includes/Parser.php

   1 <?php
   2 /**
   3  * File for Parser and related classes
   4  *
   5  * @package MediaWiki
   6  * @subpackage Parser
   7  */
   8
   9 /** */
  10 require_once( 'Sanitizer.php' );
  11 require_once( 'HttpFunctions.php' );
  12
  13 /**
  14  * Update this version number when the ParserOutput format
  15  * changes in an incompatible way, so the parser cache
  16  * can automatically discard old data.
  17  */
  18 define( 'MW_PARSER_VERSION', '1.6.0' );
  19
  20 /**
  21  * Variable substitution O(N^2) attack
  22  *
  23  * Without countermeasures, it would be possible to attack the parser by saving
  24  * a page filled with a large number of inclusions of large pages. The size of
  25  * the generated page would be proportional to the square of the input size.
  26  * Hence, we limit the number of inclusions of any given page, thus bringing any
  27  * attack back to O(N).
  28  */
  29
  30 define( 'MAX_INCLUDE_REPEAT', 100 );
  31 define( 'MAX_INCLUDE_SIZE', 1000000 ); // 1 Million
  32
  33 define( 'RLH_FOR_UPDATE', 1 );
  34
  35 # Allowed values for $mOutputType
  36 define( 'OT_HTML', 1 );
  37 define( 'OT_WIKI', 2 );
  38 define( 'OT_MSG' , 3 );
  39
  40 # string parameter for extractTags which will cause it
  41 # to strip HTML comments in addition to regular
  42 # <XML>-style tags. This should not be anything we
  43 # may want to use in wikisyntax
  44 define( 'STRIP_COMMENTS', 'HTMLCommentStrip' );
  45
  46 # Constants needed for external link processing
  47 define( 'HTTP_PROTOCOLS', 'http:\/\/|https:\/\/' );
  48 # Everything except bracket, space, or control characters
  49 define( 'EXT_LINK_URL_CLASS', '[^][<>"\\x00-\\x20\\x7F]' );
  50 # Including space
  51 define( 'EXT_LINK_TEXT_CLASS', '[^\]\\x00-\\x1F\\x7F]' );
  52 define( 'EXT_IMAGE_FNAME_CLASS', '[A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]' );
  53 define( 'EXT_IMAGE_EXTENSIONS', 'gif|png|jpg|jpeg' );
  54 define( 'EXT_LINK_BRACKETED',  '/\[(\b(' . wfUrlProtocols() . ')'.EXT_LINK_URL_CLASS.'+) *('.EXT_LINK_TEXT_CLASS.'*?)\]/S' );
  55 define( 'EXT_IMAGE_REGEX',
  56         '/^('.HTTP_PROTOCOLS.')'.  # Protocol
  57         '('.EXT_LINK_URL_CLASS.'+)\\/'.  # Hostname and path
  58         '('.EXT_IMAGE_FNAME_CLASS.'+)\\.((?i)'.EXT_IMAGE_EXTENSIONS.')$/S' # Filename
  59 );
  60
  61 /**
  62  * PHP Parser
  63  *
  64  * Processes wiki markup
  65  *
  66  * <pre>
  67  * There are three main entry points into the Parser class:
  68  * parse()
  69  *   produces HTML output
  70  * preSaveTransform().
  71  *   produces altered wiki markup.
  72  * transformMsg()
  73  *   performs brace substitution on MediaWiki messages
  74  *
  75  * Globals used:
  76  *    objects:   $wgLang
  77  *
  78  * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
  79  *
  80  * settings:
  81  *  $wgUseTex*, $wgUseDynamicDates*, $wgInterwikiMagic*,
  82  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  83  *  $wgLocaltimezone, $wgAllowSpecialInclusion*
  84  *
  85  *  * only within ParserOptions
  86  * </pre>
  87  *
  88  * @package MediaWiki
  89  */
  90 class Parser
  91 {
  92         /**#@+
  93          * @access private
  94          */
  95         # Persistent:
  96         var $mTagHooks;
  97
  98         # Cleared with clearState():
  99         var $mOutput, $mAutonumber, $mDTopen, $mStripState = array();
 100         var $mVariables, $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 101         var $mInterwikiLinkHolders, $mLinkHolders, $mUniqPrefix;
 102         var $mTemplates,        // cache of already loaded templates, avoids
 103                                 // multiple SQL queries for the same string
 104             $mTemplatePath;     // stores an unsorted hash of all the templates already loaded
 105                                 // in this path. Used for loop detection.
 106
 107         # Temporary
 108         # These are variables reset at least once per parse regardless of $clearState
 109         var $mOptions,      // ParserOptions object
 110                 $mTitle,        // Title context, used for self-link rendering and similar things
 111                 $mOutputType,   // Output type, one of the OT_xxx constants
 112                 $mRevisionId;   // ID to display in {{REVISIONID}} tags
 113
 114         /**#@-*/
 115
 116         /**
 117          * Constructor
 118          *
 119          * @access public
 120          */
 121         function Parser() {
 122                 $this->mTagHooks = array();
 123                 $this->clearState();
 124         }
 125
 126         /**
 127          * Clear Parser state
 128          *
 129          * @access private
 130          */
 131         function clearState() {
 132                 $this->mOutput = new ParserOutput;
 133                 $this->mAutonumber = 0;
 134                 $this->mLastSection = '';
 135                 $this->mDTopen = false;
 136                 $this->mVariables = false;
 137                 $this->mIncludeCount = array();
 138                 $this->mStripState = array();
 139                 $this->mArgStack = array();
 140                 $this->mInPre = false;
 141                 $this->mInterwikiLinkHolders = array(
 142                         'texts' => array(),
 143                         'titles' => array()
 144                 );
 145                 $this->mLinkHolders = array(
 146                         'namespaces' => array(),
 147                         'dbkeys' => array(),
 148                         'queries' => array(),
 149                         'texts' => array(),
 150                         'titles' => array()
 151                 );
 152                 $this->mRevisionId = null;
 153                 $this->mUniqPrefix = 'UNIQ' . Parser::getRandomString();
 154
 155                 # Clear these on every parse, bug 4549
 156                 $this->mTemplates = array();
 157                 $this->mTemplatePath = array();
 158
 159                 wfRunHooks( 'ParserClearState', array( &$this ) );
 160         }
 161
 162         /**
 163          * Accessor for mUniqPrefix.
 164          *
 165          * @access public
 166          */
 167         function UniqPrefix() {
 168                 return $this->mUniqPrefix;
 169         }
 170
 171         /**
 172          * Convert wikitext to HTML
 173          * Do not call this function recursively.
 174          *
 175          * @access private
 176          * @param string $text Text we want to parse
 177          * @param Title &$title A title object
 178          * @param array $options
 179          * @param boolean $linestart
 180          * @param boolean $clearState
 181          * @param int $revid number to pass in {{REVISIONID}}
 182          * @return ParserOutput a ParserOutput
 183          */
 184         function parse( $text, &$title, $options, $linestart = true, $clearState = true, $revid = null ) {
 185                 /**
 186                  * First pass--just handle <nowiki> sections, pass the rest off
 187                  * to internalParse() which does all the real work.
 188                  */
 189
 190                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
 191                 $fname = 'Parser::parse';
 192                 wfProfileIn( $fname );
 193
 194                 if ( $clearState ) {
 195                         $this->clearState();
 196                 }
 197
 198                 $this->mOptions = $options;
 199                 $this->mTitle =& $title;
 200                 $this->mRevisionId = $revid;
 201                 $this->mOutputType = OT_HTML;
 202
 203                 $this->mStripState = NULL;
 204
 205                 //$text = $this->strip( $text, $this->mStripState );
 206                 // VOODOO MAGIC FIX! Sometimes the above segfaults in PHP5.
 207                 $x =& $this->mStripState;
 208
 209                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$x ) );
 210                 $text = $this->strip( $text, $x );
 211                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$x ) );
 212
 213                 # Hook to suspend the parser in this state
 214                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$x ) ) ) {
 215                         wfProfileOut( $fname );
 216                         return $text ;
 217                 }
 218
 219                 $text = $this->internalParse( $text );
 220
 221                 $text = $this->unstrip( $text, $this->mStripState );
 222
 223                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 224                 $fixtags = array(
 225                         # french spaces, last one Guillemet-left
 226                         # only if there is something before the space
 227                         '/(.) (?=\\?|:|;|!|\\302\\273)/' => '\\1&nbsp;\\2',
 228                         # french spaces, Guillemet-right
 229                         '/(\\302\\253) /' => '\\1&nbsp;',
 230                         '/<center *>(.*)<\\/center *>/i' => '<div class="center">\\1</div>',
 231                 );
 232                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 233
 234                 # only once and last
 235                 $text = $this->doBlockLevels( $text, $linestart );
 236
 237                 $this->replaceLinkHolders( $text );
 238
 239                 # the position of the convert() call should not be changed. it
 240                 # assumes that the links are all replaces and the only thing left
 241                 # is the <nowiki> mark.
