 318                         }
 319                         $attributes = $p[1];
 320                         $empty      = $p[2];
 321                         $inside     = $p[3];
 322
 323                         $marker = $rnd . sprintf('%08X', $n++);
 324                         $stripped .= $marker;
 325
 326                         $tags[$marker] = "<$tag$attributes$empty>";
 327                         $params[$marker] = Sanitizer::decodeTagAttributes( $attributes );
 328
 329                         if ( $empty === '/' ) {
 330                                 // Empty element tag, <tag />
 331                                 $content[$marker] = null;
 332                                 $text = $inside;
 333                         } else {
 334                                 $q = preg_split( $end, $inside, 2 );
 335                                 $content[$marker] = $q[0];
 336                                 if( count( $q ) < 2 ) {
 337                                         # No end tag -- let it run out to the end of the text.
 338                                         break;
 339                                 } else {
 340                                         $text = $q[1];
 341                                 }
 342                         }
 343                 }
 344                 return $stripped;
 345         }
 346
 347         /**
 348          * Wrapper function for extractTagsAndParams
 349          * for cases where $tags and $params isn't needed
 350          * i.e. where tags will never have params, like <nowiki>
 351          *
 352          * @access private
 353          * @static
 354          */
 355         function extractTags( $tag, $text, &$content, $uniq_prefix = '' ) {
 356                 $dummy_tags = array();
 357                 $dummy_params = array();
 358
 359                 return Parser::extractTagsAndParams( $tag, $text, $content,
 360                         $dummy_tags, $dummy_params, $uniq_prefix );
 361         }
 362
 363         /**
 364          * Strips and renders nowiki, pre, math, hiero
 365          * If $render is set, performs necessary rendering operations on plugins
 366          * Returns the text, and fills an array with data needed in unstrip()
 367          * If the $state is already a valid strip state, it adds to the state
 368          *
 369          * @param bool $stripcomments when set, HTML comments <!-- like this -->
 370          *  will be stripped in addition to other tags. This is important
 371          *  for section editing, where these comments cause confusion when
 372          *  counting the sections in the wikisource
 373          *
 374          * @access private
 375          */
 376         function strip( $text, &$state, $stripcomments = false ) {
 377                 $render = ($this->mOutputType == OT_HTML);
 378                 $html_content = array();
 379                 $nowiki_content = array();
 380                 $math_content = array();
 381                 $pre_content = array();
 382                 $comment_content = array();
 383                 $ext_content = array();
 384                 $ext_tags = array();
 385                 $ext_params = array();
 386                 $gallery_content = array();
 387
 388                 # Replace any instances of the placeholders
 389                 $uniq_prefix = $this->mUniqPrefix;
 390                 #$text = str_replace( $uniq_prefix, wfHtmlEscapeFirst( $uniq_prefix ), $text );
 391
 392                 # html
 393                 global $wgRawHtml;
 394                 if( $wgRawHtml ) {
 395                         $text = Parser::extractTags('html', $text, $html_content, $uniq_prefix);
 396                         foreach( $html_content as $marker => $content ) {
 397                                 if ($render ) {
 398                                         # Raw and unchecked for validity.
 399                                         $html_content[$marker] = $content;
 400                                 } else {
 401                                         $html_content[$marker] = '<html>'.$content.'</html>';
 402                                 }
 403                         }
 404                 }
 405
 406                 # nowiki
 407                 $text = Parser::extractTags('nowiki', $text, $nowiki_content, $uniq_prefix);
 408                 foreach( $nowiki_content as $marker => $content ) {
 409                         if( $render ){
 410                                 $nowiki_content[$marker] = wfEscapeHTMLTagsOnly( $content );
 411                         } else {
 412                                 $nowiki_content[$marker] = '<nowiki>'.$content.'</nowiki>';
 413                         }
 414                 }
 415
 416                 # math
 417                 if( $this->mOptions->getUseTeX() ) {
 418                         $text = Parser::extractTags('math', $text, $math_content, $uniq_prefix);
 419                         foreach( $math_content as $marker => $content ){
 420                                 if( $render ) {
 421                                         $math_content[$marker] = renderMath( $content );
 422                                 } else {
 423                                         $math_content[$marker] = '<math>'.$content.'</math>';
 424                                 }
 425                         }
 426                 }
 427
 428                 # pre
 429                 $text = Parser::extractTags('pre', $text, $pre_content, $uniq_prefix);
 430                 foreach( $pre_content as $marker => $content ){
 431                         if( $render ){
 432                                 $pre_content[$marker] = '<pre>' . wfEscapeHTMLTagsOnly( $content ) . '</pre>';
 433                         } else {
 434                                 $pre_content[$marker] = '<pre>'.$content.'</pre>';
 435                         }
 436                 }
 437
 438                 # gallery
 439                 $text = Parser::extractTags('gallery', $text, $gallery_content, $uniq_prefix);
 440                 foreach( $gallery_content as $marker => $content ) {
 441                         require_once( 'ImageGallery.php' );
 442                         if ( $render ) {
 443                                 $gallery_content[$marker] = $this->renderImageGallery( $content );
 444                         } else {
 445                                 $gallery_content[$marker] = '<gallery>'.$content.'</gallery>';
 446                         }
 447                 }
 448
 449                 # Comments
 450                 if($stripcomments) {
 451                         $text = Parser::extractTags(STRIP_COMMENTS, $text, $comment_content, $uniq_prefix);
 452                         foreach( $comment_content as $marker => $content ){
 453                                 $comment_content[$marker] = '<!--'.$content.'-->';
 454                         }
 455                 }
 456
 457                 # Extensions
 458                 foreach ( $this->mTagHooks as $tag => $callback ) {
 459                         $ext_content[$tag] = array();
 460                         $text = Parser::extractTagsAndParams( $tag, $text, $ext_content[$tag],
 461                                 $ext_tags[$tag], $ext_params[$tag], $uniq_prefix );
 462                         foreach( $ext_content[$tag] as $marker => $content ) {
 463                                 $full_tag = $ext_tags[$tag][$marker];
 464                                 $params = $ext_params[$tag][$marker];
 465                                 if ( $render )
 466                                         $ext_content[$tag][$marker] = call_user_func_array( $callback, array( $content, $params, &$this ) );
 467                                 else {
 468                                         if ( is_null( $content ) ) {
 469                                                 // Empty element tag
 470                                                 $ext_content[$tag][$marker] = $full_tag;
 471                                         } else {
 472                                                 $ext_content[$tag][$marker] = "$full_tag$content</$tag>";
 473                                         }
 474                                 }
 475                         }
 476                 }
 477
 478                 # Merge state with the pre-existing state, if there is one
 479                 if ( $state ) {
 480                         $state['html'] = $state['html'] + $html_content;
 481                         $state['nowiki'] = $state['nowiki'] + $nowiki_content;
 482                         $state['math'] = $state['math'] + $math_content;
 483                         $state['pre'] = $state['pre'] + $pre_content;
 484                         $state['gallery'] = $state['gallery'] + $gallery_content;
 485                         $state['comment'] = $state['comment'] + $comment_content;
 486
 487                         foreach( $ext_content as $tag => $array ) {
 488                                 if ( array_key_exists( $tag, $state ) ) {
 489                                         $state[$tag] = $state[$tag] + $array;
 490                                 }
 491                         }
 492                 } else {
 493                         $state = array(
 494                           'html' => $html_content,
 495                           'nowiki' => $nowiki_content,
 496                           'math' => $math_content,
 497                           'pre' => $pre_content,
 498                           'gallery' => $gallery_content,
 499                           'comment' => $comment_content,
 500                         ) + $ext_content;
 501                 }
 502                 return $text;
 503         }
 504
 505         /**
 506          * restores pre, math, and hiero removed by strip()
 507          *
 508          * always call unstripNoWiki() after this one
 509          * @access private
 510          */
 511         function unstrip( $text, &$state ) {
 512                 if ( !is_array( $state ) ) {
 513                         return $text;
 514                 }
 515
 516                 # Must expand in reverse order, otherwise nested tags will be corrupted
 517                 foreach( array_reverse( $state, true ) as $tag => $contentDict ) {
 518                         if( $tag != 'nowiki' && $tag != 'html' ) {
 519                                 foreach( array_reverse( $contentDict, true ) as $uniq => $content ) {
 520                                         $text = str_replace( $uniq, $content, $text );
 521                                 }
 522                         }
 523                 }
 524
 525                 return $text;
 526         }
 527
 528         /**
 529          * always call this after unstrip() to preserve the order
 530          *
 531          * @access private
 532          */
 533         function unstripNoWiki( $text, &$state ) {
 534                 if ( !is_array( $state ) ) {
 535                         return $text;
 536                 }
 537
 538                 # Must expand in reverse order, otherwise nested tags will be corrupted
 539                 for ( $content = end($state['nowiki']); $content !== false; $content = prev( $state['nowiki'] ) ) {
 540                         $text = str_replace( key( $state['nowiki'] ), $content, $text );
 541                 }
 542
 543                 global $wgRawHtml;
 544                 if ($wgRawHtml) {
 545                         for ( $content = end($state['html']); $content !== false; $content = prev( $state['html'] ) ) {
 546                                 $text = str_replace( key( $state['html'] ), $content, $text );
 547                         }
 548                 }
 549
 550                 return $text;
 551         }
 552
 553         /**
 554          * Add an item to the strip state
 555          * Returns the unique tag which must be inserted into the stripped text
 556          * The tag will be replaced with the original text in unstrip()
 557          *
 558          * @access private
 559          */
 560         function insertStripItem( $text, &$state ) {
 561                 $rnd = $this->mUniqPrefix . '-item' . Parser::getRandomString();
 562                 if ( !$state ) {
 563                         $state = array(
 564                           'html' => array(),
 565                           'nowiki' => array(),
 566                           'math' => array(),
 567                           'pre' => array(),
 568                           'comment' => array(),
 569                           'gallery' => array(),
 570                         );
 571                 }
 572                 $state['item'][$rnd] = $text;
 573                 return $rnd;
 574         }
 575
 576         /**
 577          * Interface with html tidy, used if $wgUseTidy = true.
 578          * If tidy isn't able to correct the markup, the original will be
 579          * returned in all its glory with a warning comment appended.
 580          *
 581          * Either the external tidy program or the in-process tidy extension
 582          * will be used depending on availability. Override the default
 583          * $wgTidyInternal setting to disable the internal if it's not working.
 584          *
 585          * @param string $text Hideous HTML input
 586          * @return string Corrected HTML output
 587          * @access public
 588          * @static
 589          */
 590         function tidy( $text ) {
 591                 global $wgTidyInternal;
 592                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 593 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 594 '<head><title>test</title></head><body>'.$text.'</body></html>';
 595                 if( $wgTidyInternal ) {
 596                         $correctedtext = Parser::internalTidy( $wrappedtext );
 597                 } else {
 598                         $correctedtext = Parser::externalTidy( $wrappedtext );
 599                 }
 600                 if( is_null( $correctedtext ) ) {
 601                         wfDebug( "Tidy error detected!\n" );
 602                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 603                 }
 604                 return $correctedtext;
 605         }
 606
 607         /**
 608          * Spawn an external HTML tidy process and get corrected markup back from it.
 609          *
 610          * @access private
 611          * @static
 612          */
 613         function externalTidy( $text ) {
 614                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 615                 $fname = 'Parser::externalTidy';
 616                 wfProfileIn( $fname );
 617
 618                 $cleansource = '';
 619                 $opts = ' -utf8';
 620
 621                 $descriptorspec = array(
 622                         0 => array('pipe', 'r'),
 623                         1 => array('pipe', 'w'),
 624                         2 => array('file', '/dev/null', 'a')
 625                 );
 626                 $pipes = array();
 627                 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 628                 if (is_resource($process)) {
 629                         // Theoretically, this style of communication could cause a deadlock
 630                         // here. If the stdout buffer fills up, then writes to stdin could
 631                         // block. This doesn't appear to happen with tidy, because tidy only
 632                         // writes to stdout after it's finished reading from stdin. Search
 633                         // for tidyParseStdin and tidySaveStdout in console/tidy.c
 634                         fwrite($pipes[0], $text);
 635                         fclose($pipes[0]);
 636                         while (!feof($pipes[1])) {
 637                                 $cleansource .= fgets($pipes[1], 1024);
 638                         }
 639                         fclose($pipes[1]);
 640                         proc_close($process);
 641                 }
 642
 643                 wfProfileOut( $fname );
 644
 645                 if( $cleansource == '' && $text != '') {
 646                         // Some kind of error happened, so we couldn't get the corrected text.
 647                         // Just give up; we'll use the source text and append a warning.
 648                         return null;
 649                 } else {
 650                         return $cleansource;
 651                 }
 652         }
 653
 654         /**
 655          * Use the HTML tidy PECL extension to use the tidy library in-process,
 656          * saving the overhead of spawning a new process. Currently written to
 657          * the PHP 4.3.x version of the extension, may not work on PHP 5.
 658          *
 659          * 'pear install tidy' should be able to compile the extension module.
 660          *
 661          * @access private
 662          * @static
 663          */
 664         function internalTidy( $text ) {
 665                 global $wgTidyConf;
 666                 $fname = 'Parser::internalTidy';
 667                 wfProfileIn( $fname );
 668
 669                 tidy_load_config( $wgTidyConf );
 670                 tidy_set_encoding( 'utf8' );
 671                 tidy_parse_string( $text );
 672                 tidy_clean_repair();
 673                 if( tidy_get_status() == 2 ) {
 674                         // 2 is magic number for fatal error
 675                         // http://www.php.net/manual/en/function.tidy-get-status.php
 676                         $cleansource = null;
 677                 } else {
 678                         $cleansource = tidy_get_output();
 679                 }
 680                 wfProfileOut( $fname );
 681                 return $cleansource;
 682         }
 683
 684         /**
 685          * parse the wiki syntax used to render tables
 686          *
 687          * @access private
 688          */
 689         function doTableStuff ( $t ) {
 690                 $fname = 'Parser::doTableStuff';
 691                 wfProfileIn( $fname );
 692
 693                 $t = explode ( "\n" , $t ) ;
 694                 $td = array () ; # Is currently a td tag open?
 695                 $ltd = array () ; # Was it TD or TH?
 696                 $tr = array () ; # Is currently a tr tag open?
 697                 $ltr = array () ; # tr attributes
 698                 $has_opened_tr = array(); # Did this table open a <tr> element?
 699                 $indent_level = 0; # indent level of the table
 700                 foreach ( $t AS $k => $x )
 701                 {
 702                         $x = trim ( $x ) ;
 703                         $fc = substr ( $x , 0 , 1 ) ;
 704                         if ( preg_match( '/^(:*)\{\|(.*)$/', $x, $matches ) ) {
 705                                 $indent_level = strlen( $matches[1] );
 706
 707                                 $attributes = $this->unstripForHTML( $matches[2] );
 708
 709                                 $t[$k] = str_repeat( '<dl><dd>', $indent_level ) .
 710                                         '<table' . Sanitizer::fixTagAttributes ( $attributes, 'table' ) . '>' ;
 711                                 array_push ( $td , false ) ;
 712                                 array_push ( $ltd , '' ) ;
 713                                 array_push ( $tr , false ) ;
 714                                 array_push ( $ltr , '' ) ;
 715                                 array_push ( $has_opened_tr, false );
 716                         }
 717                         else if ( count ( $td ) == 0 ) { } # Don't do any of the following
 718                         else if ( '|}' == substr ( $x , 0 , 2 ) ) {
 719                                 $z = "</table>" . substr ( $x , 2);
 720                                 $l = array_pop ( $ltd ) ;
 721                                 if ( !array_pop ( $has_opened_tr ) ) $z = "<tr><td></td></tr>" . $z ;
 722                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 723                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 724                                 array_pop ( $ltr ) ;
 725                                 $t[$k] = $z . str_repeat( '</dd></dl>', $indent_level );
 726                         }
 727                         else if ( '|-' == substr ( $x , 0 , 2 ) ) { # Allows for |---------------
 728                                 $x = substr ( $x , 1 ) ;
 729                                 while ( $x != '' && substr ( $x , 0 , 1 ) == '-' ) $x = substr ( $x , 1 ) ;
 730                                 $z = '' ;
 731                                 $l = array_pop ( $ltd ) ;
 732                                 array_pop ( $has_opened_tr );
 733                                 array_push ( $has_opened_tr , true ) ;
 734                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 735                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 736                                 array_pop ( $ltr ) ;
 737                                 $t[$k] = $z ;
 738                                 array_push ( $tr , false ) ;
 739                                 array_push ( $td , false ) ;
 740                                 array_push ( $ltd , '' ) ;
 741                                 $attributes = $this->unstripForHTML( $x );
 742                                 array_push ( $ltr , Sanitizer::fixTagAttributes ( $attributes, 'tr' ) ) ;
 743                         }
 744                         else if ( '|' == $fc || '!' == $fc || '|+' == substr ( $x , 0 , 2 ) ) { # Caption
 745                                 # $x is a table row
 746                                 if ( '|+' == substr ( $x , 0 , 2 ) ) {
 747                                         $fc = '+' ;
 748                                         $x = substr ( $x , 1 ) ;
 749                                 }
 750                                 $after = substr ( $x , 1 ) ;
 751                                 if ( $fc == '!' ) $after = str_replace ( '!!' , '||' , $after ) ;
 752                                 $after = explode ( '||' , $after ) ;
 753                                 $t[$k] = '' ;
 754
 755                                 # Loop through each table cell
 756                                 foreach ( $after AS $theline )
 757                                 {
 758                                         $z = '' ;
 759                                         if ( $fc != '+' )
 760                                         {
 761                                                 $tra = array_pop ( $ltr ) ;
 762                                                 if ( !array_pop ( $tr ) ) $z = '<tr'.$tra.">\n" ;
 763                                                 array_push ( $tr , true ) ;
 764                                                 array_push ( $ltr , '' ) ;
 765                                                 array_pop ( $has_opened_tr );
 766                                                 array_push ( $has_opened_tr , true ) ;
 767                                         }
 768
 769                                         $l = array_pop ( $ltd ) ;
 770                                         if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 771                                         if ( $fc == '|' ) $l = 'td' ;
 772                                         else if ( $fc == '!' ) $l = 'th' ;
 773                                         else if ( $fc == '+' ) $l = 'caption' ;
 774                                         else $l = '' ;
 775                                         array_push ( $ltd , $l ) ;
 776
 777                                         # Cell parameters
 778                                         $y = explode ( '|' , $theline , 2 ) ;
 779                                         # Note that a '|' inside an invalid link should not
 780                                         # be mistaken as delimiting cell parameters
 781                                         if ( strpos( $y[0], '[[' ) !== false ) {
 782                                                 $y = array ($theline);
 783                                         }
 784                                         if ( count ( $y ) == 1 )
 785                                                 $y = "{$z}<{$l}>{$y[0]}" ;
 786                                         else {
 787                                                 $attributes = $this->unstripForHTML( $y[0] );
 788                                                 $y = "{$z}<{$l}".Sanitizer::fixTagAttributes($attributes, $l).">{$y[1]}" ;
 789                                         }
 790                                         $t[$k] .= $y ;
 791                                         array_push ( $td , true ) ;
 792                                 }
 793                         }
 794                 }
 795
 796                 # Closing open td, tr && table
 797                 while ( count ( $td ) > 0 )
 798                 {
 799                         $l = array_pop ( $ltd ) ;
 800                         if ( array_pop ( $td ) ) $t[] = '</td>' ;
 801                         if ( array_pop ( $tr ) ) $t[] = '</tr>' ;
 802                         if ( !array_pop ( $has_opened_tr ) ) $t[] = "<tr><td></td></tr>" ;
 803                         $t[] = '</table>' ;
 804                 }
 805
 806                 $t = implode ( "\n" , $t ) ;
 807                 wfProfileOut( $fname );
 808                 return $t ;
 809         }
 810
 811         /**
 812          * Helper function for parse() that transforms wiki markup into
 813          * HTML. Only called for $mOutputType == OT_HTML.