 242                 $text = $wgContLang->convert($text);
 243                 $this->mOutput->setTitleText($wgContLang->getParsedTitle());
 244
 245                 $text = $this->unstripNoWiki( $text, $this->mStripState );
 246
 247                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 248
 249                 $text = Sanitizer::normalizeCharReferences( $text );
 250
 251                 if (($wgUseTidy and $this->mOptions->mTidy) or $wgAlwaysUseTidy) {
 252                         $text = Parser::tidy($text);
 253                 }
 254
 255                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 256
 257                 $this->mOutput->setText( $text );
 258                 wfProfileOut( $fname );
 259
 260                 return $this->mOutput;
 261         }
 262
 263         /**
 264          * Get a random string
 265          *
 266          * @access private
 267          * @static
 268          */
 269         function getRandomString() {
 270                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 271         }
 272
 273         /**
 274          * Replaces all occurrences of <$tag>content</$tag> in the text
 275          * with a random marker and returns the new text. the output parameter
 276          * $content will be an associative array filled with data on the form
 277          * $unique_marker => content.
 278          *
 279          * If $content is already set, the additional entries will be appended
 280          * If $tag is set to STRIP_COMMENTS, the function will extract
 281          * <!-- HTML comments -->
 282          *
 283          * @access private
 284          * @static
 285          */
 286         function extractTagsAndParams($tag, $text, &$content, &$tags, &$params, $uniq_prefix = ''){
 287                 $rnd = $uniq_prefix . '-' . $tag . Parser::getRandomString();
 288                 if ( !$content ) {
 289                         $content = array( );
 290                 }
 291                 $n = 1;
 292                 $stripped = '';
 293
 294                 if ( !$tags ) {
 295                         $tags = array( );
 296                 }
 297
 298                 if ( !$params ) {
 299                         $params = array( );
 300                 }
 301
 302                 if( $tag == STRIP_COMMENTS ) {
 303                         $start = '/<!--()()/';
 304                         $end   = '/-->/';
 305                 } else {
 306                         $start = "/<$tag(\\s+[^\\/>]*|\\s*)(\\/?)>/i";
 307                         $end   = "/<\\/$tag\\s*>/i";
 308                 }
 309
 310                 while ( '' != $text ) {
 311                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 312                         $stripped .= $p[0];
 313                         if( count( $p ) < 4 ) {
 314                                 break;
 315                         }
 316                         $attributes = $p[1];
 317                         $empty      = $p[2];
 318                         $inside     = $p[3];
 319
 320                         $marker = $rnd . sprintf('%08X', $n++);
 321                         $stripped .= $marker;
 322
 323                         $tags[$marker] = "<$tag$attributes$empty>";
 324                         $params[$marker] = Sanitizer::decodeTagAttributes( $attributes );
 325
 326                         if ( $empty === '/' ) {
 327                                 // Empty element tag, <tag />
 328                                 $content[$marker] = null;
 329                                 $text = $inside;
 330                         } else {
 331                                 $q = preg_split( $end, $inside, 2 );
 332                                 $content[$marker] = $q[0];
 333                                 if( count( $q ) < 2 ) {
 334                                         # No end tag -- let it run out to the end of the text.
 335                                         break;
 336                                 } else {
 337                                         $text = $q[1];
 338                                 }
 339                         }
 340                 }
 341                 return $stripped;
 342         }
 343
 344         /**
 345          * Wrapper function for extractTagsAndParams
 346          * for cases where $tags and $params isn't needed
 347          * i.e. where tags will never have params, like <nowiki>
 348          *
 349          * @access private
 350          * @static
 351          */
 352         function extractTags( $tag, $text, &$content, $uniq_prefix = '' ) {
 353                 $dummy_tags = array();
 354                 $dummy_params = array();
 355
 356                 return Parser::extractTagsAndParams( $tag, $text, $content,
 357                         $dummy_tags, $dummy_params, $uniq_prefix );
 358         }
 359
 360         /**
 361          * Strips and renders nowiki, pre, math, hiero
 362          * If $render is set, performs necessary rendering operations on plugins
 363          * Returns the text, and fills an array with data needed in unstrip()
 364          * If the $state is already a valid strip state, it adds to the state
 365          *
 366          * @param bool $stripcomments when set, HTML comments <!-- like this -->
 367          *  will be stripped in addition to other tags. This is important
 368          *  for section editing, where these comments cause confusion when
 369          *  counting the sections in the wikisource
 370          *
 371          * @access private
 372          */
 373         function strip( $text, &$state, $stripcomments = false ) {
 374                 $render = ($this->mOutputType == OT_HTML);
 375                 $html_content = array();
 376                 $nowiki_content = array();
 377                 $math_content = array();
 378                 $pre_content = array();
 379                 $comment_content = array();
 380                 $ext_content = array();
 381                 $ext_tags = array();
 382                 $ext_params = array();
 383                 $gallery_content = array();
 384
 385                 # Replace any instances of the placeholders
 386                 $uniq_prefix = $this->mUniqPrefix;
 387                 #$text = str_replace( $uniq_prefix, wfHtmlEscapeFirst( $uniq_prefix ), $text );
 388
 389                 # html
 390                 global $wgRawHtml;
 391                 if( $wgRawHtml ) {
 392                         $text = Parser::extractTags('html', $text, $html_content, $uniq_prefix);
 393                         foreach( $html_content as $marker => $content ) {
 394                                 if ($render ) {
 395                                         # Raw and unchecked for validity.
 396                                         $html_content[$marker] = $content;
 397                                 } else {
 398                                         $html_content[$marker] = '<html>'.$content.'</html>';
 399                                 }
 400                         }
 401                 }
 402
 403                 # nowiki
 404                 $text = Parser::extractTags('nowiki', $text, $nowiki_content, $uniq_prefix);
 405                 foreach( $nowiki_content as $marker => $content ) {
 406                         if( $render ){
 407                                 $nowiki_content[$marker] = wfEscapeHTMLTagsOnly( $content );
 408                         } else {
 409                                 $nowiki_content[$marker] = '<nowiki>'.$content.'</nowiki>';
 410                         }
 411                 }
 412
 413                 # math
 414                 if( $this->mOptions->getUseTeX() ) {
 415                         $text = Parser::extractTags('math', $text, $math_content, $uniq_prefix);
 416                         foreach( $math_content as $marker => $content ){
 417                                 if( $render ) {
 418                                         $math_content[$marker] = renderMath( $content );
 419                                 } else {
 420                                         $math_content[$marker] = '<math>'.$content.'</math>';
 421                                 }
 422                         }
 423                 }
 424
 425                 # pre
 426                 $text = Parser::extractTags('pre', $text, $pre_content, $uniq_prefix);
 427                 foreach( $pre_content as $marker => $content ){
 428                         if( $render ){
 429                                 $pre_content[$marker] = '<pre>' . wfEscapeHTMLTagsOnly( $content ) . '</pre>';
 430                         } else {
 431                                 $pre_content[$marker] = '<pre>'.$content.'</pre>';
 432                         }
 433                 }
 434
 435                 # gallery
 436                 $text = Parser::extractTags('gallery', $text, $gallery_content, $uniq_prefix);
 437                 foreach( $gallery_content as $marker => $content ) {
 438                         require_once( 'ImageGallery.php' );
 439                         if ( $render ) {
 440                                 $gallery_content[$marker] = $this->renderImageGallery( $content );
 441                         } else {
 442                                 $gallery_content[$marker] = '<gallery>'.$content.'</gallery>';
 443                         }
 444                 }
 445
 446                 # Comments
 447                 if($stripcomments) {
 448                         $text = Parser::extractTags(STRIP_COMMENTS, $text, $comment_content, $uniq_prefix);
 449                         foreach( $comment_content as $marker => $content ){
 450                                 $comment_content[$marker] = '<!--'.$content.'-->';
 451                         }
 452                 }
 453
 454                 # Extensions
 455                 foreach ( $this->mTagHooks as $tag => $callback ) {
 456                         $ext_content[$tag] = array();
 457                         $text = Parser::extractTagsAndParams( $tag, $text, $ext_content[$tag],
 458                                 $ext_tags[$tag], $ext_params[$tag], $uniq_prefix );
 459                         foreach( $ext_content[$tag] as $marker => $content ) {
 460                                 $full_tag = $ext_tags[$tag][$marker];
 461                                 $params = $ext_params[$tag][$marker];
 462                                 if ( $render )