 814          *
 815          * @access private
 816          */
 817         function internalParse( $text ) {
 818                 global $wgContLang;
 819                 $args = array();
 820                 $isMain = true;
 821                 $fname = 'Parser::internalParse';
 822                 wfProfileIn( $fname );
 823
 824                 # Remove <noinclude> tags and <includeonly> sections
 825                 $text = strtr( $text, array( '<onlyinclude>' => '' , '</onlyinclude>' => '' ) );
 826                 $text = strtr( $text, array( '<noinclude>' => '', '</noinclude>' => '') );
 827                 $text = preg_replace( '/<includeonly>.*?<\/includeonly>/s', '', $text );
 828
 829                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ) );
 830                 $text = $this->replaceVariables( $text, $args );
 831
 832                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 833
 834                 $text = $this->doHeadings( $text );
 835                 if($this->mOptions->getUseDynamicDates()) {
 836                         $df =& DateFormatter::getInstance();
 837                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 838                 }
 839                 $text = $this->doAllQuotes( $text );
 840                 $text = $this->replaceInternalLinks( $text );
 841                 $text = $this->replaceExternalLinks( $text );
 842
 843                 # replaceInternalLinks may sometimes leave behind
 844                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 845                 $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
 846
 847                 $text = $this->doMagicLinks( $text );
 848                 $text = $this->doTableStuff( $text );
 849                 $text = $this->formatHeadings( $text, $isMain );
 850
 851                 wfProfileOut( $fname );
 852                 return $text;
 853         }
 854
 855         /**
 856          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 857          * magic external links.
 858          *
 859          * @access private
 860          */
 861         function &doMagicLinks( &$text ) {
 862                 $text = $this->magicISBN( $text );
 863                 $text = $this->magicRFC( $text, 'RFC ', 'rfcurl' );
 864                 $text = $this->magicRFC( $text, 'PMID ', 'pubmedurl' );
 865                 return $text;
 866         }
 867
 868         /**
 869          * Parse headers and return html
 870          *
 871          * @access private
 872          */
 873         function doHeadings( $text ) {
 874                 $fname = 'Parser::doHeadings';
 875                 wfProfileIn( $fname );
 876                 for ( $i = 6; $i >= 1; --$i ) {
 877                         $h = str_repeat( '=', $i );
 878                         $text = preg_replace( "/^{$h}(.+){$h}(\\s|$)/m",
 879                           "<h{$i}>\\1</h{$i}>\\2", $text );
 880                 }
 881                 wfProfileOut( $fname );
 882                 return $text;
 883         }
 884
 885         /**
 886          * Replace single quotes with HTML markup
 887          * @access private
 888          * @return string the altered text
 889          */
 890         function doAllQuotes( $text ) {
 891                 $fname = 'Parser::doAllQuotes';
 892                 wfProfileIn( $fname );
 893                 $outtext = '';
 894                 $lines = explode( "\n", $text );
 895                 foreach ( $lines as $line ) {
 896                         $outtext .= $this->doQuotes ( $line ) . "\n";
 897                 }
 898                 $outtext = substr($outtext, 0,-1);
 899                 wfProfileOut( $fname );
 900                 return $outtext;
 901         }
 902
 903         /**
 904          * Helper function for doAllQuotes()
 905          * @access private
 906          */
 907         function doQuotes( $text ) {
 908                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
 909                 if ( count( $arr ) == 1 )
 910                         return $text;
 911                 else
 912                 {
 913                         # First, do some preliminary work. This may shift some apostrophes from
 914                         # being mark-up to being text. It also counts the number of occurrences
 915                         # of bold and italics mark-ups.
 916                         $i = 0;
 917                         $numbold = 0;
 918                         $numitalics = 0;
 919                         foreach ( $arr as $r )
 920                         {
 921                                 if ( ( $i % 2 ) == 1 )
 922                                 {
 923                                         # If there are ever four apostrophes, assume the first is supposed to
 924                                         # be text, and the remaining three constitute mark-up for bold text.
 925                                         if ( strlen( $arr[$i] ) == 4 )
 926                                         {
 927                                                 $arr[$i-1] .= "'";
 928                                                 $arr[$i] = "'''";
 929                                         }
 930                                         # If there are more than 5 apostrophes in a row, assume they're all
 931                                         # text except for the last 5.
 932                                         else if ( strlen( $arr[$i] ) > 5 )
 933                                         {
 934                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
 935                                                 $arr[$i] = "'''''";
 936                                         }
 937                                         # Count the number of occurrences of bold and italics mark-ups.
 938                                         # We are not counting sequences of five apostrophes.
 939                                         if ( strlen( $arr[$i] ) == 2 ) $numitalics++;  else
 940                                         if ( strlen( $arr[$i] ) == 3 ) $numbold++;     else
 941                                         if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
 942                                 }
 943                                 $i++;
 944                         }
 945
 946                         # If there is an odd number of both bold and italics, it is likely
 947                         # that one of the bold ones was meant to be an apostrophe followed
 948                         # by italics. Which one we cannot know for certain, but it is more
 949                         # likely to be one that has a single-letter word before it.
 950                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
 951                         {
 952                                 $i = 0;
 953                                 $firstsingleletterword = -1;
 954                                 $firstmultiletterword = -1;
 955                                 $firstspace = -1;
 956                                 foreach ( $arr as $r )
 957                                 {
 958                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
 959                                         {
 960                                                 $x1 = substr ($arr[$i-1], -1);
 961                                                 $x2 = substr ($arr[$i-1], -2, 1);
 962                                                 if ($x1 == ' ') {
 963                                                         if ($firstspace == -1) $firstspace = $i;
 964                                                 } else if ($x2 == ' ') {
 965                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
 966                                                 } else {
 967                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
 968                                                 }
 969                                         }
 970                                         $i++;
 971                                 }
 972
 973                                 # If there is a single-letter word, use it!
 974                                 if ($firstsingleletterword > -1)
 975                                 {
 976                                         $arr [ $firstsingleletterword ] = "''";
 977                                         $arr [ $firstsingleletterword-1 ] .= "'";
 978                                 }
 979                                 # If not, but there's a multi-letter word, use that one.
 980                                 else if ($firstmultiletterword > -1)
 981                                 {
 982                                         $arr [ $firstmultiletterword ] = "''";
 983                                         $arr [ $firstmultiletterword-1 ] .= "'";
 984                                 }
 985                                 # ... otherwise use the first one that has neither.
 986                                 # (notice that it is possible for all three to be -1 if, for example,
 987                                 # there is only one pentuple-apostrophe in the line)
 988                                 else if ($firstspace > -1)
 989                                 {
 990                                         $arr [ $firstspace ] = "''";
 991                                         $arr [ $firstspace-1 ] .= "'";
 992                                 }
 993                         }
 994
 995                         # Now let's actually convert our apostrophic mush to HTML!
 996                         $output = '';
 997                         $buffer = '';
 998                         $state = '';
 999                         $i = 0;
1000                         foreach ($arr as $r)
1001                         {
1002                                 if (($i % 2) == 0)
1003                                 {
1004                                         if ($state == 'both')
1005                                                 $buffer .= $r;
1006                                         else
1007                                                 $output .= $r;
1008                                 }
1009                                 else
1010                                 {
1011                                         if (strlen ($r) == 2)
1012                                         {
1013                                                 if ($state == 'i')
1014                                                 { $output .= '</i>'; $state = ''; }
1015                                                 else if ($state == 'bi')
1016                                                 { $output .= '</i>'; $state = 'b'; }
1017                                                 else if ($state == 'ib')
1018                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1019                                                 else if ($state == 'both')
1020                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1021                                                 else # $state can be 'b' or ''
1022                                                 { $output .= '<i>'; $state .= 'i'; }
1023                                         }
1024                                         else if (strlen ($r) == 3)
1025                                         {
1026                                                 if ($state == 'b')
1027                                                 { $output .= '</b>'; $state = ''; }
1028                                                 else if ($state == 'bi')
1029                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1030                                                 else if ($state == 'ib')
1031                                                 { $output .= '</b>'; $state = 'i'; }
1032                                                 else if ($state == 'both')
1033                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1034                                                 else # $state can be 'i' or ''
1035                                                 { $output .= '<b>'; $state .= 'b'; }
1036                                         }
1037                                         else if (strlen ($r) == 5)
1038                                         {
1039                                                 if ($state == 'b')
1040                                                 { $output .= '</b><i>'; $state = 'i'; }
1041                                                 else if ($state == 'i')
1042                                                 { $output .= '</i><b>'; $state = 'b'; }
1043                                                 else if ($state == 'bi')
1044                                                 { $output .= '</i></b>'; $state = ''; }
1045                                                 else if ($state == 'ib')
1046                                                 { $output .= '</b></i>'; $state = ''; }
1047                                                 else if ($state == 'both')
1048                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1049                                                 else # ($state == '')
1050                                                 { $buffer = ''; $state = 'both'; }
1051                                         }
1052                                 }
1053                                 $i++;
1054                         }
1055                         # Now close all remaining tags.  Notice that the order is important.
1056                         if ($state == 'b' || $state == 'ib')
1057                                 $output .= '</b>';
1058                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
1059                                 $output .= '</i>';
1060                         if ($state == 'bi')
1061                                 $output .= '</b>';
1062                         if ($state == 'both')
1063                                 $output .= '<b><i>'.$buffer.'</i></b>';
1064                         return $output;
1065                 }
1066         }
1067
1068         /**
1069          * Replace external links
1070          *
1071          * Note: this is all very hackish and the order of execution matters a lot.
1072          * Make sure to run maintenance/parserTests.php if you change this code.
1073          *
1074          * @access private
1075          */
1076         function replaceExternalLinks( $text ) {
1077                 global $wgContLang;
1078                 $fname = 'Parser::replaceExternalLinks';
1079                 wfProfileIn( $fname );
1080
1081                 $sk =& $this->mOptions->getSkin();
1082
1083                 $bits = preg_split( EXT_LINK_BRACKETED, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1084
1085                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
1086
1087                 $i = 0;
1088                 while ( $i<count( $bits ) ) {
1089                         $url = $bits[$i++];
1090                         $protocol = $bits[$i++];
1091                         $text = $bits[$i++];
1092                         $trail = $bits[$i++];
1093
1094                         # The characters '<' and '>' (which were escaped by
1095                         # removeHTMLtags()) should not be included in
1096                         # URLs, per RFC 2396.
1097                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1098                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1099                                 $url = substr($url, 0, $m2[0][1]);
1100                         }
1101
1102                         # If the link text is an image URL, replace it with an <img> tag
1103                         # This happened by accident in the original parser, but some people used it extensively
1104                         $img = $this->maybeMakeExternalImage( $text );
1105                         if ( $img !== false ) {
1106                                 $text = $img;
1107                         }
1108
1109                         $dtrail = '';
1110
1111                         # Set linktype for CSS - if URL==text, link is essentially free
1112                         $linktype = ($text == $url) ? 'free' : 'text';
1113
1114                         # No link text, e.g. [http://domain.tld/some.link]
1115                         if ( $text == '' ) {
1116                                 # Autonumber if allowed
1117                                 if ( strpos( HTTP_PROTOCOLS, str_replace('/','\/', $protocol) ) !== false ) {
1118                                         $text = '[' . ++$this->mAutonumber . ']';
1119                                         $linktype = 'autonumber';
1120                                 } else {
1121                                         # Otherwise just use the URL
1122                                         $text = htmlspecialchars( $url );
1123                                         $linktype = 'free';
1124                                 }
1125                         } else {
1126                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1127                                 # Check for trail
1128                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1129                         }
1130
1131                         $text = $wgContLang->markNoConversion($text);
1132
1133                         # Replace &amp; from obsolete syntax with &.
1134                         # All HTML entities will be escaped by makeExternalLink()
1135                         $url = str_replace( '&amp;', '&', $url );
1136                         # Replace unnecessary URL escape codes with the referenced character
1137                         # This prevents spammers from hiding links from the filters
1138                         $url = Parser::replaceUnusualEscapes( $url );
1139
1140                         # Process the trail (i.e. everything after this link up until start of the next link),
1141                         # replacing any non-bracketed links
1142                         $trail = $this->replaceFreeExternalLinks( $trail );
1143
1144                         # Use the encoded URL
1145                         # This means that users can paste URLs directly into the text
1146                         # Funny characters like &ouml; aren't valid in URLs anyway
1147                         # This was changed in August 2004
1148                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype ) . $dtrail . $trail;
1149
1150                         # Register link in the output object
1151                         $this->mOutput->addExternalLink( $url );
1152                 }
1153
1154                 wfProfileOut( $fname );
1155                 return $s;
1156         }
1157
1158         /**
1159          * Replace anything that looks like a URL with a link
1160          * @access private
1161          */
1162         function replaceFreeExternalLinks( $text ) {
1163                 global $wgContLang;
1164                 $fname = 'Parser::replaceFreeExternalLinks';
1165                 wfProfileIn( $fname );
1166
1167                 $bits = preg_split( '/(\b(?:' . wfUrlProtocols() . '))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1168                 $s = array_shift( $bits );
1169                 $i = 0;
1170
1171                 $sk =& $this->mOptions->getSkin();
1172
1173                 while ( $i < count( $bits ) ){
1174                         $protocol = $bits[$i++];
1175                         $remainder = $bits[$i++];
1176
1177                         if ( preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1178                                 # Found some characters after the protocol that look promising
1179                                 $url = $protocol . $m[1];
1180                                 $trail = $m[2];
1181
1182                                 # The characters '<' and '>' (which were escaped by
1183                                 # removeHTMLtags()) should not be included in
1184                                 # URLs, per RFC 2396.
1185                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1186                                         $trail = substr($url, $m2[0][1]) . $trail;
1187                                         $url = substr($url, 0, $m2[0][1]);
1188                                 }
1189
1190                                 # Move trailing punctuation to $trail
1191                                 $sep = ',;\.:!?';
1192                                 # If there is no left bracket, then consider right brackets fair game too
1193                                 if ( strpos( $url, '(' ) === false ) {
1194                                         $sep .= ')';
1195                                 }
1196
1197                                 $numSepChars = strspn( strrev( $url ), $sep );
1198                                 if ( $numSepChars ) {
1199                                         $trail = substr( $url, -$numSepChars ) . $trail;
1200                                         $url = substr( $url, 0, -$numSepChars );
1201                                 }
1202
1203                                 # Replace &amp; from obsolete syntax with &.
1204                                 # All HTML entities will be escaped by makeExternalLink()
1205                                 # or maybeMakeExternalImage()
1206                                 $url = str_replace( '&amp;', '&', $url );
1207                                 # Replace unnecessary URL escape codes with their equivalent characters
1208                                 $url = Parser::replaceUnusualEscapes( $url );
1209
1210                                 # Is this an external image?
1211                                 $text = $this->maybeMakeExternalImage( $url );
1212                                 if ( $text === false ) {
1213                                         # Not an image, make a link
1214                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free' );
1215                                         # Register it in the output object
1216                                         $this->mOutput->addExternalLink( $url );
1217                                 }
1218                                 $s .= $text . $trail;
1219                         } else {
1220                                 $s .= $protocol . $remainder;
1221                         }
1222                 }
1223                 wfProfileOut( $fname );
1224                 return $s;
1225         }
1226
1227         /**
1228          * Replace unusual URL escape codes with their equivalent characters
1229          * @param string
1230          * @return string
1231          * @static
1232          */
1233         function replaceUnusualEscapes( $url ) {
1234                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1235                         array( 'Parser', 'replaceUnusualEscapesCallback' ), $url );
1236         }
1237
1238         /**
1239          * Callback function used in replaceUnusualEscapes().
1240          * Replaces unusual URL escape codes with their equivalent character
1241          * @static
1242          * @access private
1243          */
1244         function replaceUnusualEscapesCallback( $matches ) {
1245                 $char = urldecode( $matches[0] );
1246                 $ord = ord( $char );
1247                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1248                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1249                         // No, shouldn't be escaped
1250                         return $char;
1251                 } else {
1252                         // Yes, leave it escaped
1253                         return $matches[0];
1254                 }
1255         }
1256
1257         /**
1258          * make an image if it's allowed, either through the global
1259          * option or through the exception
1260          * @access private
1261          */
1262         function maybeMakeExternalImage( $url ) {
1263                 $sk =& $this->mOptions->getSkin();
1264                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1265                 $imagesexception = !empty($imagesfrom);
1266                 $text = false;
1267                 if ( $this->mOptions->getAllowExternalImages()
1268                      || ( $imagesexception && strpos( $url, $imagesfrom ) === 0 ) ) {
1269                         if ( preg_match( EXT_IMAGE_REGEX, $url ) ) {
1270                                 # Image found
1271                                 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
1272                         }
1273                 }
1274                 return $text;
1275         }
1276
1277         /**
1278          * Process [[ ]] wikilinks
1279          *
1280          * @access private
1281          */
1282         function replaceInternalLinks( $s ) {
1283                 global $wgContLang;
1284                 static $fname = 'Parser::replaceInternalLinks' ;
1285
1286                 wfProfileIn( $fname );
1287
1288                 wfProfileIn( $fname.'-setup' );
1289                 static $tc = FALSE;
1290                 # the % is needed to support urlencoded titles as well
1291                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1292
1293                 $sk =& $this->mOptions->getSkin();
1294
1295                 #split the entire text string on occurences of [[
1296                 $a = explode( '[[', ' ' . $s );
1297                 #get the first element (all text up to first [[), and remove the space we added
1298                 $s = array_shift( $a );
1299                 $s = substr( $s, 1 );
1300
1301                 # Match a link having the form [[namespace:link|alternate]]trail
1302                 static $e1 = FALSE;
1303                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1304                 # Match cases where there is no "]]", which might still be images
1305                 static $e1_img = FALSE;
1306                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1307                 # Match the end of a line for a word that's not followed by whitespace,
1308                 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1309                 $e2 = wfMsgForContent( 'linkprefix' );
1310
1311                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1312
1313                 if( is_null( $this->mTitle ) ) {
1314                         wfDebugDieBacktrace( 'nooo' );
1315                 }
1316                 $nottalk = !$this->mTitle->isTalkPage();
1317
1318                 if ( $useLinkPrefixExtension ) {
1319                         if ( preg_match( $e2, $s, $m ) ) {
1320                                 $first_prefix = $m[2];
1321                         } else {
1322                                 $first_prefix = false;
1323                         }
1324                 } else {
1325                         $prefix = '';
1326                 }
1327
1328                 $selflink = $this->mTitle->getPrefixedText();
1329                 wfProfileOut( $fname.'-setup' );
1330
1331                 $checkVariantLink = sizeof($wgContLang->getVariants())>1;
1332                 $useSubpages = $this->areSubpagesAllowed();
1333
1334                 # Loop for each link
1335                 for ($k = 0; isset( $a[$k] ); $k++) {
1336                         $line = $a[$k];
1337                         if ( $useLinkPrefixExtension ) {
1338                                 wfProfileIn( $fname.'-prefixhandling' );
1339                                 if ( preg_match( $e2, $s, $m ) ) {
1340                                         $prefix = $m[2];
1341                                         $s = $m[1];
1342                                 } else {
1343                                         $prefix='';
1344                                 }
1345                                 # first link
1346                                 if($first_prefix) {
1347                                         $prefix = $first_prefix;
1348                                         $first_prefix = false;
1349                                 }
1350                                 wfProfileOut( $fname.'-prefixhandling' );
1351                         }
1352
1353                         $might_be_img = false;
1354
1355                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1356                                 $text = $m[2];
1357                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1358                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1359                                 # the real problem is with the $e1 regex
1360                                 # See bug 1300.
1361                                 #
1362                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1363                                 # and no image is in sight. See bug 2095.
1364                                 #
1365                                 if( $text !== '' && preg_match( "/^\](.*)/s", $m[3], $n ) ) {
1366                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1367                                         $m[3] = $n[1];
1368                                 }
1369                                 # fix up urlencoded title texts
1370                                 if(preg_match('/%/', $m[1] )) $m[1] = urldecode($m[1]);
1371                                 $trail = $m[3];
1372                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1373                                 $might_be_img = true;
1374                                 $text = $m[2];
1375                                 if(preg_match('/%/', $m[1] )) $m[1] = urldecode($m[1]);
1376                                 $trail = "";
1377                         } else { # Invalid form; output directly
1378                                 $s .= $prefix . '[[' . $line ;
1379                                 continue;
1380                         }
1381
1382                         # Don't allow internal links to pages containing
1383                         # PROTO: where PROTO is a valid URL protocol; these
1384                         # should be external links.
1385                         if (preg_match('/^(\b(?:' . wfUrlProtocols() . '))/', $m[1])) {
1386                                 $s .= $prefix . '[[' . $line ;
1387                                 continue;
1388                         }
1389
1390                         # Make subpage if necessary
1391                         if( $useSubpages ) {
1392                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1393                         } else {
1394                                 $link = $m[1];
1395                         }
1396
1397                         $noforce = (substr($m[1], 0, 1) != ':');
1398                         if (!$noforce) {
1399                                 # Strip off leading ':'
1400                                 $link = substr($link, 1);
1401                         }
1402
1403                         $nt = Title::newFromText( $this->unstripNoWiki($link, $this->mStripState) );
1404                         if( !$nt ) {
1405                                 $s .= $prefix . '[[' . $line;
1406                                 continue;
1407                         }
1408
1409                         #check other language variants of the link
1410                         #if the article does not exist
1411                         if( $checkVariantLink
1412                             && $nt->getArticleID() == 0 ) {
1413                                 $wgContLang->findVariantLink($link, $nt);
1414                         }
1415
1416                         $ns = $nt->getNamespace();
1417                         $iw = $nt->getInterWiki();
1418
1419                         if ($might_be_img) { # if this is actually an invalid link
1420                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1421                                         $found = false;
1422                                         while (isset ($a[$k+1]) ) {
1423                                                 #look at the next 'line' to see if we can close it there
1424                                                 $spliced = array_splice( $a, $k + 1, 1 );
1425                                                 $next_line = array_shift( $spliced );
1426                                                 if( preg_match("/^(.*?]].*?)]](.*)$/sD", $next_line, $m) ) {
1427                                                 # the first ]] closes the inner link, the second the image
1428                                                         $found = true;
1429                                                         $text .= '[[' . $m[1];
1430                                                         $trail = $m[2];
1431                                                         break;
1432                                                 } elseif( preg_match("/^.*?]].*$/sD", $next_line, $m) ) {
1433                                                         #if there's exactly one ]] that's fine, we'll keep looking
1434                                                         $text .= '[[' . $m[0];
1435                                                 } else {
1436                                                         #if $next_line is invalid too, we need look no further
1437                                                         $text .= '[[' . $next_line;
1438                                                         break;
1439                                                 }
1440                                         }
1441                                         if ( !$found ) {
1442                                                 # we couldn't find the end of this imageLink, so output it raw
1443                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1444                                                 $text = $this->replaceInternalLinks($text);
1445                                                 $s .= $prefix . '[[' . $link . '|' . $text;
1446                                                 # note: no $trail, because without an end, there *is* no trail
1447                                                 continue;
1448                                         }
1449                                 } else { #it's not an image, so output it raw
1450                                         $s .= $prefix . '[[' . $link . '|' . $text;
1451                                         # note: no $trail, because without an end, there *is* no trail
1452                                         continue;
1453                                 }
1454                         }
1455
1456                         $wasblank = ( '' == $text );
1457                         if( $wasblank ) $text = $link;
1458
1459
1460                         # Link not escaped by : , create the various objects
1461                         if( $noforce ) {
1462
1463                                 # Interwikis
1464                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1465                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1466                                         $s = rtrim($s . "\n");
1467                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1468                                         continue;
1469                                 }
1470
1471                                 if ( $ns == NS_IMAGE ) {
1472                                         wfProfileIn( "$fname-image" );
1473                                         if ( !wfIsBadImage( $nt->getDBkey() ) ) {
1474                                                 # recursively parse links inside the image caption
1475                                                 # actually, this will parse them in any other parameters, too,
1476                                                 # but it might be hard to fix that, and it doesn't matter ATM
1477                                                 $text = $this->replaceExternalLinks($text);
1478                                                 $text = $this->replaceInternalLinks($text);
1479
1480                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1481                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text ) ) . $trail;
1482                                                 $this->mOutput->addImage( $nt->getDBkey() );
1483
1484                                                 wfProfileOut( "$fname-image" );
1485                                                 continue;
1486                                         }
1487                                         wfProfileOut( "$fname-image" );
1488
1489                                 }
1490
1491                                 if ( $ns == NS_CATEGORY ) {
1492                                         wfProfileIn( "$fname-category" );
1493                                         $s = rtrim($s . "\n"); # bug 87
1494
1495                                         if ( $wasblank ) {
1496                                                 if ( $this->mTitle->getNamespace() == NS_CATEGORY ) {
1497                                                         $sortkey = $this->mTitle->getText();
1498                                                 } else {
1499                                                         $sortkey = $this->mTitle->getPrefixedText();
1500                                                 }
1501                                         } else {
1502                                                 $sortkey = $text;
1503                                         }
1504                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1505                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1506                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1507
1508                                         /**
1509                                          * Strip the whitespace Category links produce, see bug 87
1510                                          * @todo We might want to use trim($tmp, "\n") here.
1511                                          */
1512                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1513
1514                                         wfProfileOut( "$fname-category" );
1515                                         continue;
1516                                 }
1517                         }
1518
1519                         if( ( $nt->getPrefixedText() === $selflink ) &&
1520                             ( $nt->getFragment() === '' ) ) {
1521                                 # Self-links are handled specially; generally de-link and change to bold.
1522                                 $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1523                                 continue;
1524                         }
1525
1526                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1527                         if( $ns == NS_MEDIA ) {
1528                                 $link = $sk->makeMediaLinkObj( $nt, $text );
1529                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1530                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1531                                 $this->mOutput->addImage( $nt->getDBkey() );
1532                                 continue;
1533                         } elseif( $ns == NS_SPECIAL ) {
1534                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1535                                 continue;
1536                         } elseif( $ns == NS_IMAGE ) {
1537                                 $img = Image::newFromTitle( $nt );
1538                                 if( $img->exists() ) {
1539                                         // Force a blue link if the file exists; may be a remote
1540                                         // upload on the shared repository, and we want to see its
1541                                         // auto-generated page.
1542                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1543                                         continue;
1544                                 }
1545                         }
1546                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1547                 }
1548                 wfProfileOut( $fname );
1549                 return $s;
1550         }
1551
1552         /**
1553          * Make a link placeholder. The text returned can be later resolved to a real link with
1554          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1555          * parsing of interwiki links, and secondly to allow all extistence checks and
1556          * article length checks (for stub links) to be bundled into a single query.
1557          *
1558          */
1559         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1560                 if ( ! is_object($nt) ) {
1561                         # Fail gracefully
1562                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1563                 } else {
1564                         # Separate the link trail from the rest of the link
1565                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1566
1567                         if ( $nt->isExternal() ) {
1568                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1569                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1570                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1571                         } else {
1572                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1573                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1574                                 $this->mLinkHolders['queries'][] = $query;
1575                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1576                                 $this->mLinkHolders['titles'][] = $nt;
1577
1578                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1579                         }
1580                 }
1581                 return $retVal;
1582         }
1583
1584         /**
1585          * Render a forced-blue link inline; protect against double expansion of
1586          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1587          * Since this little disaster has to split off the trail text to avoid
1588          * breaking URLs in the following text without breaking trails on the
1589          * wiki links, it's been made into a horrible function.
1590          *
1591          * @param Title $nt
1592          * @param string $text
1593          * @param string $query
1594          * @param string $trail
1595          * @param string $prefix
1596          * @return string HTML-wikitext mix oh yuck
1597          */
1598         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1599                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1600                 $sk =& $this->mOptions->getSkin();
1601                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1602                 return $this->armorLinks( $link ) . $trail;
1603         }
1604
1605         /**
1606          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1607          * going to go through further parsing steps before inline URL expansion.
1608          *
1609          * In particular this is important when using action=render, which causes
1610          * full URLs to be included.
1611          *
1612          * Oh man I hate our multi-layer parser!
1613          *
1614          * @param string more-or-less HTML
1615          * @return string less-or-more HTML with NOPARSE bits
1616          */
1617         function armorLinks( $text ) {
1618                 return preg_replace( "/\b(" . wfUrlProtocols() . ')/',
1619                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1620         }
1621
1622         /**
1623          * Return true if subpage links should be expanded on this page.
1624          * @return bool
1625          */
1626         function areSubpagesAllowed() {
1627                 # Some namespaces don't allow subpages
1628                 global $wgNamespacesWithSubpages;
1629                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1630         }
1631
1632         /**
1633          * Handle link to subpage if necessary
1634          * @param string $target the source of the link
1635          * @param string &$text the link text, modified as necessary
1636          * @return string the full name of the link
1637          * @access private
1638          */
1639         function maybeDoSubpageLink($target, &$text) {
1640                 # Valid link forms:
1641                 # Foobar -- normal
1642                 # :Foobar -- override special treatment of prefix (images, language links)
1643                 # /Foobar -- convert to CurrentPage/Foobar
1644                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1645                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1646                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1647
1648                 $fname = 'Parser::maybeDoSubpageLink';
1649                 wfProfileIn( $fname );
1650                 $ret = $target; # default return value is no change
1651
1652                 # Some namespaces don't allow subpages,
1653                 # so only perform processing if subpages are allowed
1654                 if( $this->areSubpagesAllowed() ) {
1655                         # Look at the first character
1656                         if( $target != '' && $target{0} == '/' ) {
1657                                 # / at end means we don't want the slash to be shown
1658                                 if( substr( $target, -1, 1 ) == '/' ) {
1659                                         $target = substr( $target, 1, -1 );
1660                                         $noslash = $target;
1661                                 } else {
1662                                         $noslash = substr( $target, 1 );
1663                                 }
1664
1665                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash);
1666                                 if( '' === $text ) {
1667                                         $text = $target;
1668                                 } # this might be changed for ugliness reasons
1669                         } else {
1670                                 # check for .. subpage backlinks
1671                                 $dotdotcount = 0;
1672                                 $nodotdot = $target;
1673                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1674                                         ++$dotdotcount;
1675                                         $nodotdot = substr( $nodotdot, 3 );
1676                                 }
1677                                 if($dotdotcount > 0) {
1678                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1679                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1680                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1681                                                 # / at the end means don't show full path
1682                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1683                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1684                                                         if( '' === $text ) {
1685                                                                 $text = $nodotdot;
1686                                                         }
1687                                                 }
1688                                                 $nodotdot = trim( $nodotdot );
1689                                                 if( $nodotdot != '' ) {
1690                                                         $ret .= '/' . $nodotdot;
1691                                                 }
1692                                         }
1693                                 }
1694                         }
1695                 }
1696
1697                 wfProfileOut( $fname );
1698                 return $ret;
1699         }
1700
1701         /**#@+
1702          * Used by doBlockLevels()
1703          * @access private
1704          */
1705         /* private */ function closeParagraph() {
1706                 $result = '';
1707                 if ( '' != $this->mLastSection ) {
1708                         $result = '</' . $this->mLastSection  . ">\n";
1709                 }
1710                 $this->mInPre = false;
1711                 $this->mLastSection = '';
1712                 return $result;
1713         }
1714         # getCommon() returns the length of the longest common substring
1715         # of both arguments, starting at the beginning of both.
1716         #
1717         /* private */ function getCommon( $st1, $st2 ) {
1718                 $fl = strlen( $st1 );
1719                 $shorter = strlen( $st2 );
1720                 if ( $fl < $shorter ) { $shorter = $fl; }
1721
1722                 for ( $i = 0; $i < $shorter; ++$i ) {
1723                         if ( $st1{$i} != $st2{$i} ) { break; }
1724                 }
1725                 return $i;
1726         }
1727         # These next three functions open, continue, and close the list
1728         # element appropriate to the prefix character passed into them.
1729         #
1730         /* private */ function openList( $char ) {
1731                 $result = $this->closeParagraph();
1732
1733                 if ( '*' == $char ) { $result .= '<ul><li>'; }
1734                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
1735                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
1736                 else if ( ';' == $char ) {
1737                         $result .= '<dl><dt>';
1738                         $this->mDTopen = true;
1739                 }
1740                 else { $result = '<!-- ERR 1 -->'; }
1741
1742                 return $result;
1743         }
1744
1745         /* private */ function nextItem( $char ) {
1746                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
1747                 else if ( ':' == $char || ';' == $char ) {
1748                         $close = '</dd>';
1749                         if ( $this->mDTopen ) { $close = '</dt>'; }
1750                         if ( ';' == $char ) {
1751                                 $this->mDTopen = true;
1752                                 return $close . '<dt>';
1753                         } else {
1754                                 $this->mDTopen = false;
1755                                 return $close . '<dd>';
1756                         }
1757                 }
1758                 return '<!-- ERR 2 -->';
1759         }
1760
1761         /* private */ function closeList( $char ) {
1762                 if ( '*' == $char ) { $text = '</li></ul>'; }
1763                 else if ( '#' == $char ) { $text = '</li></ol>'; }
1764                 else if ( ':' == $char ) {
1765                         if ( $this->mDTopen ) {
1766                                 $this->mDTopen = false;
1767                                 $text = '</dt></dl>';
1768                         } else {
1769                                 $text = '</dd></dl>';
1770                         }
1771                 }
1772                 else {  return '<!-- ERR 3 -->'; }
1773                 return $text."\n";
1774         }
1775         /**#@-*/
1776
1777         /**
1778          * Make lists from lines starting with ':', '*', '#', etc.
1779          *
1780          * @access private
1781          * @return string the lists rendered as HTML
1782          */
1783         function doBlockLevels( $text, $linestart ) {
1784                 $fname = 'Parser::doBlockLevels';
1785                 wfProfileIn( $fname );
1786
1787                 # Parsing through the text line by line.  The main thing
1788                 # happening here is handling of block-level elements p, pre,
1789                 # and making lists from lines starting with * # : etc.
1790                 #
1791                 $textLines = explode( "\n", $text );
1792
1793                 $lastPrefix = $output = '';
1794                 $this->mDTopen = $inBlockElem = false;
1795                 $prefixLength = 0;
1796                 $paragraphStack = false;
1797
1798                 if ( !$linestart ) {
1799                         $output .= array_shift( $textLines );
1800                 }
1801                 foreach ( $textLines as $oLine ) {
1802                         $lastPrefixLength = strlen( $lastPrefix );
1803                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
1804                         $preOpenMatch = preg_match('/<pre/i', $oLine );
1805                         if ( !$this->mInPre ) {
1806                                 # Multiple prefixes may abut each other for nested lists.