 463                                         $ext_content[$tag][$marker] = call_user_func_array( $callback, array( $content, $params, &$this ) );
 464                                 else {
 465                                         if ( is_null( $content ) ) {
 466                                                 // Empty element tag
 467                                                 $ext_content[$tag][$marker] = $full_tag;
 468                                         } else {
 469                                                 $ext_content[$tag][$marker] = "$full_tag$content</$tag>";
 470                                         }
 471                                 }
 472                         }
 473                 }
 474
 475                 # Merge state with the pre-existing state, if there is one
 476                 if ( $state ) {
 477                         $state['html'] = $state['html'] + $html_content;
 478                         $state['nowiki'] = $state['nowiki'] + $nowiki_content;
 479                         $state['math'] = $state['math'] + $math_content;
 480                         $state['pre'] = $state['pre'] + $pre_content;
 481                         $state['comment'] = $state['comment'] + $comment_content;
 482                         $state['gallery'] = $state['gallery'] + $gallery_content;
 483
 484                         foreach( $ext_content as $tag => $array ) {
 485                                 if ( array_key_exists( $tag, $state ) ) {
 486                                         $state[$tag] = $state[$tag] + $array;
 487                                 }
 488                         }
 489                 } else {
 490                         $state = array(
 491                           'html' => $html_content,
 492                           'nowiki' => $nowiki_content,
 493                           'math' => $math_content,
 494                           'pre' => $pre_content,
 495                           'comment' => $comment_content,
 496                           'gallery' => $gallery_content,
 497                         ) + $ext_content;
 498                 }
 499                 return $text;
 500         }
 501
 502         /**
 503          * restores pre, math, and hiero removed by strip()
 504          *
 505          * always call unstripNoWiki() after this one
 506          * @access private
 507          */
 508         function unstrip( $text, &$state ) {
 509                 if ( !is_array( $state ) ) {
 510                         return $text;
 511                 }
 512
 513                 # Must expand in reverse order, otherwise nested tags will be corrupted
 514                 foreach( array_reverse( $state, true ) as $tag => $contentDict ) {
 515                         if( $tag != 'nowiki' && $tag != 'html' ) {
 516                                 foreach( array_reverse( $contentDict, true ) as $uniq => $content ) {
 517                                         $text = str_replace( $uniq, $content, $text );
 518                                 }
 519                         }
 520                 }
 521
 522                 return $text;
 523         }
 524
 525         /**
 526          * always call this after unstrip() to preserve the order
 527          *
 528          * @access private
 529          */
 530         function unstripNoWiki( $text, &$state ) {
 531                 if ( !is_array( $state ) ) {
 532                         return $text;
 533                 }
 534
 535                 # Must expand in reverse order, otherwise nested tags will be corrupted
 536                 for ( $content = end($state['nowiki']); $content !== false; $content = prev( $state['nowiki'] ) ) {
 537                         $text = str_replace( key( $state['nowiki'] ), $content, $text );
 538                 }
 539
 540                 global $wgRawHtml;
 541                 if ($wgRawHtml) {
 542                         for ( $content = end($state['html']); $content !== false; $content = prev( $state['html'] ) ) {
 543                                 $text = str_replace( key( $state['html'] ), $content, $text );
 544                         }
 545                 }
 546
 547                 return $text;
 548         }
 549
 550         /**
 551          * Add an item to the strip state
 552          * Returns the unique tag which must be inserted into the stripped text
 553          * The tag will be replaced with the original text in unstrip()
 554          *
 555          * @access private
 556          */
 557         function insertStripItem( $text, &$state ) {
 558                 $rnd = $this->mUniqPrefix . '-item' . Parser::getRandomString();
 559                 if ( !$state ) {
 560                         $state = array(
 561                           'html' => array(),
 562                           'nowiki' => array(),
 563                           'math' => array(),
 564                           'pre' => array(),
 565                           'comment' => array(),
 566                           'gallery' => array(),
 567                         );
 568                 }
 569                 $state['item'][$rnd] = $text;
 570                 return $rnd;
 571         }
 572
 573         /**
 574          * Interface with html tidy, used if $wgUseTidy = true.
 575          * If tidy isn't able to correct the markup, the original will be
 576          * returned in all its glory with a warning comment appended.
 577          *
 578          * Either the external tidy program or the in-process tidy extension
 579          * will be used depending on availability. Override the default
 580          * $wgTidyInternal setting to disable the internal if it's not working.
 581          *
 582          * @param string $text Hideous HTML input
 583          * @return string Corrected HTML output
 584          * @access public
 585          * @static
 586          */
 587         function tidy( $text ) {
 588                 global $wgTidyInternal;
 589                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 590 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 591 '<head><title>test</title></head><body>'.$text.'</body></html>';
 592                 if( $wgTidyInternal ) {
 593                         $correctedtext = Parser::internalTidy( $wrappedtext );
 594                 } else {
 595                         $correctedtext = Parser::externalTidy( $wrappedtext );
 596                 }
 597                 if( is_null( $correctedtext ) ) {
 598                         wfDebug( "Tidy error detected!\n" );
 599                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 600                 }
 601                 return $correctedtext;
 602         }
 603
 604         /**
 605          * Spawn an external HTML tidy process and get corrected markup back from it.
 606          *
 607          * @access private
 608          * @static
 609          */
 610         function externalTidy( $text ) {
 611                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 612                 $fname = 'Parser::externalTidy';
 613                 wfProfileIn( $fname );
 614
 615                 $cleansource = '';
 616                 $opts = ' -utf8';
 617
 618                 $descriptorspec = array(
 619                         0 => array('pipe', 'r'),
 620                         1 => array('pipe', 'w'),
 621                         2 => array('file', '/dev/null', 'a')
 622                 );
 623                 $pipes = array();
 624                 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 625                 if (is_resource($process)) {
 626                         fwrite($pipes[0], $text);
 627                         fclose($pipes[0]);
 628                         while (!feof($pipes[1])) {
 629                                 $cleansource .= fgets($pipes[1], 1024);
 630                         }
 631                         fclose($pipes[1]);
 632                         proc_close($process);
 633                 }
 634
 635                 wfProfileOut( $fname );
 636
 637                 if( $cleansource == '' && $text != '') {
 638                         // Some kind of error happened, so we couldn't get the corrected text.
 639                         // Just give up; we'll use the source text and append a warning.
 640                         return null;
 641                 } else {
 642                         return $cleansource;
 643                 }
 644         }
 645
 646         /**
 647          * Use the HTML tidy PECL extension to use the tidy library in-process,
 648          * saving the overhead of spawning a new process. Currently written to
 649          * the PHP 4.3.x version of the extension, may not work on PHP 5.
 650          *
 651          * 'pear install tidy' should be able to compile the extension module.
 652          *
 653          * @access private
 654          * @static
 655          */
 656         function internalTidy( $text ) {
 657                 global $wgTidyConf;
 658                 $fname = 'Parser::internalTidy';
 659                 wfProfileIn( $fname );
 660
 661                 tidy_load_config( $wgTidyConf );
 662                 tidy_set_encoding( 'utf8' );
 663                 tidy_parse_string( $text );
 664                 tidy_clean_repair();
 665                 if( tidy_get_status() == 2 ) {
 666                         // 2 is magic number for fatal error
 667                         // http://www.php.net/manual/en/function.tidy-get-status.php
 668                         $cleansource = null;
 669                 } else {
 670                         $cleansource = tidy_get_output();
 671                 }
 672                 wfProfileOut( $fname );
 673                 return $cleansource;
 674         }
 675
 676         /**
 677          * parse the wiki syntax used to render tables
 678          *
 679          * @access private
 680          */
 681         function doTableStuff ( $t ) {
 682                 $fname = 'Parser::doTableStuff';
 683                 wfProfileIn( $fname );
 684
 685                 $t = explode ( "\n" , $t ) ;
 686                 $td = array () ; # Is currently a td tag open?
 687                 $ltd = array () ; # Was it TD or TH?
 688                 $tr = array () ; # Is currently a tr tag open?
 689                 $ltr = array () ; # tr attributes
 690                 $has_opened_tr = array(); # Did this table open a <tr> element?