1807                                 $prefixLength = strspn( $oLine, '*#:;' );
1808                                 $pref = substr( $oLine, 0, $prefixLength );
1809
1810                                 # eh?
1811                                 $pref2 = str_replace( ';', ':', $pref );
1812                                 $t = substr( $oLine, $prefixLength );
1813                                 $this->mInPre = !empty($preOpenMatch);
1814                         } else {
1815                                 # Don't interpret any other prefixes in preformatted text
1816                                 $prefixLength = 0;
1817                                 $pref = $pref2 = '';
1818                                 $t = $oLine;
1819                         }
1820
1821                         # List generation
1822                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
1823                                 # Same as the last item, so no need to deal with nesting or opening stuff
1824                                 $output .= $this->nextItem( substr( $pref, -1 ) );
1825                                 $paragraphStack = false;
1826
1827                                 if ( substr( $pref, -1 ) == ';') {
1828                                         # The one nasty exception: definition lists work like this:
1829                                         # ; title : definition text
1830                                         # So we check for : in the remainder text to split up the
1831                                         # title and definition, without b0rking links.
1832                                         $term = $t2 = '';
1833                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
1834                                                 $t = $t2;
1835                                                 $output .= $term . $this->nextItem( ':' );
1836                                         }
1837                                 }
1838                         } elseif( $prefixLength || $lastPrefixLength ) {
1839                                 # Either open or close a level...
1840                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
1841                                 $paragraphStack = false;
1842
1843                                 while( $commonPrefixLength < $lastPrefixLength ) {
1844                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
1845                                         --$lastPrefixLength;
1846                                 }
1847                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
1848                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
1849                                 }
1850                                 while ( $prefixLength > $commonPrefixLength ) {
1851                                         $char = substr( $pref, $commonPrefixLength, 1 );
1852                                         $output .= $this->openList( $char );
1853
1854                                         if ( ';' == $char ) {
1855                                                 # FIXME: This is dupe of code above
1856                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
1857                                                         $t = $t2;
1858                                                         $output .= $term . $this->nextItem( ':' );
1859                                                 }
1860                                         }
1861                                         ++$commonPrefixLength;
1862                                 }
1863                                 $lastPrefix = $pref2;
1864                         }
1865                         if( 0 == $prefixLength ) {
1866                                 wfProfileIn( "$fname-paragraph" );
1867                                 # No prefix (not in list)--go to paragraph mode
1868                                 // XXX: use a stack for nestable elements like span, table and div
1869                                 $openmatch = preg_match('/(<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
1870                                 $closematch = preg_match(
1871                                         '/(<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
1872                                         '<td|<th|<div|<\\/div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul)/iS', $t );
1873                                 if ( $openmatch or $closematch ) {
1874                                         $paragraphStack = false;
1875                                         $output .= $this->closeParagraph();
1876                                         if ( $preOpenMatch and !$preCloseMatch ) {
1877                                                 $this->mInPre = true;
1878                                         }
1879                                         if ( $closematch ) {
1880                                                 $inBlockElem = false;
1881                                         } else {
1882                                                 $inBlockElem = true;
1883                                         }
1884                                 } else if ( !$inBlockElem && !$this->mInPre ) {
1885                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
1886                                                 // pre
1887                                                 if ($this->mLastSection != 'pre') {
1888                                                         $paragraphStack = false;
1889                                                         $output .= $this->closeParagraph().'<pre>';
1890                                                         $this->mLastSection = 'pre';
1891                                                 }
1892                                                 $t = substr( $t, 1 );
1893                                         } else {
1894                                                 // paragraph
1895                                                 if ( '' == trim($t) ) {
1896                                                         if ( $paragraphStack ) {
1897                                                                 $output .= $paragraphStack.'<br />';
1898                                                                 $paragraphStack = false;
1899                                                                 $this->mLastSection = 'p';
1900                                                         } else {
1901                                                                 if ($this->mLastSection != 'p' ) {
1902                                                                         $output .= $this->closeParagraph();
1903                                                                         $this->mLastSection = '';
1904                                                                         $paragraphStack = '<p>';
1905                                                                 } else {
1906                                                                         $paragraphStack = '</p><p>';
1907                                                                 }
1908                                                         }
1909                                                 } else {
1910                                                         if ( $paragraphStack ) {
1911                                                                 $output .= $paragraphStack;
1912                                                                 $paragraphStack = false;
1913                                                                 $this->mLastSection = 'p';
1914                                                         } else if ($this->mLastSection != 'p') {
1915                                                                 $output .= $this->closeParagraph().'<p>';
1916                                                                 $this->mLastSection = 'p';
1917                                                         }
1918                                                 }
1919                                         }
1920                                 }
1921                                 wfProfileOut( "$fname-paragraph" );
1922                         }
1923                         // somewhere above we forget to get out of pre block (bug 785)
1924                         if($preCloseMatch && $this->mInPre) {
1925                                 $this->mInPre = false;
1926                         }
1927                         if ($paragraphStack === false) {
1928                                 $output .= $t."\n";
1929                         }
1930                 }
1931                 while ( $prefixLength ) {
1932                         $output .= $this->closeList( $pref2{$prefixLength-1} );
1933                         --$prefixLength;
1934                 }
1935                 if ( '' != $this->mLastSection ) {
1936                         $output .= '</' . $this->mLastSection . '>';
1937                         $this->mLastSection = '';
1938                 }
1939
1940                 wfProfileOut( $fname );
1941                 return $output;
1942         }
1943
1944         /**
1945          * Split up a string on ':', ignoring any occurences inside
1946          * <a>..</a> or <span>...</span>
1947          * @param string $str the string to split
1948          * @param string &$before set to everything before the ':'
1949          * @param string &$after set to everything after the ':'
1950          * return string the position of the ':', or false if none found
1951          */
1952         function findColonNoLinks($str, &$before, &$after) {
1953                 # I wonder if we should make this count all tags, not just <a>
1954                 # and <span>. That would prevent us from matching a ':' that
1955                 # comes in the middle of italics other such formatting....
1956                 # -- Wil
1957                 $fname = 'Parser::findColonNoLinks';
1958                 wfProfileIn( $fname );
1959                 $pos = 0;
1960                 do {
1961                         $colon = strpos($str, ':', $pos);
1962
1963                         if ($colon !== false) {
1964                                 $before = substr($str, 0, $colon);
1965                                 $after = substr($str, $colon + 1);
1966
1967                                 # Skip any ':' within <a> or <span> pairs
1968                                 $a = substr_count($before, '<a');
1969                                 $s = substr_count($before, '<span');
1970                                 $ca = substr_count($before, '</a>');
1971                                 $cs = substr_count($before, '</span>');
1972
1973                                 if ($a <= $ca and $s <= $cs) {
1974                                         # Tags are balanced before ':'; ok
1975                                         break;
1976                                 }
1977                                 $pos = $colon + 1;
1978                         }
1979                 } while ($colon !== false);
1980                 wfProfileOut( $fname );
1981                 return $colon;
1982         }
1983
1984         /**
1985          * Return value of a magic variable (like PAGENAME)
1986          *
1987          * @access private
1988          */
1989         function getVariableValue( $index ) {
1990                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
1991
1992                 /**
1993                  * Some of these require message or data lookups and can be
1994                  * expensive to check many times.
1995                  */
1996                 static $varCache = array();
1997                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$varCache ) ) )
1998                         if ( isset( $varCache[$index] ) )
1999                                 return $varCache[$index];
2000
2001                 $ts = time();
2002                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2003
2004                 switch ( $index ) {
2005                         case MAG_CURRENTMONTH:
2006                                 return $varCache[$index] = $wgContLang->formatNum( date( 'm', $ts ) );
2007                         case MAG_CURRENTMONTHNAME:
2008                                 return $varCache[$index] = $wgContLang->getMonthName( date( 'n', $ts ) );
2009                         case MAG_CURRENTMONTHNAMEGEN:
2010                                 return $varCache[$index] = $wgContLang->getMonthNameGen( date( 'n', $ts ) );
2011                         case MAG_CURRENTMONTHABBREV:
2012                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( date( 'n', $ts ) );
2013                         case MAG_CURRENTDAY:
2014                                 return $varCache[$index] = $wgContLang->formatNum( date( 'j', $ts ) );
2015                         case MAG_CURRENTDAY2:
2016                                 return $varCache[$index] = $wgContLang->formatNum( date( 'd', $ts ) );
2017                         case MAG_PAGENAME:
2018                                 return $this->mTitle->getText();
2019                         case MAG_PAGENAMEE:
2020                                 return $this->mTitle->getPartialURL();
2021                         case MAG_FULLPAGENAME:
2022                                 return $this->mTitle->getPrefixedText();
2023                         case MAG_FULLPAGENAMEE:
2024                                 return $this->mTitle->getPrefixedURL();
2025                         case MAG_REVISIONID:
2026                                 return $this->mRevisionId;
2027                         case MAG_NAMESPACE:
2028                                 return $wgContLang->getNsText( $this->mTitle->getNamespace() );
2029                         case MAG_NAMESPACEE:
2030                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2031                         case MAG_CURRENTDAYNAME:
2032                                 return $varCache[$index] = $wgContLang->getWeekdayName( date( 'w', $ts ) + 1 );
2033                         case MAG_CURRENTYEAR:
2034                                 return $varCache[$index] = $wgContLang->formatNum( date( 'Y', $ts ), true );
2035                         case MAG_CURRENTTIME:
2036                                 return $varCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2037                         case MAG_CURRENTWEEK:
2038                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2039                                 // int to remove the padding
2040                                 return $varCache[$index] = $wgContLang->formatNum( (int)date( 'W', $ts ) );
2041                         case MAG_CURRENTDOW:
2042                                 return $varCache[$index] = $wgContLang->formatNum( date( 'w', $ts ) );
2043                         case MAG_NUMBEROFARTICLES:
2044                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfArticles() );
2045                         case MAG_NUMBEROFFILES:
2046                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfFiles() );
2047                         case MAG_SITENAME:
2048                                 return $wgSitename;
2049                         case MAG_SERVER:
2050                                 return $wgServer;
2051                         case MAG_SERVERNAME:
2052                                 return $wgServerName;
2053                         case MAG_SCRIPTPATH:
2054                                 return $wgScriptPath;
2055                         default:
2056                                 $ret = null;
2057                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$varCache, &$index, &$ret ) ) )
2058                                         return $ret;
2059                                 else
2060                                         return null;
2061                 }
2062         }
2063
2064         /**
2065          * initialise the magic variables (like CURRENTMONTHNAME)
2066          *
2067          * @access private
2068          */
2069         function initialiseVariables() {
2070                 $fname = 'Parser::initialiseVariables';
2071                 wfProfileIn( $fname );
2072                 global $wgVariableIDs;
2073                 $this->mVariables = array();
2074                 foreach ( $wgVariableIDs as $id ) {
2075                         $mw =& MagicWord::get( $id );
2076                         $mw->addToArray( $this->mVariables, $id );
2077                 }
2078                 wfProfileOut( $fname );
2079         }
2080
2081         /**
2082          * parse any parentheses in format ((title|part|part))
2083          * and call callbacks to get a replacement text for any found piece
2084          *
2085          * @param string $text The text to parse
2086          * @param array $callbacks rules in form:
2087          *     '{' => array(                            # opening parentheses
2088          *                                      'end' => '}',   # closing parentheses
2089          *                                      'cb' => array(2 => callback,    # replacement callback to call if {{..}} is found
2090          *                                                                4 => callback         # replacement callback to call if {{{{..}}}} is found
2091          *                                                                )
2092          *                                      )
2093          * @access private
2094          */
2095         function replace_callback ($text, $callbacks) {
2096                 $openingBraceStack = array();   # this array will hold a stack of parentheses which are not closed yet
2097                 $lastOpeningBrace = -1;                 # last not closed parentheses
2098
2099                 for ($i = 0; $i < strlen($text); $i++) {
2100                         # check for any opening brace
2101                         $rule = null;
2102                         $nextPos = -1;
2103                         foreach ($callbacks as $key => $value) {
2104                                 $pos = strpos ($text, $key, $i);
2105                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)) {
2106                                         $rule = $value;
2107                                         $nextPos = $pos;
2108                                 }
2109                         }
2110
2111                         if ($lastOpeningBrace >= 0) {
2112                                 $pos = strpos ($text, $openingBraceStack[$lastOpeningBrace]['braceEnd'], $i);
2113
2114                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)){
2115                                         $rule = null;
2116                                         $nextPos = $pos;
2117                                 }
2118
2119                                 $pos = strpos ($text, '|', $i);
2120
2121                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)){
2122                                         $rule = null;
2123                                         $nextPos = $pos;
2124                                 }
2125                         }
2126
2127                         if ($nextPos == -1)
2128                                 break;
2129
2130                         $i = $nextPos;
2131
2132                         # found openning brace, lets add it to parentheses stack
2133                         if (null != $rule) {
2134                                 $piece = array('brace' => $text[$i],
2135                                                            'braceEnd' => $rule['end'],
2136                                                            'count' => 1,
2137                                                            'title' => '',
2138                                                            'parts' => null);
2139
2140                                 # count openning brace characters
2141                                 while ($i+1 < strlen($text) && $text[$i+1] == $piece['brace']) {
2142                                         $piece['count']++;
2143                                         $i++;
2144                                 }
2145
2146                                 $piece['startAt'] = $i+1;
2147                                 $piece['partStart'] = $i+1;
2148
2149                                 # we need to add to stack only if openning brace count is enough for any given rule
2150                                 foreach ($rule['cb'] as $cnt => $fn) {
2151                                         if ($piece['count'] >= $cnt) {
2152                                                 $lastOpeningBrace ++;
2153                                                 $openingBraceStack[$lastOpeningBrace] = $piece;
2154                                                 break;
2155                                         }
2156                                 }
2157
2158                                 continue;
2159                         }
2160                         else if ($lastOpeningBrace >= 0) {
2161                                 # first check if it is a closing brace
2162                                 if ($openingBraceStack[$lastOpeningBrace]['braceEnd'] == $text[$i]) {
2163                                         # lets check if it is enough characters for closing brace
2164                                         $count = 1;
2165                                         while ($i+$count < strlen($text) && $text[$i+$count] == $text[$i])
2166                                                 $count++;
2167
2168                                         # if there are more closing parentheses than opening ones, we parse less
2169                                         if ($openingBraceStack[$lastOpeningBrace]['count'] < $count)
2170                                                 $count = $openingBraceStack[$lastOpeningBrace]['count'];
2171
2172                                         # check for maximum matching characters (if there are 5 closing characters, we will probably need only 3 - depending on the rules)
2173                                         $matchingCount = 0;
2174                                         $matchingCallback = null;
2175                                         foreach ($callbacks[$openingBraceStack[$lastOpeningBrace]['brace']]['cb'] as $cnt => $fn) {
2176                                                 if ($count >= $cnt && $matchingCount < $cnt) {
2177                                                         $matchingCount = $cnt;
2178                                                         $matchingCallback = $fn;
2179                                                 }
2180                                         }
2181
2182                                         if ($matchingCount == 0) {
2183                                                 $i += $count - 1;
2184                                                 continue;
2185                                         }
2186
2187                                         # lets set a title or last part (if '|' was found)
2188                                         if (null === $openingBraceStack[$lastOpeningBrace]['parts'])
2189                                                 $openingBraceStack[$lastOpeningBrace]['title'] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2190                                         else
2191                                                 $openingBraceStack[$lastOpeningBrace]['parts'][] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2192
2193                                         $pieceStart = $openingBraceStack[$lastOpeningBrace]['startAt'] - $matchingCount;
2194                                         $pieceEnd = $i + $matchingCount;
2195
2196                                         if( is_callable( $matchingCallback ) ) {
2197                                                 $cbArgs = array (
2198                                                                                  'text' => substr($text, $pieceStart, $pieceEnd - $pieceStart),
2199                                                                                  'title' => trim($openingBraceStack[$lastOpeningBrace]['title']),
2200                                                                                  'parts' => $openingBraceStack[$lastOpeningBrace]['parts'],
2201                                                                                  'lineStart' => (($pieceStart > 0) && ($text[$pieceStart-1] == '\n')),
2202                                                                                  );
2203                                                 # finally we can call a user callback and replace piece of text
2204                                                 $replaceWith = call_user_func( $matchingCallback, $cbArgs );
2205                                                 $text = substr($text, 0, $pieceStart) . $replaceWith . substr($text, $pieceEnd);
2206                                                 $i = $pieceStart + strlen($replaceWith) - 1;
2207                                         }
2208                                         else {
2209                                                 # null value for callback means that parentheses should be parsed, but not replaced
2210                                                 $i += $matchingCount - 1;
2211                                         }
2212
2213                                         # reset last openning parentheses, but keep it in case there are unused characters
2214                                         $piece = array('brace' => $openingBraceStack[$lastOpeningBrace]['brace'],
2215                                                                    'braceEnd' => $openingBraceStack[$lastOpeningBrace]['braceEnd'],
2216                                                                    'count' => $openingBraceStack[$lastOpeningBrace]['count'],
2217                                                                    'title' => '',
2218                                                                    'parts' => null,
2219                                                                    'startAt' => $openingBraceStack[$lastOpeningBrace]['startAt']);
2220                                         $openingBraceStack[$lastOpeningBrace--] = null;
2221
2222                                         if ($matchingCount < $piece['count']) {
2223                                                 $piece['count'] -= $matchingCount;
2224                                                 $piece['startAt'] -= $matchingCount;
2225                                                 $piece['partStart'] = $piece['startAt'];
2226                                                 # do we still qualify for any callback with remaining count?
2227                                                 foreach ($callbacks[$piece['brace']]['cb'] as $cnt => $fn) {
2228                                                         if ($piece['count'] >= $cnt) {
2229                                                                 $lastOpeningBrace ++;
2230                                                                 $openingBraceStack[$lastOpeningBrace] = $piece;
2231                                                                 break;
2232                                                         }
2233                                                 }
2234                                         }
2235                                         continue;
2236                                 }
2237
2238                                 # lets set a title if it is a first separator, or next part otherwise
2239                                 if ($text[$i] == '|') {
2240                                         if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2241                                                 $openingBraceStack[$lastOpeningBrace]['title'] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2242                                                 $openingBraceStack[$lastOpeningBrace]['parts'] = array();
2243                                         }
2244                                         else
2245                                                 $openingBraceStack[$lastOpeningBrace]['parts'][] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2246
2247                                         $openingBraceStack[$lastOpeningBrace]['partStart'] = $i + 1;
2248                                 }
2249                         }
2250                 }
2251
2252                 return $text;
2253         }
2254
2255         /**
2256          * Replace magic variables, templates, and template arguments
2257          * with the appropriate text. Templates are substituted recursively,
2258          * taking care to avoid infinite loops.
2259          *
2260          * Note that the substitution depends on value of $mOutputType:
2261          *  OT_WIKI: only {{subst:}} templates
2262          *  OT_MSG: only magic variables
2263          *  OT_HTML: all templates and magic variables
2264          *
2265          * @param string $tex The text to transform
2266          * @param array $args Key-value pairs representing template parameters to substitute
2267          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2268          * @access private
2269          */
2270         function replaceVariables( $text, $args = array(), $argsOnly = false ) {
2271                 # Prevent too big inclusions
2272                 if( strlen( $text ) > MAX_INCLUDE_SIZE ) {
2273                         return $text;
2274                 }
2275
2276                 $fname = 'Parser::replaceVariables';
2277                 wfProfileIn( $fname );
2278
2279                 # This function is called recursively. To keep track of arguments we need a stack:
2280                 array_push( $this->mArgStack, $args );
2281
2282                 $braceCallbacks = array();
2283                 if ( !$argsOnly ) {
2284                         $braceCallbacks[2] = array( &$this, 'braceSubstitution' );
2285                 }
2286                 if ( $this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI ) {
2287                         $braceCallbacks[3] = array( &$this, 'argSubstitution' );
2288                 }
2289                 $callbacks = array();
2290                 $callbacks['{'] = array('end' => '}', 'cb' => $braceCallbacks);
2291                 $callbacks['['] = array('end' => ']', 'cb' => array(2=>null));
2292                 $text = $this->replace_callback ($text, $callbacks);
2293
2294                 array_pop( $this->mArgStack );
2295
2296                 wfProfileOut( $fname );
2297                 return $text;
2298         }
2299
2300         /**
2301          * Replace magic variables
2302          * @access private
2303          */
2304         function variableSubstitution( $matches ) {
2305                 $fname = 'Parser::variableSubstitution';
2306                 $varname = $matches[1];
2307                 wfProfileIn( $fname );
2308                 if ( !$this->mVariables ) {
2309                         $this->initialiseVariables();
2310                 }
2311                 $skip = false;
2312                 if ( $this->mOutputType == OT_WIKI ) {
2313                         # Do only magic variables prefixed by SUBST
2314                         $mwSubst =& MagicWord::get( MAG_SUBST );
2315                         if (!$mwSubst->matchStartAndRemove( $varname ))
2316                                 $skip = true;
2317                         # Note that if we don't substitute the variable below,
2318                         # we don't remove the {{subst:}} magic word, in case
2319                         # it is a template rather than a magic variable.
2320                 }
2321                 if ( !$skip && array_key_exists( $varname, $this->mVariables ) ) {
2322                         $id = $this->mVariables[$varname];
2323                         $text = $this->getVariableValue( $id );
2324                         $this->mOutput->mContainsOldMagic = true;
2325                 } else {
2326                         $text = $matches[0];
2327                 }
2328                 wfProfileOut( $fname );
2329                 return $text;
2330         }
2331
2332         # Split template arguments
2333         function getTemplateArgs( $argsString ) {
2334                 if ( $argsString === '' ) {
2335                         return array();
2336                 }
2337
2338                 $args = explode( '|', substr( $argsString, 1 ) );
2339
2340                 # If any of the arguments contains a '[[' but no ']]', it needs to be
2341                 # merged with the next arg because the '|' character between belongs
2342                 # to the link syntax and not the template parameter syntax.
2343                 $argc = count($args);
2344
2345                 for ( $i = 0; $i < $argc-1; $i++ ) {
2346                         if ( substr_count ( $args[$i], '[[' ) != substr_count ( $args[$i], ']]' ) ) {
2347                                 $args[$i] .= '|'.$args[$i+1];
2348                                 array_splice($args, $i+1, 1);
2349                                 $i--;
2350                                 $argc--;
2351                         }
2352                 }
2353
2354                 return $args;
2355         }
2356
2357         /**
2358          * Return the text of a template, after recursively
2359          * replacing any variables or templates within the template.
2360          *
2361          * @param array $piece The parts of the template
2362          *  $piece['text']: matched text
2363          *  $piece['title']: the title, i.e. the part before the |
2364          *  $piece['parts']: the parameter array
2365          * @return string the text of the template
2366          * @access private
2367          */
2368         function braceSubstitution( $piece ) {
2369                 global $wgContLang;
2370                 $fname = 'Parser::braceSubstitution';
2371                 wfProfileIn( $fname );
2372
2373                 # Flags
2374                 $found = false;             # $text has been filled
2375                 $nowiki = false;            # wiki markup in $text should be escaped
2376                 $noparse = false;           # Unsafe HTML tags should not be stripped, etc.
2377                 $noargs = false;            # Don't replace triple-brace arguments in $text
2378                 $replaceHeadings = false;   # Make the edit section links go to the template not the article
2379                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2380                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2381
2382                 # Title object, where $text came from
2383                 $title = NULL;
2384
2385                 $linestart = '';
2386
2387                 # $part1 is the bit before the first |, and must contain only title characters
2388                 # $args is a list of arguments, starting from index 0, not including $part1
2389
2390                 $part1 = $piece['title'];
2391                 # If the third subpattern matched anything, it will start with |
2392
2393                 if (null == $piece['parts']) {
2394                         $replaceWith = $this->variableSubstitution (array ($piece['text'], $piece['title']));
2395                         if ($replaceWith != $piece['text']) {
2396                                 $text = $replaceWith;
2397                                 $found = true;
2398                                 $noparse = true;
2399                                 $noargs = true;
2400                         }
2401                 }
2402
2403                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2404                 $argc = count( $args );
2405
2406                 # SUBST
2407                 if ( !$found ) {
2408                         $mwSubst =& MagicWord::get( MAG_SUBST );
2409                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor ($this->mOutputType == OT_WIKI) ) {
2410                                 # One of two possibilities is true:
2411                                 # 1) Found SUBST but not in the PST phase
2412                                 # 2) Didn't find SUBST and in the PST phase
2413                                 # In either case, return without further processing
2414                                 $text = $piece['text'];
2415                                 $found = true;
2416                                 $noparse = true;
2417                                 $noargs = true;
2418                         }
2419                 }
2420
2421                 # MSG, MSGNW, INT and RAW
2422                 if ( !$found ) {
2423                         # Check for MSGNW:
2424                         $mwMsgnw =& MagicWord::get( MAG_MSGNW );
2425                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2426                                 $nowiki = true;
2427                         } else {
2428                                 # Remove obsolete MSG:
2429                                 $mwMsg =& MagicWord::get( MAG_MSG );
2430                                 $mwMsg->matchStartAndRemove( $part1 );
2431                         }
2432
2433                         # Check for RAW:
2434                         $mwRaw =& MagicWord::get( MAG_RAW );
2435                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2436                                 $forceRawInterwiki = true;
2437                         }
2438
2439                         # Check if it is an internal message
2440                         $mwInt =& MagicWord::get( MAG_INT );
2441                         if ( $mwInt->matchStartAndRemove( $part1 ) ) {
2442                                 if ( $this->incrementIncludeCount( 'int:'.$part1 ) ) {
2443                                         $text = $linestart . wfMsgReal( $part1, $args, true );
2444                                         $found = true;
2445                                 }
2446                         }
2447                 }
2448
2449                 # NS
2450                 if ( !$found ) {
2451                         # Check for NS: (namespace expansion)
2452                         $mwNs = MagicWord::get( MAG_NS );
2453                         if ( $mwNs->matchStartAndRemove( $part1 ) ) {
2454                                 if ( intval( $part1 ) || $part1 == "0" ) {
2455                                         $text = $linestart . $wgContLang->getNsText( intval( $part1 ) );
2456                                         $found = true;
2457                                 } else {
2458                                         $index = Namespace::getCanonicalIndex( strtolower( $part1 ) );
2459                                         if ( !is_null( $index ) ) {
2460                                                 $text = $linestart . $wgContLang->getNsText( $index );
2461                                                 $found = true;
2462                                         }
2463                                 }
2464                         }
2465                 }
2466
2467                 # LCFIRST, UCFIRST, LC and UC
2468                 if ( !$found ) {
2469                         $lcfirst =& MagicWord::get( MAG_LCFIRST );
2470                         $ucfirst =& MagicWord::get( MAG_UCFIRST );
2471                         $lc =& MagicWord::get( MAG_LC );
2472                         $uc =& MagicWord::get( MAG_UC );
2473                         if ( $lcfirst->matchStartAndRemove( $part1 ) ) {
2474                                 $text = $linestart . $wgContLang->lcfirst( $part1 );
2475                                 $found = true;
2476                         } else if ( $ucfirst->matchStartAndRemove( $part1 ) ) {
2477                                 $text = $linestart . $wgContLang->ucfirst( $part1 );
2478                                 $found = true;
2479                         } else if ( $lc->matchStartAndRemove( $part1 ) ) {
2480                                 $text = $linestart . $wgContLang->lc( $part1 );
2481                                 $found = true;
2482                         } else if ( $uc->matchStartAndRemove( $part1 ) ) {
2483                                  $text = $linestart . $wgContLang->uc( $part1 );
2484                                  $found = true;
2485                         }
2486                 }
2487
2488                 # LOCALURL and FULLURL
2489                 if ( !$found ) {
2490                         $mwLocal =& MagicWord::get( MAG_LOCALURL );
2491                         $mwLocalE =& MagicWord::get( MAG_LOCALURLE );
2492                         $mwFull =& MagicWord::get( MAG_FULLURL );
2493                         $mwFullE =& MagicWord::get( MAG_FULLURLE );
2494
2495
2496                         if ( $mwLocal->matchStartAndRemove( $part1 ) ) {
2497                                 $func = 'getLocalURL';
2498                         } elseif ( $mwLocalE->matchStartAndRemove( $part1 ) ) {
2499                                 $func = 'escapeLocalURL';
2500                         } elseif ( $mwFull->matchStartAndRemove( $part1 ) ) {
2501                                 $func = 'getFullURL';
2502                         } elseif ( $mwFullE->matchStartAndRemove( $part1 ) ) {
2503                                 $func = 'escapeFullURL';
2504                         } else {
2505                                 $func = false;
2506                         }
2507
2508                         if ( $func !== false ) {
2509                                 $title = Title::newFromText( $part1 );
2510                                 if ( !is_null( $title ) ) {
2511                                         if ( $argc > 0 ) {
2512                                                 $text = $linestart . $title->$func( $args[0] );
2513                                         } else {
2514                                                 $text = $linestart . $title->$func();
2515                                         }
2516                                         $found = true;
2517                                 }
2518                         }
2519                 }
2520
2521                 # GRAMMAR
2522                 if ( !$found && $argc == 1 ) {
2523                         $mwGrammar =& MagicWord::get( MAG_GRAMMAR );
2524                         if ( $mwGrammar->matchStartAndRemove( $part1 ) ) {
2525                                 $text = $linestart . $wgContLang->convertGrammar( $args[0], $part1 );
2526                                 $found = true;
2527                         }
2528                 }
2529
2530                 # PLURAL
2531                 if ( !$found && $argc >= 2 ) {
2532                         $mwPluralForm =& MagicWord::get( MAG_PLURAL );
2533                         if ( $mwPluralForm->matchStartAndRemove( $part1 ) ) {
2534                                 if ($argc==2) {$args[2]=$args[1];}
2535                                 $text = $linestart . $wgContLang->convertPlural( $part1, $args[0], $args[1], $args[2]);
2536                                 $found = true;
2537                         }
2538                 }
2539
2540                 # Template table test
2541
2542                 # Did we encounter this template already? If yes, it is in the cache
2543                 # and we need to check for loops.
2544                 if ( !$found && isset( $this->mTemplates[$piece['title']] ) ) {
2545                         $found = true;
2546
2547                         # Infinite loop test
2548                         if ( isset( $this->mTemplatePath[$part1] ) ) {
2549                                 $noparse = true;
2550                                 $noargs = true;
2551                                 $found = true;
2552                                 $text = $linestart .
2553                                         '{{' . $part1 . '}}' .
2554                                         '<!-- WARNING: template loop detected -->';
2555                                 wfDebug( "$fname: template loop broken at '$part1'\n" );
2556                         } else {
2557                                 # set $text to cached message.
2558                                 $text = $linestart . $this->mTemplates[$piece['title']];
2559                         }
2560                 }
2561
2562                 # Load from database
2563                 $lastPathLevel = $this->mTemplatePath;
2564                 if ( !$found ) {
2565                         $ns = NS_TEMPLATE;
2566                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage='' );
2567                         if ($subpage !== '') {
2568                                 $ns = $this->mTitle->getNamespace();
2569                         }
2570                         $title = Title::newFromText( $part1, $ns );
2571
2572                         if ( !is_null( $title ) ) {
2573                                 if ( !$title->isExternal() ) {
2574                                         # Check for excessive inclusion
2575                                         $dbk = $title->getPrefixedDBkey();
2576                                         if ( $this->incrementIncludeCount( $dbk ) ) {
2577                                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() ) {
2578                                                         # Capture special page output
2579                                                         $text = SpecialPage::capturePath( $title );
2580                                                         if ( is_string( $text ) ) {
2581                                                                 $found = true;
2582                                                                 $noparse = true;
2583                                                                 $noargs = true;
2584                                                                 $isHTML = true;
2585                                                                 $this->disableCache();
2586                                                         }
2587                                                 } else {
2588                                                         $articleContent = $this->fetchTemplate( $title );
2589                                                         if ( $articleContent !== false ) {
2590                                                                 $found = true;
2591                                                                 $text = $articleContent;
2592                                                                 $replaceHeadings = true;
2593                                                         }
2594                                                 }
2595                                         }
2596
2597                                         # If the title is valid but undisplayable, make a link to it
2598                                         if ( $this->mOutputType == OT_HTML && !$found ) {
2599                                                 $text = '[['.$title->getPrefixedText().']]';
2600                                                 $found = true;
2601                                         }
2602                                 } elseif ( $title->isTrans() ) {
2603                                         // Interwiki transclusion
2604                                         if ( $this->mOutputType == OT_HTML && !$forceRawInterwiki ) {
2605                                                 $text = $this->interwikiTransclude( $title, 'render' );
2606                                                 $isHTML = true;
2607                                                 $noparse = true;
2608                                         } else {
2609                                                 $text = $this->interwikiTransclude( $title, 'raw' );
2610                                                 $replaceHeadings = true;
2611                                         }
2612                                         $found = true;
2613                                 }
2614
2615                                 # Template cache array insertion
2616                                 # Use the original $piece['title'] not the mangled $part1, so that
2617                                 # modifiers such as RAW: produce separate cache entries
2618                                 if( $found ) {
2619                                         $this->mTemplates[$piece['title']] = $text;
2620                                         $text = $linestart . $text;
2621                                 }
2622                         }
2623                 }
2624
2625                 # Recursive parsing, escaping and link table handling
2626                 # Only for HTML output
2627                 if ( $nowiki && $found && $this->mOutputType == OT_HTML ) {
2628                         $text = wfEscapeWikiText( $text );
2629                 } elseif ( ($this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI) && $found ) {
2630                         if ( !$noargs ) {
2631                                 # Clean up argument array
2632                                 $assocArgs = array();
2633                                 $index = 1;
2634                                 foreach( $args as $arg ) {
2635                                         $eqpos = strpos( $arg, '=' );
2636                                         if ( $eqpos === false ) {
2637                                                 $assocArgs[$index++] = $arg;
2638                                         } else {
2639                                                 $name = trim( substr( $arg, 0, $eqpos ) );
2640                                                 $value = trim( substr( $arg, $eqpos+1 ) );
2641                                                 if ( $value === false ) {
2642                                                         $value = '';
2643                                                 }
2644                                                 if ( $name !== false ) {
2645                                                         $assocArgs[$name] = $value;
2646                                                 }
2647                                         }
2648                                 }
2649
2650                                 # Add a new element to the templace recursion path
2651                                 $this->mTemplatePath[$part1] = 1;
2652                         }
2653
2654                         if ( !$noparse ) {
2655                                 # If there are any <onlyinclude> tags, only include them
2656                                 if ( in_string( '<onlyinclude>', $text ) && in_string( '</onlyinclude>', $text ) ) {
2657                                         preg_match_all( '/<onlyinclude>(.*?)\n?<\/onlyinclude>/s', $text, $m );
2658                                         $text = '';
2659                                         foreach ($m[1] as $piece)
2660                                                 $text .= $piece;
2661                                 }
2662                                 # Remove <noinclude> sections and <includeonly> tags
2663                                 $text = preg_replace( '/<noinclude>.*?<\/noinclude>/s', '', $text );
2664                                 $text = strtr( $text, array( '<includeonly>' => '' , '</includeonly>' => '' ) );
2665
2666                                 if( $this->mOutputType == OT_HTML ) {
2667                                         # Strip <nowiki>, <pre>, etc.
2668                                         $text = $this->strip( $text, $this->mStripState );
2669                                         $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'replaceVariables' ), $assocArgs );
2670                                 }
2671                                 $text = $this->replaceVariables( $text, $assocArgs );
2672
2673                                 # If the template begins with a table or block-level
2674                                 # element, it should be treated as beginning a new line.
2675                                 if (!$piece['lineStart'] && preg_match('/^({\\||:|;|#|\*)/', $text)) {
2676                                         $text = "\n" . $text;
2677                                 }
2678                         } elseif ( !$noargs ) {
2679                                 # $noparse and !$noargs
2680                                 # Just replace the arguments, not any double-brace items
2681                                 # This is used for rendered interwiki transclusion
2682                                 $text = $this->replaceVariables( $text, $assocArgs, true );
2683                         }
2684                 }
2685                 # Prune lower levels off the recursion check path
2686                 $this->mTemplatePath = $lastPathLevel;
2687
2688                 if ( !$found ) {
2689                         wfProfileOut( $fname );
2690                         return $piece['text'];
2691                 } else {
2692                         if ( $isHTML ) {
2693                                 # Replace raw HTML by a placeholder
2694                                 # Add a blank line preceding, to prevent it from mucking up
2695                                 # immediately preceding headings
2696                                 $text = "\n\n" . $this->insertStripItem( $text, $this->mStripState );
2697                         } else {
2698                                 # replace ==section headers==
2699                                 # XXX this needs to go away once we have a better parser.