 691                 $indent_level = 0; # indent level of the table
 692                 foreach ( $t AS $k => $x )
 693                 {
 694                         $x = trim ( $x ) ;
 695                         $fc = substr ( $x , 0 , 1 ) ;
 696                         if ( preg_match( '/^(:*)\{\|(.*)$/', $x, $matches ) ) {
 697                                 $indent_level = strlen( $matches[1] );
 698
 699                                 $attributes = $this->unstripForHTML( $matches[2] );
 700
 701                                 $t[$k] = str_repeat( '<dl><dd>', $indent_level ) .
 702                                         '<table' . Sanitizer::fixTagAttributes ( $attributes, 'table' ) . '>' ;
 703                                 array_push ( $td , false ) ;
 704                                 array_push ( $ltd , '' ) ;
 705                                 array_push ( $tr , false ) ;
 706                                 array_push ( $ltr , '' ) ;
 707                                 array_push ( $has_opened_tr, false );
 708                         }
 709                         else if ( count ( $td ) == 0 ) { } # Don't do any of the following
 710                         else if ( '|}' == substr ( $x , 0 , 2 ) ) {
 711                                 $z = "</table>" . substr ( $x , 2);
 712                                 $l = array_pop ( $ltd ) ;
 713                                 if ( !array_pop ( $has_opened_tr ) ) $z = "<tr><td></td></tr>" . $z ;
 714                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 715                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 716                                 array_pop ( $ltr ) ;
 717                                 $t[$k] = $z . str_repeat( '</dd></dl>', $indent_level );
 718                         }
 719                         else if ( '|-' == substr ( $x , 0 , 2 ) ) { # Allows for |---------------
 720                                 $x = substr ( $x , 1 ) ;
 721                                 while ( $x != '' && substr ( $x , 0 , 1 ) == '-' ) $x = substr ( $x , 1 ) ;
 722                                 $z = '' ;
 723                                 $l = array_pop ( $ltd ) ;
 724                                 array_pop ( $has_opened_tr );
 725                                 array_push ( $has_opened_tr , true ) ;
 726                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 727                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 728                                 array_pop ( $ltr ) ;
 729                                 $t[$k] = $z ;
 730                                 array_push ( $tr , false ) ;
 731                                 array_push ( $td , false ) ;
 732                                 array_push ( $ltd , '' ) ;
 733                                 $attributes = $this->unstripForHTML( $x );
 734                                 array_push ( $ltr , Sanitizer::fixTagAttributes ( $attributes, 'tr' ) ) ;
 735                         }
 736                         else if ( '|' == $fc || '!' == $fc || '|+' == substr ( $x , 0 , 2 ) ) { # Caption
 737                                 # $x is a table row
 738                                 if ( '|+' == substr ( $x , 0 , 2 ) ) {
 739                                         $fc = '+' ;
 740                                         $x = substr ( $x , 1 ) ;
 741                                 }
 742                                 $after = substr ( $x , 1 ) ;
 743                                 if ( $fc == '!' ) $after = str_replace ( '!!' , '||' , $after ) ;
 744                                 $after = explode ( '||' , $after ) ;
 745                                 $t[$k] = '' ;
 746
 747                                 # Loop through each table cell
 748                                 foreach ( $after AS $theline )
 749                                 {
 750                                         $z = '' ;
 751                                         if ( $fc != '+' )
 752                                         {
 753                                                 $tra = array_pop ( $ltr ) ;
 754                                                 if ( !array_pop ( $tr ) ) $z = '<tr'.$tra.">\n" ;
 755                                                 array_push ( $tr , true ) ;
 756                                                 array_push ( $ltr , '' ) ;
 757                                                 array_pop ( $has_opened_tr );
 758                                                 array_push ( $has_opened_tr , true ) ;
 759                                         }
 760
 761                                         $l = array_pop ( $ltd ) ;
 762                                         if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 763                                         if ( $fc == '|' ) $l = 'td' ;
 764                                         else if ( $fc == '!' ) $l = 'th' ;
 765                                         else if ( $fc == '+' ) $l = 'caption' ;
 766                                         else $l = '' ;
 767                                         array_push ( $ltd , $l ) ;
 768
 769                                         # Cell parameters
 770                                         $y = explode ( '|' , $theline , 2 ) ;
 771                                         # Note that a '|' inside an invalid link should not
 772                                         # be mistaken as delimiting cell parameters
 773                                         if ( strpos( $y[0], '[[' ) !== false ) {
 774                                                 $y = array ($theline);
 775                                         }
 776                                         if ( count ( $y ) == 1 )
 777                                                 $y = "{$z}<{$l}>{$y[0]}" ;
 778                                         else {
 779                                                 $attributes = $this->unstripForHTML( $y[0] );
 780                                                 $y = "{$z}<{$l}".Sanitizer::fixTagAttributes($attributes, $l).">{$y[1]}" ;
 781                                         }
 782                                         $t[$k] .= $y ;
 783                                         array_push ( $td , true ) ;
 784                                 }
 785                         }
 786                 }
 787
 788                 # Closing open td, tr && table
 789                 while ( count ( $td ) > 0 )
 790                 {
 791                         $l = array_pop ( $ltd ) ;
 792                         if ( array_pop ( $td ) ) $t[] = '</td>' ;
 793                         if ( array_pop ( $tr ) ) $t[] = '</tr>' ;
 794                         if ( array_pop ( $has_opened_tr ) ) $t[] = "<tr><td></td></tr>" ;
 795                         $t[] = '</table>' ;
 796                 }
 797
 798                 $t = implode ( "\n" , $t ) ;
 799                 wfProfileOut( $fname );
 800                 return $t ;
 801         }
 802
 803         /**
 804          * Helper function for parse() that transforms wiki markup into
 805          * HTML. Only called for $mOutputType == OT_HTML.
 806          *
 807          * @access private
 808          */
 809         function internalParse( $text ) {
 810                 global $wgContLang;
 811                 $args = array();
 812                 $isMain = true;
 813                 $fname = 'Parser::internalParse';
 814                 wfProfileIn( $fname );
 815
 816                 # Remove <noinclude> tags and <includeonly> sections
 817                 $text = strtr( $text, array( '<onlyinclude>' => '' , '</onlyinclude>' => '' ) );
 818                 $text = strtr( $text, array( '<noinclude>' => '', '</noinclude>' => '') );
 819                 $text = preg_replace( '/<includeonly>.*?<\/includeonly>/s', '', $text );
 820
 821                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ) );
 822                 $text = $this->replaceVariables( $text, $args );
 823
 824                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 825
 826                 $text = $this->doHeadings( $text );
 827                 if($this->mOptions->getUseDynamicDates()) {
 828                         $df =& DateFormatter::getInstance();
 829                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 830                 }
 831                 $text = $this->doAllQuotes( $text );
 832                 $text = $this->replaceInternalLinks( $text );
 833                 $text = $this->replaceExternalLinks( $text );
 834
 835                 # replaceInternalLinks may sometimes leave behind
 836                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 837                 $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
 838
 839                 $text = $this->doMagicLinks( $text );
 840                 $text = $this->doTableStuff( $text );
 841                 $text = $this->formatHeadings( $text, $isMain );
 842
 843                 wfProfileOut( $fname );
 844                 return $text;
 845         }
 846
 847         /**
 848          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 849          * magic external links.
 850          *
 851          * @access private
 852          */
 853         function &doMagicLinks( &$text ) {
 854                 $text = $this->magicISBN( $text );
 855                 $text = $this->magicRFC( $text, 'RFC ', 'rfcurl' );
 856                 $text = $this->magicRFC( $text, 'PMID ', 'pubmedurl' );
 857                 return $text;
 858         }
 859
 860         /**
 861          * Parse headers and return html
 862          *
 863          * @access private
 864          */
 865         function doHeadings( $text ) {
 866                 $fname = 'Parser::doHeadings';
 867                 wfProfileIn( $fname );
 868                 for ( $i = 6; $i >= 1; --$i ) {
 869                         $h = str_repeat( '=', $i );
 870                         $text = preg_replace( "/^{$h}(.+){$h}(\\s|$)/m",
 871                           "<h{$i}>\\1</h{$i}>\\2", $text );
 872                 }
 873                 wfProfileOut( $fname );
 874                 return $text;
 875         }
 876
 877         /**
 878          * Replace single quotes with HTML markup
 879          * @access private
 880          * @return string the altered text
 881          */
 882         function doAllQuotes( $text ) {
 883                 $fname = 'Parser::doAllQuotes';
 884                 wfProfileIn( $fname );
 885                 $outtext = '';
 886                 $lines = explode( "\n", $text );
 887                 foreach ( $lines as $line ) {
 888                         $outtext .= $this->doQuotes ( $line ) . "\n";
 889                 }
 890                 $outtext = substr($outtext, 0,-1);
 891                 wfProfileOut( $fname );
 892                 return $outtext;
 893         }
 894
 895         /**
 896          * Helper function for doAllQuotes()
 897          * @access private
 898          */
 899         function doQuotes( $text ) {
 900                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
 901                 if ( count( $arr ) == 1 )
 902                         return $text;
 903                 else
 904                 {
 905                         # First, do some preliminary work. This may shift some apostrophes from
 906                         # being mark-up to being text. It also counts the number of occurrences
 907                         # of bold and italics mark-ups.
 908                         $i = 0;
 909                         $numbold = 0;
 910                         $numitalics = 0;
 911                         foreach ( $arr as $r )
 912                         {
 913                                 if ( ( $i % 2 ) == 1 )
 914                                 {
 915                                         # If there are ever four apostrophes, assume the first is supposed to
 916                                         # be text, and the remaining three constitute mark-up for bold text.
 917                                         if ( strlen( $arr[$i] ) == 4 )
 918                                         {
 919                                                 $arr[$i-1] .= "'";
 920                                                 $arr[$i] = "'''";
 921                                         }
 922                                         # If there are more than 5 apostrophes in a row, assume they're all
 923                                         # text except for the last 5.
 924                                         else if ( strlen( $arr[$i] ) > 5 )
 925                                         {
 926                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
 927                                                 $arr[$i] = "'''''";
 928                                         }
 929                                         # Count the number of occurrences of bold and italics mark-ups.
 930                                         # We are not counting sequences of five apostrophes.
 931                                         if ( strlen( $arr[$i] ) == 2 ) $numitalics++;  else
 932                                         if ( strlen( $arr[$i] ) == 3 ) $numbold++;     else
 933                                         if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
 934                                 }
 935                                 $i++;
 936                         }
 937
 938                         # If there is an odd number of both bold and italics, it is likely
 939                         # that one of the bold ones was meant to be an apostrophe followed
 940                         # by italics. Which one we cannot know for certain, but it is more
 941                         # likely to be one that has a single-letter word before it.
 942                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
 943                         {
 944                                 $i = 0;
 945                                 $firstsingleletterword = -1;
 946                                 $firstmultiletterword = -1;
 947                                 $firstspace = -1;
 948                                 foreach ( $arr as $r )
 949                                 {
 950                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
 951                                         {
 952                                                 $x1 = substr ($arr[$i-1], -1);
 953                                                 $x2 = substr ($arr[$i-1], -2, 1);
 954                                                 if ($x1 == ' ') {
 955                                                         if ($firstspace == -1) $firstspace = $i;
 956                                                 } else if ($x2 == ' ') {
 957                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
 958                                                 } else {
 959                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
 960                                                 }
 961                                         }
 962                                         $i++;
 963                                 }
 964
 965                                 # If there is a single-letter word, use it!
 966                                 if ($firstsingleletterword > -1)
 967                                 {
 968                                         $arr [ $firstsingleletterword ] = "''";
 969                                         $arr [ $firstsingleletterword-1 ] .= "'";
 970                                 }
 971                                 # If not, but there's a multi-letter word, use that one.
 972                                 else if ($firstmultiletterword > -1)
 973                                 {
 974                                         $arr [ $firstmultiletterword ] = "''";
 975                                         $arr [ $firstmultiletterword-1 ] .= "'";
 976                                 }
 977                                 # ... otherwise use the first one that has neither.
 978                                 # (notice that it is possible for all three to be -1 if, for example,
 979                                 # there is only one pentuple-apostrophe in the line)
 980                                 else if ($firstspace > -1)
 981                                 {
 982                                         $arr [ $firstspace ] = "''";
 983                                         $arr [ $firstspace-1 ] .= "'";
 984                                 }
 985                         }
 986
 987                         # Now let's actually convert our apostrophic mush to HTML!
 988                         $output = '';
 989                         $buffer = '';
 990                         $state = '';
 991                         $i = 0;
 992                         foreach ($arr as $r)
 993                         {
 994                                 if (($i % 2) == 0)
 995                                 {
 996                                         if ($state == 'both')
 997                                                 $buffer .= $r;
 998                                         else
 999                                                 $output .= $r;
1000                                 }
1001                                 else
1002                                 {
1003                                         if (strlen ($r) == 2)
1004                                         {
1005                                                 if ($state == 'i')
1006                                                 { $output .= '</i>'; $state = ''; }
1007                                                 else if ($state == 'bi')
1008                                                 { $output .= '</i>'; $state = 'b'; }
1009                                                 else if ($state == 'ib')
1010                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1011                                                 else if ($state == 'both')
1012                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1013                                                 else # $state can be 'b' or ''
1014                                                 { $output .= '<i>'; $state .= 'i'; }
1015                                         }
1016                                         else if (strlen ($r) == 3)
1017                                         {
1018                                                 if ($state == 'b')
1019                                                 { $output .= '</b>'; $state = ''; }
1020                                                 else if ($state == 'bi')
1021                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1022                                                 else if ($state == 'ib')
1023                                                 { $output .= '</b>'; $state = 'i'; }
1024                                                 else if ($state == 'both')
1025                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1026                                                 else # $state can be 'i' or ''
1027                                                 { $output .= '<b>'; $state .= 'b'; }
1028                                         }
1029                                         else if (strlen ($r) == 5)
1030                                         {
1031                                                 if ($state == 'b')
1032                                                 { $output .= '</b><i>'; $state = 'i'; }
1033                                                 else if ($state == 'i')
1034                                                 { $output .= '</i><b>'; $state = 'b'; }
1035                                                 else if ($state == 'bi')
1036                                                 { $output .= '</i></b>'; $state = ''; }
1037                                                 else if ($state == 'ib')
1038                                                 { $output .= '</b></i>'; $state = ''; }
1039                                                 else if ($state == 'both')
1040                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1041                                                 else # ($state == '')
1042                                                 { $buffer = ''; $state = 'both'; }
1043                                         }
1044                                 }
1045                                 $i++;
1046                         }
1047                         # Now close all remaining tags.  Notice that the order is important.
1048                         if ($state == 'b' || $state == 'ib')
1049                                 $output .= '</b>';
1050                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
1051                                 $output .= '</i>';
1052                         if ($state == 'bi')
1053                                 $output .= '</b>';
1054                         if ($state == 'both')
1055                                 $output .= '<b><i>'.$buffer.'</i></b>';
1056                         return $output;
1057                 }
1058         }
1059
1060         /**
1061          * Replace external links
1062          *
1063          * Note: this is all very hackish and the order of execution matters a lot.
1064          * Make sure to run maintenance/parserTests.php if you change this code.
1065          *
1066          * @access private
1067          */
1068         function replaceExternalLinks( $text ) {
1069                 global $wgContLang;
1070                 $fname = 'Parser::replaceExternalLinks';
1071                 wfProfileIn( $fname );
1072
1073                 $sk =& $this->mOptions->getSkin();
1074
1075                 $bits = preg_split( EXT_LINK_BRACKETED, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1076
1077                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
1078
1079                 $i = 0;
1080                 while ( $i<count( $bits ) ) {
1081                         $url = $bits[$i++];
1082                         $protocol = $bits[$i++];
1083                         $text = $bits[$i++];
1084                         $trail = $bits[$i++];
1085
1086                         # The characters '<' and '>' (which were escaped by
1087                         # removeHTMLtags()) should not be included in
1088                         # URLs, per RFC 2396.
1089                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1090                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1091                                 $url = substr($url, 0, $m2[0][1]);
1092                         }
1093
1094                         # If the link text is an image URL, replace it with an <img> tag
1095                         # This happened by accident in the original parser, but some people used it extensively
1096                         $img = $this->maybeMakeExternalImage( $text );
1097                         if ( $img !== false ) {
1098                                 $text = $img;
1099                         }
1100
1101                         $dtrail = '';
1102
1103                         # Set linktype for CSS - if URL==text, link is essentially free
1104                         $linktype = ($text == $url) ? 'free' : 'text';
1105
1106                         # No link text, e.g. [http://domain.tld/some.link]
1107                         if ( $text == '' ) {
1108                                 # Autonumber if allowed
1109                                 if ( strpos( HTTP_PROTOCOLS, str_replace('/','\/', $protocol) ) !== false ) {
1110                                         $text = '[' . ++$this->mAutonumber . ']';
1111                                         $linktype = 'autonumber';
1112                                 } else {
1113                                         # Otherwise just use the URL
1114                                         $text = htmlspecialchars( $url );
1115                                         $linktype = 'free';
1116                                 }
1117                         } else {
1118                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1119                                 # Check for trail
1120                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1121                         }
1122
1123                         $text = $wgContLang->markNoConversion($text);
1124
1125                         # Replace &amp; from obsolete syntax with &.