2700                                 if ( $this->mOutputType != OT_WIKI && $replaceHeadings ) {
2701                                         if( !is_null( $title ) )
2702                                                 $encodedname = base64_encode($title->getPrefixedDBkey());
2703                                         else
2704                                                 $encodedname = base64_encode("");
2705                                         $m = preg_split('/(^={1,6}.*?={1,6}\s*?$)/m', $text, -1,
2706                                                 PREG_SPLIT_DELIM_CAPTURE);
2707                                         $text = '';
2708                                         $nsec = 0;
2709                                         for( $i = 0; $i < count($m); $i += 2 ) {
2710                                                 $text .= $m[$i];
2711                                                 if (!isset($m[$i + 1]) || $m[$i + 1] == "") continue;
2712                                                 $hl = $m[$i + 1];
2713                                                 if( strstr($hl, "<!--MWTEMPLATESECTION") ) {
2714                                                         $text .= $hl;
2715                                                         continue;
2716                                                 }
2717                                                 preg_match('/^(={1,6})(.*?)(={1,6})\s*?$/m', $hl, $m2);
2718                                                 $text .= $m2[1] . $m2[2] . "<!--MWTEMPLATESECTION="
2719                                                         . $encodedname . "&" . base64_encode("$nsec") . "-->" . $m2[3];
2720
2721                                                 $nsec++;
2722                                         }
2723                                 }
2724                         }
2725                 }
2726
2727                 # Prune lower levels off the recursion check path
2728                 $this->mTemplatePath = $lastPathLevel;
2729
2730                 if ( !$found ) {
2731                         wfProfileOut( $fname );
2732                         return $piece['text'];
2733                 } else {
2734                         wfProfileOut( $fname );
2735                         return $text;
2736                 }
2737         }
2738
2739         /**
2740          * Fetch the unparsed text of a template and register a reference to it.
2741          */
2742         function fetchTemplate( $title ) {
2743                 $text = false;
2744                 // Loop to fetch the article, with up to 1 redirect
2745                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
2746                         $rev = Revision::newFromTitle( $title );
2747                         $this->mOutput->addTemplate( $title, $title->getArticleID() );
2748                         if ( !$rev ) {
2749                                 break;
2750                         }
2751                         $text = $rev->getText();
2752                         if ( $text === false ) {
2753                                 break;
2754                         }
2755                         // Redirect?
2756                         $title = Title::newFromRedirect( $text );
2757                 }
2758                 return $text;
2759         }
2760
2761         /**
2762          * Transclude an interwiki link.
2763          */
2764         function interwikiTransclude( $title, $action ) {
2765                 global $wgEnableScaryTranscluding, $wgCanonicalNamespaceNames;
2766
2767                 if (!$wgEnableScaryTranscluding)
2768                         return wfMsg('scarytranscludedisabled');
2769
2770                 // The namespace will actually only be 0 or 10, depending on whether there was a leading :
2771                 // But we'll handle it generally anyway
2772                 if ( $title->getNamespace() ) {
2773                         // Use the canonical namespace, which should work anywhere
2774                         $articleName = $wgCanonicalNamespaceNames[$title->getNamespace()] . ':' . $title->getDBkey();
2775                 } else {
2776                         $articleName = $title->getDBkey();
2777                 }
2778
2779                 $url = str_replace('$1', urlencode($articleName), Title::getInterwikiLink($title->getInterwiki()));
2780                 $url .= "?action=$action";
2781                 if (strlen($url) > 255)
2782                         return wfMsg('scarytranscludetoolong');
2783                 return $this->fetchScaryTemplateMaybeFromCache($url);
2784         }
2785
2786         function fetchScaryTemplateMaybeFromCache($url) {
2787                 global $wgTranscludeCacheExpiry;
2788                 $dbr =& wfGetDB(DB_SLAVE);
2789                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
2790                                 array('tc_url' => $url));
2791                 if ($obj) {
2792                         $time = $obj->tc_time;
2793                         $text = $obj->tc_contents;
2794                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
2795                                 return $text;
2796                         }
2797                 }
2798
2799                 $text = wfGetHTTP($url);
2800                 if (!$text)
2801                         return wfMsg('scarytranscludefailed', $url);
2802
2803                 $dbw =& wfGetDB(DB_MASTER);
2804                 $dbw->replace('transcache', array('tc_url'), array(
2805                         'tc_url' => $url,
2806                         'tc_time' => time(),
2807                         'tc_contents' => $text));
2808                 return $text;
2809         }
2810
2811
2812         /**
2813          * Triple brace replacement -- used for template arguments
2814          * @access private
2815          */
2816         function argSubstitution( $matches ) {
2817                 $arg = trim( $matches['title'] );
2818                 $text = $matches['text'];
2819                 $inputArgs = end( $this->mArgStack );
2820
2821                 if ( array_key_exists( $arg, $inputArgs ) ) {
2822                         $text = $inputArgs[$arg];
2823                 } else if ($this->mOutputType == OT_HTML && null != $matches['parts'] && count($matches['parts']) > 0) {
2824                         $text = $matches['parts'][0];
2825                 }
2826
2827                 return $text;
2828         }
2829
2830         /**
2831          * Returns true if the function is allowed to include this entity
2832          * @access private
2833          */
2834         function incrementIncludeCount( $dbk ) {
2835                 if ( !array_key_exists( $dbk, $this->mIncludeCount ) ) {
2836                         $this->mIncludeCount[$dbk] = 0;
2837                 }
2838                 if ( ++$this->mIncludeCount[$dbk] <= MAX_INCLUDE_REPEAT ) {
2839                         return true;
2840                 } else {
2841                         return false;
2842                 }
2843         }
2844
2845         /**
2846          * This function accomplishes several tasks:
2847          * 1) Auto-number headings if that option is enabled
2848          * 2) Add an [edit] link to sections for logged in users who have enabled the option
2849          * 3) Add a Table of contents on the top for users who have enabled the option
2850          * 4) Auto-anchor headings
2851          *
2852          * It loops through all headlines, collects the necessary data, then splits up the
2853          * string and re-inserts the newly formatted headlines.
2854          *
2855          * @param string $text
2856          * @param boolean $isMain
2857          * @access private
2858          */
2859         function formatHeadings( $text, $isMain=true ) {
2860                 global $wgMaxTocLevel, $wgContLang;
2861
2862                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
2863                 $doShowToc = true;
2864                 $forceTocHere = false;
2865                 if( !$this->mTitle->userCanEdit() ) {
2866                         $showEditLink = 0;
2867                 } else {
2868                         $showEditLink = $this->mOptions->getEditSection();
2869                 }
2870
2871                 # Inhibit editsection links if requested in the page
2872                 $esw =& MagicWord::get( MAG_NOEDITSECTION );
2873                 if( $esw->matchAndRemove( $text ) ) {
2874                         $showEditLink = 0;
2875                 }
2876                 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
2877                 # do not add TOC
2878                 $mw =& MagicWord::get( MAG_NOTOC );
2879                 if( $mw->matchAndRemove( $text ) ) {
2880                         $doShowToc = false;
2881                 }
2882
2883                 # Get all headlines for numbering them and adding funky stuff like [edit]
2884                 # links - this is for later, but we need the number of headlines right now
2885                 $numMatches = preg_match_all( '/<H([1-6])(.*?'.'>)(.*?)<\/H[1-6] *>/i', $text, $matches );
2886
2887                 # if there are fewer than 4 headlines in the article, do not show TOC
2888                 if( $numMatches < 4 ) {
2889                         $doShowToc = false;
2890                 }
2891
2892                 # if the string __TOC__ (not case-sensitive) occurs in the HTML,
2893                 # override above conditions and always show TOC at that place
2894
2895                 $mw =& MagicWord::get( MAG_TOC );
2896                 if($mw->match( $text ) ) {
2897                         $doShowToc = true;
2898                         $forceTocHere = true;
2899                 } else {
2900                         # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
2901                         # override above conditions and always show TOC above first header
2902                         $mw =& MagicWord::get( MAG_FORCETOC );
2903                         if ($mw->matchAndRemove( $text ) ) {
2904                                 $doShowToc = true;
2905                         }
2906                 }
2907
2908                 # Never ever show TOC if no headers
2909                 if( $numMatches < 1 ) {
2910                         $doShowToc = false;
2911                 }
2912
2913                 # We need this to perform operations on the HTML
2914                 $sk =& $this->mOptions->getSkin();
2915
2916                 # headline counter
2917                 $headlineCount = 0;
2918                 $sectionCount = 0; # headlineCount excluding template sections
2919
2920                 # Ugh .. the TOC should have neat indentation levels which can be
2921                 # passed to the skin functions. These are determined here
2922                 $toc = '';
2923                 $full = '';
2924                 $head = array();
2925                 $sublevelCount = array();
2926                 $levelCount = array();
2927                 $toclevel = 0;
2928                 $level = 0;
2929                 $prevlevel = 0;
2930                 $toclevel = 0;
2931                 $prevtoclevel = 0;
2932
2933                 foreach( $matches[3] as $headline ) {
2934                         $istemplate = 0;
2935                         $templatetitle = '';
2936                         $templatesection = 0;
2937                         $numbering = '';
2938
2939                         if (preg_match("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", $headline, $mat)) {
2940                                 $istemplate = 1;
2941                                 $templatetitle = base64_decode($mat[1]);
2942                                 $templatesection = 1 + (int)base64_decode($mat[2]);
2943                                 $headline = preg_replace("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", "", $headline);
2944                         }
2945
2946                         if( $toclevel ) {
2947                                 $prevlevel = $level;
2948                                 $prevtoclevel = $toclevel;
2949                         }
2950                         $level = $matches[1][$headlineCount];
2951
2952                         if( $doNumberHeadings || $doShowToc ) {
2953
2954                                 if ( $level > $prevlevel ) {
2955                                         # Increase TOC level
2956                                         $toclevel++;
2957                                         $sublevelCount[$toclevel] = 0;
2958                                         $toc .= $sk->tocIndent();
2959                                 }
2960                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
2961                                         # Decrease TOC level, find level to jump to
2962
2963                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
2964                                                 # Can only go down to level 1
2965                                                 $toclevel = 1;
2966                                         } else {
2967                                                 for ($i = $toclevel; $i > 0; $i--) {
2968                                                         if ( $levelCount[$i] == $level ) {
2969                                                                 # Found last matching level
2970                                                                 $toclevel = $i;
2971                                                                 break;
2972                                                         }
2973                                                         elseif ( $levelCount[$i] < $level ) {
2974                                                                 # Found first matching level below current level
2975                                                                 $toclevel = $i + 1;
2976                                                                 break;
2977                                                         }
2978                                                 }
2979                                         }
2980
2981                                         $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
2982                                 }
2983                                 else {
2984                                         # No change in level, end TOC line
2985                                         $toc .= $sk->tocLineEnd();
2986                                 }
2987
2988                                 $levelCount[$toclevel] = $level;
2989
2990                                 # count number of headlines for each level
2991                                 @$sublevelCount[$toclevel]++;
2992                                 $dot = 0;
2993                                 for( $i = 1; $i <= $toclevel; $i++ ) {
2994                                         if( !empty( $sublevelCount[$i] ) ) {
2995                                                 if( $dot ) {
2996                                                         $numbering .= '.';
2997                                                 }
2998                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
2999                                                 $dot = 1;
3000                                         }
3001                                 }
3002                         }
3003
3004                         # The canonized header is a version of the header text safe to use for links
3005                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3006                         $canonized_headline = $this->unstrip( $headline, $this->mStripState );
3007                         $canonized_headline = $this->unstripNoWiki( $canonized_headline, $this->mStripState );
3008
3009                         # Remove link placeholders by the link text.
3010                         #     <!--LINK number-->
3011                         # turns into
3012                         #     link text with suffix
3013                         $canonized_headline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
3014                                                             "\$this->mLinkHolders['texts'][\$1]",
3015                                                             $canonized_headline );
3016                         $canonized_headline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
3017                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
3018                                                             $canonized_headline );
3019
3020                         # strip out HTML
3021                         $canonized_headline = preg_replace( '/<.*?' . '>/','',$canonized_headline );
3022                         $tocline = trim( $canonized_headline );
3023                         # Save headline for section edit hint before it's escaped
3024                         $headline_hint = trim( $canonized_headline );
3025                         $canonized_headline = Sanitizer::escapeId( $tocline );
3026                         $refers[$headlineCount] = $canonized_headline;
3027
3028                         # count how many in assoc. array so we can track dupes in anchors
3029                         @$refers[$canonized_headline]++;
3030                         $refcount[$headlineCount]=$refers[$canonized_headline];
3031
3032                         # Don't number the heading if it is the only one (looks silly)
3033                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3034                                 # the two are different if the line contains a link
3035                                 $headline=$numbering . ' ' . $headline;
3036                         }
3037
3038                         # Create the anchor for linking from the TOC to the section
3039                         $anchor = $canonized_headline;
3040                         if($refcount[$headlineCount] > 1 ) {
3041                                 $anchor .= '_' . $refcount[$headlineCount];
3042                         }
3043                         if( $doShowToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3044                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3045                         }
3046                         if( $showEditLink && ( !$istemplate || $templatetitle !== "" ) ) {
3047                                 if ( empty( $head[$headlineCount] ) ) {
3048                                         $head[$headlineCount] = '';
3049                                 }
3050                                 if( $istemplate )
3051                                         $head[$headlineCount] .= $sk->editSectionLinkForOther($templatetitle, $templatesection);
3052                                 else
3053                                         $head[$headlineCount] .= $sk->editSectionLink($this->mTitle, $sectionCount+1, $headline_hint);
3054                         }
3055
3056                         # give headline the correct <h#> tag
3057                         @$head[$headlineCount] .= "<a name=\"$anchor\"></a><h".$level.$matches[2][$headlineCount] .$headline.'</h'.$level.'>';
3058
3059                         $headlineCount++;
3060                         if( !$istemplate )
3061                                 $sectionCount++;
3062                 }
3063
3064                 if( $doShowToc ) {
3065                         $toc .= $sk->tocUnindent( $toclevel - 1 );
3066                         $toc = $sk->tocList( $toc );
3067                 }
3068
3069                 # split up and insert constructed headlines
3070
3071                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3072                 $i = 0;
3073
3074                 foreach( $blocks as $block ) {
3075                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
3076                                 # This is the [edit] link that appears for the top block of text when
3077                                 # section editing is enabled
3078
3079                                 # Disabled because it broke block formatting
3080                                 # For example, a bullet point in the top line
3081                                 # $full .= $sk->editSectionLink(0);
3082                         }
3083                         $full .= $block;
3084                         if( $doShowToc && !$i && $isMain && !$forceTocHere) {
3085                         # Top anchor now in skin
3086                                 $full = $full.$toc;
3087                         }
3088
3089                         if( !empty( $head[$i] ) ) {
3090                                 $full .= $head[$i];
3091                         }
3092                         $i++;
3093                 }
3094                 if($forceTocHere) {
3095                         $mw =& MagicWord::get( MAG_TOC );
3096                         return $mw->replace( $toc, $full );
3097                 } else {
3098                         return $full;
3099                 }
3100         }
3101
3102         /**
3103          * Return an HTML link for the "ISBN 123456" text
3104          * @access private
3105          */
3106         function magicISBN( $text ) {
3107                 $fname = 'Parser::magicISBN';
3108                 wfProfileIn( $fname );
3109
3110                 $a = split( 'ISBN ', ' '.$text );
3111                 if ( count ( $a ) < 2 ) {
3112                         wfProfileOut( $fname );
3113                         return $text;
3114                 }
3115                 $text = substr( array_shift( $a ), 1);
3116                 $valid = '0123456789-Xx';
3117
3118                 foreach ( $a as $x ) {
3119                         $isbn = $blank = '' ;
3120                         while ( ' ' == $x{0} ) {
3121                                 $blank .= ' ';
3122                                 $x = substr( $x, 1 );
3123                         }
3124                         if ( $x == '' ) { # blank isbn
3125                                 $text .= "ISBN $blank";
3126                                 continue;
3127                         }
3128                         while ( strstr( $valid, $x{0} ) != false ) {
3129                                 $isbn .= $x{0};
3130                                 $x = substr( $x, 1 );
3131                         }
3132                         $num = str_replace( '-', '', $isbn );
3133                         $num = str_replace( ' ', '', $num );
3134                         $num = str_replace( 'x', 'X', $num );
3135
3136                         if ( '' == $num ) {
3137                                 $text .= "ISBN $blank$x";
3138                         } else {
3139                                 $titleObj = Title::makeTitle( NS_SPECIAL, 'Booksources' );
3140                                 $text .= '<a href="' .
3141                                 $titleObj->escapeLocalUrl( 'isbn='.$num ) .
3142                                         "\" class=\"internal\">ISBN $isbn</a>";
3143                                 $text .= $x;
3144                         }
3145                 }
3146                 wfProfileOut( $fname );
3147                 return $text;
3148         }
3149
3150         /**
3151          * Return an HTML link for the "RFC 1234" text
3152          *
3153          * @access private
3154          * @param string $text     Text to be processed
3155          * @param string $keyword  Magic keyword to use (default RFC)
3156          * @param string $urlmsg   Interface message to use (default rfcurl)
3157          * @return string
3158          */
3159         function magicRFC( $text, $keyword='RFC ', $urlmsg='rfcurl'  ) {
3160
3161                 $valid = '0123456789';
3162                 $internal = false;
3163
3164                 $a = split( $keyword, ' '.$text );
3165                 if ( count ( $a ) < 2 ) {
3166                         return $text;
3167                 }
3168                 $text = substr( array_shift( $a ), 1);
3169
3170                 /* Check if keyword is preceed by [[.
3171                  * This test is made here cause of the array_shift above
3172                  * that prevent the test to be done in the foreach.
3173                  */
3174                 if ( substr( $text, -2 ) == '[[' ) {
3175                         $internal = true;
3176                 }
3177
3178                 foreach ( $a as $x ) {
3179                         /* token might be empty if we have RFC RFC 1234 */
3180                         if ( $x=='' ) {
3181                                 $text.=$keyword;
3182                                 continue;
3183                                 }
3184
3185                         $id = $blank = '' ;
3186
3187                         /** remove and save whitespaces in $blank */
3188                         while ( $x{0} == ' ' ) {
3189                                 $blank .= ' ';
3190                                 $x = substr( $x, 1 );
3191                         }
3192
3193                         /** remove and save the rfc number in $id */
3194                         while ( strstr( $valid, $x{0} ) != false ) {
3195                                 $id .= $x{0};
3196                                 $x = substr( $x, 1 );
3197                         }
3198
3199                         if ( $id == '' ) {
3200                                 /* call back stripped spaces*/
3201                                 $text .= $keyword.$blank.$x;
3202                         } elseif( $internal ) {
3203                                 /* normal link */
3204                                 $text .= $keyword.$id.$x;
3205                         } else {
3206                                 /* build the external link*/
3207                                 $url = wfMsg( $urlmsg, $id);
3208                                 $sk =& $this->mOptions->getSkin();
3209                                 $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
3210                                 $text .= "<a href='{$url}'{$la}>{$keyword}{$id}</a>{$x}";
3211                         }
3212
3213                         /* Check if the next RFC keyword is preceed by [[ */
3214                         $internal = ( substr($x,-2) == '[[' );
3215                 }
3216                 return $text;
3217         }
3218
3219         /**
3220          * Transform wiki markup when saving a page by doing \r\n -> \n
3221          * conversion, substitting signatures, {{subst:}} templates, etc.
3222          *
3223          * @param string $text the text to transform
3224          * @param Title &$title the Title object for the current article
3225          * @param User &$user the User object describing the current user
3226          * @param ParserOptions $options parsing options
3227          * @param bool $clearState whether to clear the parser state first
3228          * @return string the altered wiki markup
3229          * @access public
3230          */
3231         function preSaveTransform( $text, &$title, &$user, $options, $clearState = true ) {
3232                 $this->mOptions = $options;
3233                 $this->mTitle =& $title;
3234                 $this->mOutputType = OT_WIKI;
3235
3236                 if ( $clearState ) {
3237                         $this->clearState();
3238                 }
3239
3240                 $stripState = false;
3241                 $pairs = array(
3242                         "\r\n" => "\n",
3243                 );
3244                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3245                 $text = $this->strip( $text, $stripState, true );
3246                 $text = $this->pstPass2( $text, $user );
3247                 $text = $this->unstrip( $text, $stripState );
3248                 $text = $this->unstripNoWiki( $text, $stripState );
3249                 return $text;
3250         }
3251
3252         /**
3253          * Pre-save transform helper function
3254          * @access private
3255          */
3256         function pstPass2( $text, &$user ) {
3257                 global $wgContLang, $wgLocaltimezone;
3258
3259                 /* Note: This is the timestamp saved as hardcoded wikitext to
3260                  * the database, we use $wgContLang here in order to give
3261                  * everyone the same signiture and use the default one rather
3262                  * than the one selected in each users preferences.
3263                  */
3264                 if ( isset( $wgLocaltimezone ) ) {
3265                         $oldtz = getenv( 'TZ' );
3266                         putenv( 'TZ='.$wgLocaltimezone );
3267                 }
3268                 $d = $wgContLang->timeanddate( date( 'YmdHis' ), false, false) .
3269                   ' (' . date( 'T' ) . ')';
3270                 if ( isset( $wgLocaltimezone ) ) {
3271                         putenv( 'TZ='.$oldtz );
3272                 }
3273
3274                 # Variable replacement
3275                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3276                 $text = $this->replaceVariables( $text );
3277
3278                 # Signatures
3279                 $sigText = $this->getUserSig( $user );
3280                 $text = strtr( $text, array(
3281                         '~~~~~' => $d,
3282                         '~~~~' => "$sigText $d",
3283                         '~~~' => $sigText
3284                 ) );
3285
3286                 # Context links: [[|name]] and [[name (context)|]]
3287                 #
3288                 global $wgLegalTitleChars;
3289                 $tc = "[$wgLegalTitleChars]";
3290                 $np = str_replace( array( '(', ')' ), array( '', '' ), $tc ); # No parens
3291
3292                 $namespacechar = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
3293                 $conpat = "/^({$np}+) \\(({$tc}+)\\)$/";
3294
3295                 $p1 = "/\[\[({$np}+) \\(({$np}+)\\)\\|]]/";             # [[page (context)|]]
3296                 $p2 = "/\[\[\\|({$tc}+)]]/";                                    # [[|page]]
3297                 $p3 = "/\[\[(:*$namespacechar+):({$np}+)\\|]]/";                # [[namespace:page|]] and [[:namespace:page|]]
3298                 $p4 = "/\[\[(:*$namespacechar+):({$np}+) \\(({$np}+)\\)\\|]]/"; # [[ns:page (cont)|]] and [[:ns:page (cont)|]]
3299                 $context = '';
3300                 $t = $this->mTitle->getText();
3301                 if ( preg_match( $conpat, $t, $m ) ) {
3302                         $context = $m[2];
3303                 }
3304                 $text = preg_replace( $p4, '[[\\1:\\2 (\\3)|\\2]]', $text );
3305                 $text = preg_replace( $p1, '[[\\1 (\\2)|\\1]]', $text );
3306                 $text = preg_replace( $p3, '[[\\1:\\2|\\2]]', $text );
3307
3308                 if ( '' == $context ) {
3309                         $text = preg_replace( $p2, '[[\\1]]', $text );
3310                 } else {
3311                         $text = preg_replace( $p2, "[[\\1 ({$context})|\\1]]", $text );
3312                 }
3313
3314                 # Trim trailing whitespace
3315                 # MAG_END (__END__) tag allows for trailing
3316                 # whitespace to be deliberately included
3317                 $text = rtrim( $text );
3318                 $mw =& MagicWord::get( MAG_END );
3319                 $mw->matchAndRemove( $text );
3320
3321                 return $text;
3322         }
3323
3324         /**
3325          * Fetch the user's signature text, if any, and normalize to
3326          * validated, ready-to-insert wikitext.
3327          *
3328          * @param User $user
3329          * @return string
3330          * @access private
3331          */
3332         function getUserSig( &$user ) {
3333                 global $wgContLang;
3334
3335                 $username = $user->getName();
3336                 $nickname = $user->getOption( 'nickname' );
3337                 $nickname = $nickname === '' ? $username : $nickname;
3338
3339                 if( $user->getBoolOption( 'fancysig' ) !== false ) {
3340                         # Sig. might contain markup; validate this
3341                         if( $this->validateSig( $nickname ) !== false ) {
3342                                 # Validated; clean up (if needed) and return it
3343                                 return( $this->cleanSig( $nickname ) );
3344                         } else {
3345                                 # Failed to validate; fall back to the default
3346                                 $nickname = $username;
3347                                 wfDebug( "Parser::getUserSig: $username has bad XML tags in signature.\n" );
3348                         }
3349                 }
3350
3351                 # If we're still here, make it a link to the user page
3352                 $userpage = $user->getUserPage();
3353                 return( '[[' . $userpage->getPrefixedText() . '|' . wfEscapeWikiText( $nickname ) . ']]' );
3354         }
3355
3356         /**
3357          * Check that the user's signature contains no bad XML
3358          *
3359          * @param string $text
3360          * @return mixed An expanded string, or false if invalid.
3361          */
3362         function validateSig( $text ) {
3363                 return( wfIsWellFormedXmlFragment( $text ) ? $text : false );
3364         }
3365
3366         /**
3367          * Clean up signature text
3368          *
3369          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures
3370          * 2) Substitute all transclusions
3371          *
3372          * @param string $text
3373          * @return string Signature text
3374          */
3375         function cleanSig( $text ) {
3376                 $substWord = MagicWord::get( MAG_SUBST );
3377                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3378                 $substText = '{{' . $substWord->getSynonym( 0 );
3379
3380                 $text = preg_replace( $substRegex, $substText, $text );
3381                 $text = preg_replace( '/~{3,5}/', '', $text );
3382                 $text = $this->replaceVariables( $text );
3383
3384                 return $text;
3385         }
3386
3387         /**
3388          * Set up some variables which are usually set up in parse()
3389          * so that an external function can call some class members with confidence
3390          * @access public
3391          */
3392         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
3393                 $this->mTitle =& $title;
3394                 $this->mOptions = $options;
3395                 $this->mOutputType = $outputType;
3396                 if ( $clearState ) {
3397                         $this->clearState();
3398                 }
3399         }
3400
3401         /**
3402          * Transform a MediaWiki message by replacing magic variables.
3403          *
3404          * @param string $text the text to transform
3405          * @param ParserOptions $options  options
3406          * @return string the text with variables substituted
3407          * @access public
3408          */
3409         function transformMsg( $text, $options ) {
3410                 global $wgTitle;
3411                 static $executing = false;
3412
3413                 $fname = "Parser::transformMsg";
3414
3415                 # Guard against infinite recursion
3416                 if ( $executing ) {
3417                         return $text;
3418                 }
3419                 $executing = true;
3420
3421                 wfProfileIn($fname);
3422
3423                 $this->mTitle = $wgTitle;
3424                 $this->mOptions = $options;
3425                 $this->mOutputType = OT_MSG;
3426                 $this->clearState();
3427                 $text = $this->replaceVariables( $text );
3428
3429                 $executing = false;
3430                 wfProfileOut($fname);
3431                 return $text;
3432         }
3433
3434         /**
3435          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
3436          * The callback should have the following form:
3437          *    function myParserHook( $text, $params, &$parser ) { ... }
3438          *
3439          * Transform and return $text. Use $parser for any required context, e.g. use
3440          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
3441          *
3442          * @access public
3443          *
3444          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
3445          * @param mixed $callback The callback function (and object) to use for the tag
3446          *
3447          * @return The old value of the mTagHooks array associated with the hook
3448          */
3449         function setHook( $tag, $callback ) {
3450                 $oldVal = @$this->mTagHooks[$tag];
3451                 $this->mTagHooks[$tag] = $callback;
3452
3453                 return $oldVal;
3454         }
3455
3456         /**
3457          * Replace <!--LINK--> link placeholders with actual links, in the buffer
3458          * Placeholders created in Skin::makeLinkObj()
3459          * Returns an array of links found, indexed by PDBK:
3460          *  0 - broken
3461          *  1 - normal link
3462          *  2 - stub
3463          * $options is a bit field, RLH_FOR_UPDATE to select for update
3464          */
3465         function replaceLinkHolders( &$text, $options = 0 ) {
3466                 global $wgUser;
3467                 global $wgOutputReplace;
3468
3469                 $fname = 'Parser::replaceLinkHolders';
3470                 wfProfileIn( $fname );
3471
3472                 $pdbks = array();
3473                 $colours = array();
3474                 $sk =& $this->mOptions->getSkin();
3475                 $linkCache =& LinkCache::singleton();
3476
3477                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
3478                         wfProfileIn( $fname.'-check' );
3479                         $dbr =& wfGetDB( DB_SLAVE );
3480                         $page = $dbr->tableName( 'page' );
3481                         $threshold = $wgUser->getOption('stubthreshold');
3482
3483                         # Sort by namespace
3484                         asort( $this->mLinkHolders['namespaces'] );
3485
3486                         # Generate query
3487                         $query = false;
3488                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3489                                 # Make title object
3490                                 $title = $this->mLinkHolders['titles'][$key];
3491
3492                                 # Skip invalid entries.
3493                                 # Result will be ugly, but prevents crash.
3494                                 if ( is_null( $title ) ) {
3495                                         continue;
3496                                 }
3497                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
3498
3499                                 # Check if it's a static known link, e.g. interwiki
3500                                 if ( $title->isAlwaysKnown() ) {
3501                                         $colours[$pdbk] = 1;
3502                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
3503                                         $colours[$pdbk] = 1;
3504                                         $this->mOutput->addLink( $title, $id );
3505                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
3506                                         $colours[$pdbk] = 0;
3507                                 } else {
3508                                         # Not in the link cache, add it to the query
3509                                         if ( !isset( $current ) ) {
3510                                                 $current = $ns;
3511                                                 $query =  "SELECT page_id, page_namespace, page_title";
3512                                                 if ( $threshold > 0 ) {
3513                                                         $query .= ', page_len, page_is_redirect';
3514                                                 }
3515                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
3516                                         } elseif ( $current != $ns ) {
3517                                                 $current = $ns;
3518                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
3519                                         } else {
3520                                                 $query .= ', ';
3521                                         }
3522
3523                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
3524                                 }
3525                         }
3526                         if ( $query ) {
3527                                 $query .= '))';
3528                                 if ( $options & RLH_FOR_UPDATE ) {
3529                                         $query .= ' FOR UPDATE';
3530                                 }
3531
3532                                 $res = $dbr->query( $query, $fname );
3533
3534                                 # Fetch data and form into an associative array
3535                                 # non-existent = broken
3536                                 # 1 = known
3537                                 # 2 = stub
3538                                 while ( $s = $dbr->fetchObject($res) ) {
3539                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
3540                                         $pdbk = $title->getPrefixedDBkey();
3541                                         $linkCache->addGoodLinkObj( $s->page_id, $title );
3542                                         $this->mOutput->addLink( $title, $s->page_id );
3543
3544                                         if ( $threshold >  0 ) {
3545                                                 $size = $s->page_len;
3546                                                 if ( $s->page_is_redirect || $s->page_namespace != 0 || $size >= $threshold ) {
3547                                                         $colours[$pdbk] = 1;
3548                                                 } else {
3549                                                         $colours[$pdbk] = 2;
3550                                                 }
3551                                         } else {
3552                                                 $colours[$pdbk] = 1;
3553                                         }
3554                                 }
3555                         }
3556                         wfProfileOut( $fname.'-check' );
3557
3558                         # Construct search and replace arrays
3559                         wfProfileIn( $fname.'-construct' );
3560                         $wgOutputReplace = array();
3561                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3562                                 $pdbk = $pdbks[$key];
3563                                 $searchkey = "<!--LINK $key-->";
3564                                 $title = $this->mLinkHolders['titles'][$key];
3565                                 if ( empty( $colours[$pdbk] ) ) {
3566                                         $linkCache->addBadLinkObj( $title );
3567                                         $colours[$pdbk] = 0;
3568                                         $this->mOutput->addLink( $title, 0 );
3569                                         $wgOutputReplace[$searchkey] = $sk->makeBrokenLinkObj( $title,
3570                                                                         $this->mLinkHolders['texts'][$key],
3571                                                                         $this->mLinkHolders['queries'][$key] );
3572                                 } elseif ( $colours[$pdbk] == 1 ) {
3573                                         $wgOutputReplace[$searchkey] = $sk->makeKnownLinkObj( $title,
3574                                                                         $this->mLinkHolders['texts'][$key],
3575                                                                         $this->mLinkHolders['queries'][$key] );
3576                                 } elseif ( $colours[$pdbk] == 2 ) {
3577                                         $wgOutputReplace[$searchkey] = $sk->makeStubLinkObj( $title,
3578                                                                         $this->mLinkHolders['texts'][$key],
3579                                                                         $this->mLinkHolders['queries'][$key] );
3580                                 }
3581                         }
3582                         wfProfileOut( $fname.'-construct' );
3583
3584                         # Do the thing
3585                         wfProfileIn( $fname.'-replace' );
3586
3587                         $text = preg_replace_callback(
3588                                 '/(<!--LINK .*?-->)/',
3589                                 "wfOutputReplaceMatches",
3590                                 $text);
3591
3592                         wfProfileOut( $fname.'-replace' );
3593                 }
3594
3595                 # Now process interwiki link holders
3596                 # This is quite a bit simpler than internal links
3597                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
3598                         wfProfileIn( $fname.'-interwiki' );
3599                         # Make interwiki link HTML
3600                         $wgOutputReplace = array();
3601                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
3602                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
3603                                 $wgOutputReplace[$key] = $sk->makeLinkObj( $title, $link );
3604                         }
3605
3606                         $text = preg_replace_callback(
3607                                 '/<!--IWLINK (.*?)-->/',
3608                                 "wfOutputReplaceMatches",
3609                                 $text );
3610                         wfProfileOut( $fname.'-interwiki' );
3611                 }
3612
3613                 wfProfileOut( $fname );
3614                 return $colours;
3615         }
3616
3617         /**
3618          * Replace <!--LINK--> link placeholders with plain text of links
3619          * (not HTML-formatted).
3620          * @param string $text
3621          * @return string
3622          */
3623         function replaceLinkHoldersText( $text ) {
3624                 global $wgUser;
3625                 global $wgOutputReplace;
3626
3627                 $fname = 'Parser::replaceLinkHoldersText';
3628                 wfProfileIn( $fname );
3629
3630                 $text = preg_replace_callback(
3631                         '/<!--(LINK|IWLINK) (.*?)-->/',
3632                         array( &$this, 'replaceLinkHoldersTextCallback' ),
3633                         $text );
3634
3635                 wfProfileOut( $fname );
3636                 return $text;
3637         }
3638
3639         /**
3640          * @param array $matches
3641          * @return string
3642          * @access private
3643          */
3644         function replaceLinkHoldersTextCallback( $matches ) {
3645                 $type = $matches[1];
3646                 $key  = $matches[2];
3647                 if( $type == 'LINK' ) {
3648                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
3649                                 return $this->mLinkHolders['texts'][$key];
3650                         }
3651                 } elseif( $type == 'IWLINK' ) {
3652                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
3653                                 return $this->mInterwikiLinkHolders['texts'][$key];
3654                         }
3655                 }
3656                 return $matches[0];
3657         }
3658
3659         /**
3660          * Renders an image gallery from a text with one line per image.
3661          * text labels may be given by using |-style alternative text. E.g.
3662          *   Image:one.jpg|The number "1"
3663          *   Image:tree.jpg|A tree
3664          * given as text will return the HTML of a gallery with two images,
3665          * labeled 'The number "1"' and
3666          * 'A tree'.