1126                         # All HTML entities will be escaped by makeExternalLink()
1127                         $url = str_replace( '&amp;', '&', $url );
1128                         # Replace unnecessary URL escape codes with the referenced character
1129                         # This prevents spammers from hiding links from the filters
1130                         $url = Parser::replaceUnusualEscapes( $url );
1131
1132                         # Process the trail (i.e. everything after this link up until start of the next link),
1133                         # replacing any non-bracketed links
1134                         $trail = $this->replaceFreeExternalLinks( $trail );
1135
1136                         # Use the encoded URL
1137                         # This means that users can paste URLs directly into the text
1138                         # Funny characters like &ouml; aren't valid in URLs anyway
1139                         # This was changed in August 2004
1140                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype ) . $dtrail . $trail;
1141
1142                         # Register link in the output object
1143                         $this->mOutput->addExternalLink( $url );
1144                 }
1145
1146                 wfProfileOut( $fname );
1147                 return $s;
1148         }
1149
1150         /**
1151          * Replace anything that looks like a URL with a link
1152          * @access private
1153          */
1154         function replaceFreeExternalLinks( $text ) {
1155                 global $wgContLang;
1156                 $fname = 'Parser::replaceFreeExternalLinks';
1157                 wfProfileIn( $fname );
1158
1159                 $bits = preg_split( '/(\b(?:' . wfUrlProtocols() . '))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1160                 $s = array_shift( $bits );
1161                 $i = 0;
1162
1163                 $sk =& $this->mOptions->getSkin();
1164
1165                 while ( $i < count( $bits ) ){
1166                         $protocol = $bits[$i++];
1167                         $remainder = $bits[$i++];
1168
1169                         if ( preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1170                                 # Found some characters after the protocol that look promising
1171                                 $url = $protocol . $m[1];
1172                                 $trail = $m[2];
1173
1174                                 # The characters '<' and '>' (which were escaped by
1175                                 # removeHTMLtags()) should not be included in
1176                                 # URLs, per RFC 2396.
1177                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1178                                         $trail = substr($url, $m2[0][1]) . $trail;
1179                                         $url = substr($url, 0, $m2[0][1]);
1180                                 }
1181
1182                                 # Move trailing punctuation to $trail
1183                                 $sep = ',;\.:!?';
1184                                 # If there is no left bracket, then consider right brackets fair game too
1185                                 if ( strpos( $url, '(' ) === false ) {
1186                                         $sep .= ')';
1187                                 }
1188
1189                                 $numSepChars = strspn( strrev( $url ), $sep );
1190                                 if ( $numSepChars ) {
1191                                         $trail = substr( $url, -$numSepChars ) . $trail;
1192                                         $url = substr( $url, 0, -$numSepChars );
1193                                 }
1194
1195                                 # Replace &amp; from obsolete syntax with &.
1196                                 # All HTML entities will be escaped by makeExternalLink()
1197                                 # or maybeMakeExternalImage()
1198                                 $url = str_replace( '&amp;', '&', $url );
1199                                 # Replace unnecessary URL escape codes with their equivalent characters
1200                                 $url = Parser::replaceUnusualEscapes( $url );
1201
1202                                 # Is this an external image?
1203                                 $text = $this->maybeMakeExternalImage( $url );
1204                                 if ( $text === false ) {
1205                                         # Not an image, make a link
1206                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free' );
1207                                         # Register it in the output object
1208                                         $this->mOutput->addExternalLink( $url );
1209                                 }
1210                                 $s .= $text . $trail;
1211                         } else {
1212                                 $s .= $protocol . $remainder;
1213                         }
1214                 }
1215                 wfProfileOut( $fname );
1216                 return $s;
1217         }
1218
1219         /**
1220          * Replace unusual URL escape codes with their equivalent characters
1221          * @param string
1222          * @return string
1223          * @static
1224          */
1225         function replaceUnusualEscapes( $url ) {
1226                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1227                         array( 'Parser', 'replaceUnusualEscapesCallback' ), $url );
1228         }
1229
1230         /**
1231          * Callback function used in replaceUnusualEscapes().
1232          * Replaces unusual URL escape codes with their equivalent character
1233          * @static
1234          * @access private
1235          */
1236         function replaceUnusualEscapesCallback( $matches ) {
1237                 $char = urldecode( $matches[0] );
1238                 $ord = ord( $char );
1239                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1240                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1241                         // No, shouldn't be escaped
1242                         return $char;
1243                 } else {
1244                         // Yes, leave it escaped
1245                         return $matches[0];
1246                 }
1247         }
1248
1249         /**
1250          * make an image if it's allowed, either through the global
1251          * option or through the exception
1252          * @access private
1253          */
1254         function maybeMakeExternalImage( $url ) {
1255                 $sk =& $this->mOptions->getSkin();
1256                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1257                 $imagesexception = !empty($imagesfrom);
1258                 $text = false;
1259                 if ( $this->mOptions->getAllowExternalImages()
1260                      || ( $imagesexception && strpos( $url, $imagesfrom ) === 0 ) ) {
1261                         if ( preg_match( EXT_IMAGE_REGEX, $url ) ) {
1262                                 # Image found
1263                                 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
1264                         }
1265                 }
1266                 return $text;
1267         }
1268
1269         /**
1270          * Process [[ ]] wikilinks
1271          *
1272          * @access private
1273          */
1274         function replaceInternalLinks( $s ) {
1275                 global $wgContLang;
1276                 static $fname = 'Parser::replaceInternalLinks' ;
1277
1278                 wfProfileIn( $fname );
1279
1280                 wfProfileIn( $fname.'-setup' );
1281                 static $tc = FALSE;
1282                 # the % is needed to support urlencoded titles as well
1283                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1284
1285                 $sk =& $this->mOptions->getSkin();
1286
1287                 #split the entire text string on occurences of [[
1288                 $a = explode( '[[', ' ' . $s );
1289                 #get the first element (all text up to first [[), and remove the space we added
1290                 $s = array_shift( $a );
1291                 $s = substr( $s, 1 );
1292
1293                 # Match a link having the form [[namespace:link|alternate]]trail
1294                 static $e1 = FALSE;
1295                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1296                 # Match cases where there is no "]]", which might still be images
1297                 static $e1_img = FALSE;
1298                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1299                 # Match the end of a line for a word that's not followed by whitespace,
1300                 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1301                 $e2 = wfMsgForContent( 'linkprefix' );
1302
1303                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1304
1305                 if( is_null( $this->mTitle ) ) {
1306                         wfDebugDieBacktrace( 'nooo' );
1307                 }
1308                 $nottalk = !$this->mTitle->isTalkPage();
1309
1310                 if ( $useLinkPrefixExtension ) {
1311                         if ( preg_match( $e2, $s, $m ) ) {
1312                                 $first_prefix = $m[2];
1313                         } else {
1314                                 $first_prefix = false;
1315                         }
1316                 } else {
1317                         $prefix = '';
1318                 }
1319
1320                 $selflink = $this->mTitle->getPrefixedText();
1321                 wfProfileOut( $fname.'-setup' );
1322
1323                 $checkVariantLink = sizeof($wgContLang->getVariants())>1;
1324                 $useSubpages = $this->areSubpagesAllowed();
1325
1326                 # Loop for each link
1327                 for ($k = 0; isset( $a[$k] ); $k++) {
1328                         $line = $a[$k];
1329                         if ( $useLinkPrefixExtension ) {
1330                                 wfProfileIn( $fname.'-prefixhandling' );
1331                                 if ( preg_match( $e2, $s, $m ) ) {
1332                                         $prefix = $m[2];
1333                                         $s = $m[1];
1334                                 } else {
1335                                         $prefix='';
1336                                 }
1337                                 # first link
1338                                 if($first_prefix) {
1339                                         $prefix = $first_prefix;
1340                                         $first_prefix = false;
1341                                 }
1342                                 wfProfileOut( $fname.'-prefixhandling' );
1343                         }
1344
1345                         $might_be_img = false;
1346
1347                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1348                                 $text = $m[2];
1349                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1350                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1351                                 # the real problem is with the $e1 regex
1352                                 # See bug 1300.
1353                                 #
1354                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1355                                 # and no image is in sight. See bug 2095.
1356                                 #
1357                                 if( $text !== '' && preg_match( "/^\](.*)/s", $m[3], $n ) ) {
1358                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1359                                         $m[3] = $n[1];
1360                                 }
1361                                 # fix up urlencoded title texts
1362                                 if(preg_match('/%/', $m[1] )) $m[1] = urldecode($m[1]);
1363                                 $trail = $m[3];
1364                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1365                                 $might_be_img = true;
1366                                 $text = $m[2];
1367                                 if(preg_match('/%/', $m[1] )) $m[1] = urldecode($m[1]);
1368                                 $trail = "";
1369                         } else { # Invalid form; output directly
1370                                 $s .= $prefix . '[[' . $line ;
1371                                 continue;
1372                         }
1373
1374                         # Don't allow internal links to pages containing
1375                         # PROTO: where PROTO is a valid URL protocol; these
1376                         # should be external links.
1377                         if (preg_match('/^(\b(?:' . wfUrlProtocols() . '))/', $m[1])) {
1378                                 $s .= $prefix . '[[' . $line ;
1379                                 continue;
1380                         }
1381
1382                         # Make subpage if necessary
1383                         if( $useSubpages ) {
1384                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1385                         } else {
1386                                 $link = $m[1];
1387                         }
1388
1389                         $noforce = (substr($m[1], 0, 1) != ':');
1390                         if (!$noforce) {
1391                                 # Strip off leading ':'
1392                                 $link = substr($link, 1);
1393                         }
1394
1395                         $nt = Title::newFromText( $this->unstripNoWiki($link, $this->mStripState) );
1396                         if( !$nt ) {
1397                                 $s .= $prefix . '[[' . $line;
1398                                 continue;
1399                         }
1400
1401                         #check other language variants of the link
1402                         #if the article does not exist
1403                         if( $checkVariantLink
1404                             && $nt->getArticleID() == 0 ) {
1405                                 $wgContLang->findVariantLink($link, $nt);
1406                         }
1407
1408                         $ns = $nt->getNamespace();
1409                         $iw = $nt->getInterWiki();
1410
1411                         if ($might_be_img) { # if this is actually an invalid link
1412                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1413                                         $found = false;
1414                                         while (isset ($a[$k+1]) ) {
1415                                                 #look at the next 'line' to see if we can close it there
1416                                                 $spliced = array_splice( $a, $k + 1, 1 );
1417                                                 $next_line = array_shift( $spliced );
1418                                                 if( preg_match("/^(.*?]].*?)]](.*)$/sD", $next_line, $m) ) {
1419                                                 # the first ]] closes the inner link, the second the image
1420                                                         $found = true;
1421                                                         $text .= '[[' . $m[1];
1422                                                         $trail = $m[2];
1423                                                         break;
1424                                                 } elseif( preg_match("/^.*?]].*$/sD", $next_line, $m) ) {
1425                                                         #if there's exactly one ]] that's fine, we'll keep looking
1426                                                         $text .= '[[' . $m[0];
1427                                                 } else {
1428                                                         #if $next_line is invalid too, we need look no further
1429                                                         $text .= '[[' . $next_line;
1430                                                         break;
1431                                                 }
1432                                         }
1433                                         if ( !$found ) {
1434                                                 # we couldn't find the end of this imageLink, so output it raw
1435                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1436                                                 $text = $this->replaceInternalLinks($text);
1437                                                 $s .= $prefix . '[[' . $link . '|' . $text;
1438                                                 # note: no $trail, because without an end, there *is* no trail
1439                                                 continue;
1440                                         }
1441                                 } else { #it's not an image, so output it raw
1442                                         $s .= $prefix . '[[' . $link . '|' . $text;
1443                                         # note: no $trail, because without an end, there *is* no trail
1444                                         continue;
1445                                 }
1446                         }
1447
1448                         $wasblank = ( '' == $text );
1449                         if( $wasblank ) $text = $link;
1450
1451
1452                         # Link not escaped by : , create the various objects
1453                         if( $noforce ) {
1454
1455                                 # Interwikis
1456                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1457                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1458                                         $s = rtrim($s . "\n");
1459                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1460                                         continue;
1461                                 }
1462
1463                                 if ( $ns == NS_IMAGE ) {
1464                                         wfProfileIn( "$fname-image" );
1465                                         if ( !wfIsBadImage( $nt->getDBkey() ) ) {
1466                                                 # recursively parse links inside the image caption
1467                                                 # actually, this will parse them in any other parameters, too,
1468                                                 # but it might be hard to fix that, and it doesn't matter ATM
1469                                                 $text = $this->replaceExternalLinks($text);
1470                                                 $text = $this->replaceInternalLinks($text);
1471
1472                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1473                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text ) ) . $trail;
1474                                                 $this->mOutput->addImage( $nt->getDBkey() );
1475
1476                                                 wfProfileOut( "$fname-image" );
1477                                                 continue;
1478                                         }
1479                                         wfProfileOut( "$fname-image" );
1480
1481                                 }
1482
1483                                 if ( $ns == NS_CATEGORY ) {
1484                                         wfProfileIn( "$fname-category" );
1485                                         $s = rtrim($s . "\n"); # bug 87
1486
1487                                         if ( $wasblank ) {
1488                                                 if ( $this->mTitle->getNamespace() == NS_CATEGORY ) {
1489                                                         $sortkey = $this->mTitle->getText();
1490                                                 } else {
1491                                                         $sortkey = $this->mTitle->getPrefixedText();
1492                                                 }
1493                                         } else {
1494                                                 $sortkey = $text;
1495                                         }
1496                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1497                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1498                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1499
1500                                         /**
1501                                          * Strip the whitespace Category links produce, see bug 87
1502                                          * @todo We might want to use trim($tmp, "\n") here.
1503                                          */
1504                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1505
1506                                         wfProfileOut( "$fname-category" );
1507                                         continue;
1508                                 }
1509                         }
1510
1511                         if( ( $nt->getPrefixedText() === $selflink ) &&
1512                             ( $nt->getFragment() === '' ) ) {
1513                                 # Self-links are handled specially; generally de-link and change to bold.
1514                                 $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1515                                 continue;
1516                         }
1517
1518                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1519                         if( $ns == NS_MEDIA ) {
1520                                 $link = $sk->makeMediaLinkObj( $nt, $text );
1521                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1522                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1523                                 $this->mOutput->addImage( $nt->getDBkey() );
1524                                 continue;
1525                         } elseif( $ns == NS_SPECIAL ) {
1526                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1527                                 continue;
1528                         } elseif( $ns == NS_IMAGE ) {
1529                                 $img = Image::newFromTitle( $nt );
1530                                 if( $img->exists() ) {
1531                                         // Force a blue link if the file exists; may be a remote
1532                                         // upload on the shared repository, and we want to see its
1533                                         // auto-generated page.
1534                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1535                                         continue;
1536                                 }
1537                         }
1538                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1539                 }
1540                 wfProfileOut( $fname );
1541                 return $s;
1542         }
1543
1544         /**
1545          * Make a link placeholder. The text returned can be later resolved to a real link with
1546          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1547          * parsing of interwiki links, and secondly to allow all extistence checks and
1548          * article length checks (for stub links) to be bundled into a single query.
1549          *
1550          */
1551         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1552                 if ( ! is_object($nt) ) {
1553                         # Fail gracefully
1554                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1555                 } else {
1556                         # Separate the link trail from the rest of the link
1557                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1558
1559                         if ( $nt->isExternal() ) {
1560                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1561                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1562                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1563                         } else {
1564                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1565                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1566                                 $this->mLinkHolders['queries'][] = $query;
1567                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1568                                 $this->mLinkHolders['titles'][] = $nt;
1569
1570                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1571                         }
1572                 }
1573                 return $retVal;
1574         }
1575
1576         /**
1577          * Render a forced-blue link inline; protect against double expansion of
1578          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1579          * Since this little disaster has to split off the trail text to avoid
1580          * breaking URLs in the following text without breaking trails on the
1581          * wiki links, it's been made into a horrible function.
1582          *
1583          * @param Title $nt
1584          * @param string $text
1585          * @param string $query
1586          * @param string $trail
1587          * @param string $prefix
1588          * @return string HTML-wikitext mix oh yuck
1589          */
1590         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1591                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1592                 $sk =& $this->mOptions->getSkin();
1593                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1594                 return $this->armorLinks( $link ) . $trail;
1595         }
1596
1597         /**
1598          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1599          * going to go through further parsing steps before inline URL expansion.