3667          */
3668         function renderImageGallery( $text ) {
3669                 # Setup the parser
3670                 $parserOptions = new ParserOptions;
3671                 $localParser = new Parser();
3672
3673                 $ig = new ImageGallery();
3674                 $ig->setShowBytes( false );
3675                 $ig->setShowFilename( false );
3676                 $lines = explode( "\n", $text );
3677
3678                 foreach ( $lines as $line ) {
3679                         # match lines like these:
3680                         # Image:someimage.jpg|This is some image
3681                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
3682                         # Skip empty lines
3683                         if ( count( $matches ) == 0 ) {
3684                                 continue;
3685                         }
3686                         $nt =& Title::newFromText( $matches[1] );
3687                         if( is_null( $nt ) ) {
3688                                 # Bogus title. Ignore these so we don't bomb out later.
3689                                 continue;
3690                         }
3691                         if ( isset( $matches[3] ) ) {
3692                                 $label = $matches[3];
3693                         } else {
3694                                 $label = '';
3695                         }
3696
3697                         $pout = $localParser->parse( $label , $this->mTitle, $parserOptions );
3698                         $html = $pout->getText();
3699
3700                         $ig->add( new Image( $nt ), $html );
3701                         $this->mOutput->addImage( $nt->getDBkey() );
3702                 }
3703                 return $ig->toHTML();
3704         }
3705
3706         /**
3707          * Parse image options text and use it to make an image
3708          */
3709         function makeImage( &$nt, $options ) {
3710                 global $wgContLang, $wgUseImageResize, $wgUser;
3711
3712                 $align = '';
3713
3714                 # Check if the options text is of the form "options|alt text"
3715                 # Options are:
3716                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
3717                 #  * left               no resizing, just left align. label is used for alt= only
3718                 #  * right              same, but right aligned
3719                 #  * none               same, but not aligned
3720                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
3721                 #  * center             center the image
3722                 #  * framed             Keep original image size, no magnify-button.
3723
3724                 $part = explode( '|', $options);
3725
3726                 $mwThumb  =& MagicWord::get( MAG_IMG_THUMBNAIL );
3727                 $mwManualThumb =& MagicWord::get( MAG_IMG_MANUALTHUMB );
3728                 $mwLeft   =& MagicWord::get( MAG_IMG_LEFT );
3729                 $mwRight  =& MagicWord::get( MAG_IMG_RIGHT );
3730                 $mwNone   =& MagicWord::get( MAG_IMG_NONE );
3731                 $mwWidth  =& MagicWord::get( MAG_IMG_WIDTH );
3732                 $mwCenter =& MagicWord::get( MAG_IMG_CENTER );
3733                 $mwFramed =& MagicWord::get( MAG_IMG_FRAMED );
3734                 $caption = '';
3735
3736                 $width = $height = $framed = $thumb = false;
3737                 $manual_thumb = '' ;
3738
3739                 foreach( $part as $key => $val ) {
3740                         if ( $wgUseImageResize && ! is_null( $mwThumb->matchVariableStartToEnd($val) ) ) {
3741                                 $thumb=true;
3742                         } elseif ( ! is_null( $match = $mwManualThumb->matchVariableStartToEnd($val) ) ) {
3743                                 # use manually specified thumbnail
3744                                 $thumb=true;
3745                                 $manual_thumb = $match;
3746                         } elseif ( ! is_null( $mwRight->matchVariableStartToEnd($val) ) ) {
3747                                 # remember to set an alignment, don't render immediately
3748                                 $align = 'right';
3749                         } elseif ( ! is_null( $mwLeft->matchVariableStartToEnd($val) ) ) {
3750                                 # remember to set an alignment, don't render immediately
3751                                 $align = 'left';
3752                         } elseif ( ! is_null( $mwCenter->matchVariableStartToEnd($val) ) ) {
3753                                 # remember to set an alignment, don't render immediately
3754                                 $align = 'center';
3755                         } elseif ( ! is_null( $mwNone->matchVariableStartToEnd($val) ) ) {
3756                                 # remember to set an alignment, don't render immediately
3757                                 $align = 'none';
3758                         } elseif ( $wgUseImageResize && ! is_null( $match = $mwWidth->matchVariableStartToEnd($val) ) ) {
3759                                 wfDebug( "MAG_IMG_WIDTH match: $match\n" );
3760                                 # $match is the image width in pixels
3761                                 if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $match, $m ) ) {
3762                                         $width = intval( $m[1] );
3763                                         $height = intval( $m[2] );
3764                                 } else {
3765                                         $width = intval($match);
3766                                 }
3767                         } elseif ( ! is_null( $mwFramed->matchVariableStartToEnd($val) ) ) {
3768                                 $framed=true;
3769                         } else {
3770                                 $caption = $val;
3771                         }
3772                 }
3773                 # Strip bad stuff out of the alt text
3774                 $alt = $this->replaceLinkHoldersText( $caption );
3775                 $alt = Sanitizer::stripAllTags( $alt );
3776
3777                 # Linker does the rest
3778                 $sk =& $this->mOptions->getSkin();
3779                 return $sk->makeImageLinkObj( $nt, $caption, $alt, $align, $width, $height, $framed, $thumb, $manual_thumb );
3780         }
3781
3782         /**
3783          * Set a flag in the output object indicating that the content is dynamic and
3784          * shouldn't be cached.
3785          */
3786         function disableCache() {
3787                 $this->mOutput->mCacheTime = -1;
3788         }
3789
3790         /**#@+
3791          * Callback from the Sanitizer for expanding items found in HTML attribute
3792          * values, so they can be safely tested and escaped.
3793          * @param string $text
3794          * @param array $args
3795          * @return string
3796          * @access private
3797          */
3798         function attributeStripCallback( &$text, $args ) {
3799                 $text = $this->replaceVariables( $text, $args );
3800                 $text = $this->unstripForHTML( $text );
3801                 return $text;
3802         }
3803
3804         function unstripForHTML( $text ) {
3805                 $text = $this->unstrip( $text, $this->mStripState );
3806                 $text = $this->unstripNoWiki( $text, $this->mStripState );
3807                 return $text;
3808         }
3809         /**#@-*/
3810
3811         /**#@+
3812          * Accessor/mutator
3813          */
3814         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
3815         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
3816         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
3817         /**#@-*/
3818
3819         /**#@+
3820          * Accessor
3821          */
3822         function getTags() { return array_keys( $this->mTagHooks ); }
3823         /**#@-*/
3824 }
3825
3826 /**
3827  * @todo document
3828  * @package MediaWiki
3829  */
3830 class ParserOutput
3831 {
3832         var $mText,             # The output text
3833                 $mLanguageLinks,    # List of the full text of language links, in the order they appear
3834                 $mCategories,       # Map of category names to sort keys
3835                 $mContainsOldMagic, # Boolean variable indicating if the input contained variables like {{CURRENTDAY}}
3836                 $mCacheTime,        # Timestamp on this article, or -1 for uncacheable. Used in ParserCache.
3837                 $mVersion,          # Compatibility check
3838                 $mTitleText,        # title text of the chosen language variant
3839                 $mLinks,            # 2-D map of NS/DBK to ID for the links in the document. ID=zero for broken.
3840                 $mTemplates,        # 2-D map of NS/DBK to ID for the template references. ID=zero for broken.
3841                 $mImages,           # DB keys of the images used, in the array key only
3842                 $mExternalLinks;    # External link URLs, in the key only
3843
3844         function ParserOutput( $text = '', $languageLinks = array(), $categoryLinks = array(),
3845                 $containsOldMagic = false, $titletext = '' )
3846         {
3847                 $this->mText = $text;
3848                 $this->mLanguageLinks = $languageLinks;
3849                 $this->mCategories = $categoryLinks;
3850                 $this->mContainsOldMagic = $containsOldMagic;
3851                 $this->mCacheTime = '';
3852                 $this->mVersion = MW_PARSER_VERSION;
3853                 $this->mTitleText = $titletext;
3854                 $this->mLinks = array();
3855                 $this->mTemplates = array();
3856                 $this->mImages = array();
3857                 $this->mExternalLinks = array();
3858         }
3859
3860         function getText()                   { return $this->mText; }
3861         function getLanguageLinks()          { return $this->mLanguageLinks; }
3862         function getCategoryLinks()          { return array_keys( $this->mCategories ); }
3863         function &getCategories()            { return $this->mCategories; }
3864         function getCacheTime()              { return $this->mCacheTime; }
3865         function getTitleText()              { return $this->mTitleText; }
3866         function &getLinks()                 { return $this->mLinks; }
3867         function &getTemplates()             { return $this->mTemplates; }
3868         function &getImages()                { return $this->mImages; }
3869         function &getExternalLinks()         { return $this->mExternalLinks; }
3870
3871         function containsOldMagic()          { return $this->mContainsOldMagic; }
3872         function setText( $text )            { return wfSetVar( $this->mText, $text ); }
3873         function setLanguageLinks( $ll )     { return wfSetVar( $this->mLanguageLinks, $ll ); }
3874         function setCategoryLinks( $cl )     { return wfSetVar( $this->mCategories, $cl ); }
3875         function setContainsOldMagic( $com ) { return wfSetVar( $this->mContainsOldMagic, $com ); }
3876         function setCacheTime( $t )          { return wfSetVar( $this->mCacheTime, $t ); }
3877         function setTitleText( $t )          { return wfSetVar ($this->mTitleText, $t); }
3878
3879         function addCategory( $c, $sort )    { $this->mCategories[$c] = $sort; }
3880         function addImage( $name )           { $this->mImages[$name] = 1; }
3881         function addLanguageLink( $t )       { $this->mLanguageLinks[] = $t; }
3882         function addExternalLink( $url )     { $this->mExternalLinks[$url] = 1; }
3883
3884         function addLink( $title, $id ) {
3885                 $ns = $title->getNamespace();
3886                 $dbk = $title->getDBkey();
3887                 if ( !isset( $this->mLinks[$ns] ) ) {
3888                         $this->mLinks[$ns] = array();
3889                 }
3890                 $this->mLinks[$ns][$dbk] = $id;
3891         }
3892
3893         function addTemplate( $title, $id ) {
3894                 $ns = $title->getNamespace();
3895                 $dbk = $title->getDBkey();
3896                 if ( !isset( $this->mTemplates[$ns] ) ) {
3897                         $this->mTemplates[$ns] = array();
3898                 }
3899                 $this->mTemplates[$ns][$dbk] = $id;
3900         }
3901
3902         /**
3903          * @deprecated
3904          */
3905         /*
3906         function merge( $other ) {
3907                 $this->mLanguageLinks = array_merge( $this->mLanguageLinks, $other->mLanguageLinks );
3908                 $this->mCategories = array_merge( $this->mCategories, $this->mLanguageLinks );
3909                 $this->mContainsOldMagic = $this->mContainsOldMagic || $other->mContainsOldMagic;
3910         }*/
3911
3912         /**
3913          * Return true if this cached output object predates the global or
3914          * per-article cache invalidation timestamps, or if it comes from
3915          * an incompatible older version.
3916          *
3917          * @param string $touched the affected article's last touched timestamp
3918          * @return bool
3919          * @access public
3920          */
3921         function expired( $touched ) {
3922                 global $wgCacheEpoch;
3923                 return $this->getCacheTime() == -1 || // parser says it's uncacheable
3924                        $this->getCacheTime() < $touched ||
3925                        $this->getCacheTime() <= $wgCacheEpoch ||
3926                        !isset( $this->mVersion ) ||
3927                        version_compare( $this->mVersion, MW_PARSER_VERSION, "lt" );
3928         }
3929 }
3930
3931 /**
3932  * Set options of the Parser
3933  * @todo document
3934  * @package MediaWiki
3935  */
3936 class ParserOptions
3937 {
3938         # All variables are private
3939         var $mUseTeX;                    # Use texvc to expand <math> tags
3940         var $mUseDynamicDates;           # Use DateFormatter to format dates
3941         var $mInterwikiMagic;            # Interlanguage links are removed and returned in an array
3942         var $mAllowExternalImages;       # Allow external images inline
3943         var $mAllowExternalImagesFrom;   # If not, any exception?
3944         var $mSkin;                      # Reference to the preferred skin
3945         var $mDateFormat;                # Date format index
3946         var $mEditSection;               # Create "edit section" links
3947         var $mNumberHeadings;            # Automatically number headings
3948         var $mAllowSpecialInclusion;     # Allow inclusion of special pages
3949         var $mTidy;                      # Ask for tidy cleanup
3950
3951         function getUseTeX()                        { return $this->mUseTeX; }
3952         function getUseDynamicDates()               { return $this->mUseDynamicDates; }
3953         function getInterwikiMagic()                { return $this->mInterwikiMagic; }
3954         function getAllowExternalImages()           { return $this->mAllowExternalImages; }
3955         function getAllowExternalImagesFrom()       { return $this->mAllowExternalImagesFrom; }
3956         function &getSkin()                         { return $this->mSkin; }
3957         function getDateFormat()                    { return $this->mDateFormat; }
3958         function getEditSection()                   { return $this->mEditSection; }
3959         function getNumberHeadings()                { return $this->mNumberHeadings; }
3960         function getAllowSpecialInclusion()         { return $this->mAllowSpecialInclusion; }
3961         function getTidy()                          { return $this->mTidy; }
3962
3963         function setUseTeX( $x )                    { return wfSetVar( $this->mUseTeX, $x ); }
3964         function setUseDynamicDates( $x )           { return wfSetVar( $this->mUseDynamicDates, $x ); }
3965         function setInterwikiMagic( $x )            { return wfSetVar( $this->mInterwikiMagic, $x ); }
3966         function setAllowExternalImages( $x )       { return wfSetVar( $this->mAllowExternalImages, $x ); }
3967         function setAllowExternalImagesFrom( $x )   { return wfSetVar( $this->mAllowExternalImagesFrom, $x ); }
3968         function setDateFormat( $x )                { return wfSetVar( $this->mDateFormat, $x ); }
3969         function setEditSection( $x )               { return wfSetVar( $this->mEditSection, $x ); }
3970         function setNumberHeadings( $x )            { return wfSetVar( $this->mNumberHeadings, $x ); }
3971         function setAllowSpecialInclusion( $x )     { return wfSetVar( $this->mAllowSpecialInclusion, $x ); }
3972         function setTidy( $x )                      { return wfSetVar( $this->mTidy, $x); }
3973         function setSkin( &$x ) { $this->mSkin =& $x; }
3974
3975         function ParserOptions() {
3976                 global $wgUser;
3977                 $this->initialiseFromUser( $wgUser );
3978         }
3979
3980         /**
3981          * Get parser options
3982          * @static
3983          */
3984         function newFromUser( &$user ) {
3985                 $popts = new ParserOptions;
3986                 $popts->initialiseFromUser( $user );
3987                 return $popts;
3988         }
3989
3990         /** Get user options */
3991         function initialiseFromUser( &$userInput ) {
3992                 global $wgUseTeX, $wgUseDynamicDates, $wgInterwikiMagic, $wgAllowExternalImages,
3993                        $wgAllowExternalImagesFrom, $wgAllowSpecialInclusion;
3994                 $fname = 'ParserOptions::initialiseFromUser';
3995                 wfProfileIn( $fname );
3996                 if ( !$userInput ) {
3997                         $user = new User;
3998                         $user->setLoaded( true );
3999                 } else {
4000                         $user =& $userInput;
4001                 }
4002
4003                 $this->mUseTeX = $wgUseTeX;
4004                 $this->mUseDynamicDates = $wgUseDynamicDates;
4005                 $this->mInterwikiMagic = $wgInterwikiMagic;
4006                 $this->mAllowExternalImages = $wgAllowExternalImages;
4007                 $this->mAllowExternalImagesFrom = $wgAllowExternalImagesFrom;
4008                 wfProfileIn( $fname.'-skin' );
4009                 $this->mSkin =& $user->getSkin();
4010                 wfProfileOut( $fname.'-skin' );
4011                 $this->mDateFormat = $user->getOption( 'date' );
4012                 $this->mEditSection = true;
4013                 $this->mNumberHeadings = $user->getOption( 'numberheadings' );
4014                 $this->mAllowSpecialInclusion = $wgAllowSpecialInclusion;
4015                 $this->mTidy = false;
4016                 wfProfileOut( $fname );
4017         }
4018 }
4019
4020 /**
4021  * Callback function used by Parser::replaceLinkHolders()
4022  * to substitute link placeholders.
4023  */
4024 function &wfOutputReplaceMatches( $matches ) {
4025         global $wgOutputReplace;
4026         return $wgOutputReplace[$matches[1]];
4027 }
4028
4029 /**
4030  * Return the total number of articles
4031  */
4032 function wfNumberOfArticles() {
4033         global $wgNumberOfArticles;
4034
4035         wfLoadSiteStats();
4036         return $wgNumberOfArticles;
4037 }
4038
4039 /**
4040  * Return the number of files
4041  */
4042 function wfNumberOfFiles() {
4043         $fname = 'Parser::wfNumberOfFiles';
4044
4045         wfProfileIn( $fname );
4046         $dbr =& wfGetDB( DB_SLAVE );
4047         $res = $dbr->selectField('image', 'COUNT(*)', array(), $fname );
4048         wfProfileOut( $fname );
4049
4050         return $res;
4051 }
4052
4053 /**
4054  * Get various statistics from the database
4055  * @access private
4056  */
4057 function wfLoadSiteStats() {
4058         global $wgNumberOfArticles, $wgTotalViews, $wgTotalEdits;
4059         $fname = 'wfLoadSiteStats';
4060
4061         if ( -1 != $wgNumberOfArticles ) return;
4062         $dbr =& wfGetDB( DB_SLAVE );
4063         $s = $dbr->selectRow( 'site_stats',
4064                 array( 'ss_total_views', 'ss_total_edits', 'ss_good_articles' ),
4065                 array( 'ss_row_id' => 1 ), $fname
4066         );
4067
4068         if ( $s === false ) {
4069                 return;
4070         } else {
4071                 $wgTotalViews = $s->ss_total_views;
4072                 $wgTotalEdits = $s->ss_total_edits;
4073                 $wgNumberOfArticles = $s->ss_good_articles;
4074         }
4075 }
4076
4077 /**
4078  * Escape html tags
4079  * Basically replacing " > and < with HTML entities ( &quot;, &gt;, &lt;)
4080  *
4081  * @param string $in Text that might contain HTML tags
4082  * @return string Escaped string
4083  */
4084 function wfEscapeHTMLTagsOnly( $in ) {
4085         return str_replace(
4086                 array( '"', '>', '<' ),
4087                 array( '&quot;', '&gt;', '&lt;' ),
4088                 $in );
4089 }
4090
4091 ?>