1600          *
1601          * In particular this is important when using action=render, which causes
1602          * full URLs to be included.
1603          *
1604          * Oh man I hate our multi-layer parser!
1605          *
1606          * @param string more-or-less HTML
1607          * @return string less-or-more HTML with NOPARSE bits
1608          */
1609         function armorLinks( $text ) {
1610                 return preg_replace( "/\b(" . wfUrlProtocols() . ')/',
1611                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1612         }
1613
1614         /**
1615          * Return true if subpage links should be expanded on this page.
1616          * @return bool
1617          */
1618         function areSubpagesAllowed() {
1619                 # Some namespaces don't allow subpages
1620                 global $wgNamespacesWithSubpages;
1621                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1622         }
1623
1624         /**
1625          * Handle link to subpage if necessary
1626          * @param string $target the source of the link
1627          * @param string &$text the link text, modified as necessary
1628          * @return string the full name of the link
1629          * @access private
1630          */
1631         function maybeDoSubpageLink($target, &$text) {
1632                 # Valid link forms:
1633                 # Foobar -- normal
1634                 # :Foobar -- override special treatment of prefix (images, language links)
1635                 # /Foobar -- convert to CurrentPage/Foobar
1636                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1637                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1638                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1639
1640                 $fname = 'Parser::maybeDoSubpageLink';
1641                 wfProfileIn( $fname );
1642                 $ret = $target; # default return value is no change
1643
1644                 # Some namespaces don't allow subpages,
1645                 # so only perform processing if subpages are allowed
1646                 if( $this->areSubpagesAllowed() ) {
1647                         # Look at the first character
1648                         if( $target != '' && $target{0} == '/' ) {
1649                                 # / at end means we don't want the slash to be shown
1650                                 if( substr( $target, -1, 1 ) == '/' ) {
1651                                         $target = substr( $target, 1, -1 );
1652                                         $noslash = $target;
1653                                 } else {
1654                                         $noslash = substr( $target, 1 );
1655                                 }
1656
1657                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash);
1658                                 if( '' === $text ) {
1659                                         $text = $target;
1660                                 } # this might be changed for ugliness reasons
1661                         } else {
1662                                 # check for .. subpage backlinks
1663                                 $dotdotcount = 0;
1664                                 $nodotdot = $target;
1665                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1666                                         ++$dotdotcount;
1667                                         $nodotdot = substr( $nodotdot, 3 );
1668                                 }
1669                                 if($dotdotcount > 0) {
1670                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1671                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1672                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1673                                                 # / at the end means don't show full path
1674                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1675                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1676                                                         if( '' === $text ) {
1677                                                                 $text = $nodotdot;
1678                                                         }
1679                                                 }
1680                                                 $nodotdot = trim( $nodotdot );
1681                                                 if( $nodotdot != '' ) {
1682                                                         $ret .= '/' . $nodotdot;
1683                                                 }
1684                                         }
1685                                 }
1686                         }
1687                 }
1688
1689                 wfProfileOut( $fname );
1690                 return $ret;
1691         }
1692
1693         /**#@+
1694          * Used by doBlockLevels()
1695          * @access private
1696          */
1697         /* private */ function closeParagraph() {
1698                 $result = '';
1699                 if ( '' != $this->mLastSection ) {
1700                         $result = '</' . $this->mLastSection  . ">\n";
1701                 }
1702                 $this->mInPre = false;
1703                 $this->mLastSection = '';
1704                 return $result;
1705         }
1706         # getCommon() returns the length of the longest common substring
1707         # of both arguments, starting at the beginning of both.
1708         #
1709         /* private */ function getCommon( $st1, $st2 ) {
1710                 $fl = strlen( $st1 );
1711                 $shorter = strlen( $st2 );
1712                 if ( $fl < $shorter ) { $shorter = $fl; }
1713
1714                 for ( $i = 0; $i < $shorter; ++$i ) {
1715                         if ( $st1{$i} != $st2{$i} ) { break; }
1716                 }
1717                 return $i;
1718         }
1719         # These next three functions open, continue, and close the list
1720         # element appropriate to the prefix character passed into them.
1721         #
1722         /* private */ function openList( $char ) {
1723                 $result = $this->closeParagraph();
1724
1725                 if ( '*' == $char ) { $result .= '<ul><li>'; }
1726                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
1727                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
1728                 else if ( ';' == $char ) {
1729                         $result .= '<dl><dt>';
1730                         $this->mDTopen = true;
1731                 }
1732                 else { $result = '<!-- ERR 1 -->'; }
1733
1734                 return $result;
1735         }
1736
1737         /* private */ function nextItem( $char ) {
1738                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
1739                 else if ( ':' == $char || ';' == $char ) {
1740                         $close = '</dd>';
1741                         if ( $this->mDTopen ) { $close = '</dt>'; }
1742                         if ( ';' == $char ) {
1743                                 $this->mDTopen = true;
1744                                 return $close . '<dt>';
1745                         } else {
1746                                 $this->mDTopen = false;
1747                                 return $close . '<dd>';
1748                         }
1749                 }
1750                 return '<!-- ERR 2 -->';
1751         }
1752
1753         /* private */ function closeList( $char ) {
1754                 if ( '*' == $char ) { $text = '</li></ul>'; }
1755                 else if ( '#' == $char ) { $text = '</li></ol>'; }
1756                 else if ( ':' == $char ) {
1757                         if ( $this->mDTopen ) {
1758                                 $this->mDTopen = false;
1759                                 $text = '</dt></dl>';
1760                         } else {
1761                                 $text = '</dd></dl>';
1762                         }
1763                 }
1764                 else {  return '<!-- ERR 3 -->'; }
1765                 return $text."\n";
1766         }
1767         /**#@-*/
1768
1769         /**
1770          * Make lists from lines starting with ':', '*', '#', etc.
1771          *
1772          * @access private
1773          * @return string the lists rendered as HTML
1774          */
1775         function doBlockLevels( $text, $linestart ) {
1776                 $fname = 'Parser::doBlockLevels';
1777                 wfProfileIn( $fname );
1778
1779                 # Parsing through the text line by line.  The main thing
1780                 # happening here is handling of block-level elements p, pre,
1781                 # and making lists from lines starting with * # : etc.
1782                 #
1783                 $textLines = explode( "\n", $text );
1784
1785                 $lastPrefix = $output = '';
1786                 $this->mDTopen = $inBlockElem = false;
1787                 $prefixLength = 0;
1788                 $paragraphStack = false;
1789
1790                 if ( !$linestart ) {
1791                         $output .= array_shift( $textLines );
1792                 }
1793                 foreach ( $textLines as $oLine ) {
1794                         $lastPrefixLength = strlen( $lastPrefix );
1795                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
1796                         $preOpenMatch = preg_match('/<pre/i', $oLine );
1797                         if ( !$this->mInPre ) {
1798                                 # Multiple prefixes may abut each other for nested lists.
1799                                 $prefixLength = strspn( $oLine, '*#:;' );
1800                                 $pref = substr( $oLine, 0, $prefixLength );
1801
1802                                 # eh?
1803                                 $pref2 = str_replace( ';', ':', $pref );
1804                                 $t = substr( $oLine, $prefixLength );
1805                                 $this->mInPre = !empty($preOpenMatch);
1806                         } else {
1807                                 # Don't interpret any other prefixes in preformatted text
1808                                 $prefixLength = 0;
1809                                 $pref = $pref2 = '';
1810                                 $t = $oLine;
1811                         }
1812
1813                         # List generation
1814                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
1815                                 # Same as the last item, so no need to deal with nesting or opening stuff
1816                                 $output .= $this->nextItem( substr( $pref, -1 ) );
1817                                 $paragraphStack = false;
1818
1819                                 if ( substr( $pref, -1 ) == ';') {
1820                                         # The one nasty exception: definition lists work like this:
1821                                         # ; title : definition text
1822                                         # So we check for : in the remainder text to split up the
1823                                         # title and definition, without b0rking links.
1824                                         $term = $t2 = '';
1825                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
1826                                                 $t = $t2;
1827                                                 $output .= $term . $this->nextItem( ':' );
1828                                         }
1829                                 }
1830                         } elseif( $prefixLength || $lastPrefixLength ) {
1831                                 # Either open or close a level...
1832                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
1833                                 $paragraphStack = false;
1834
1835                                 while( $commonPrefixLength < $lastPrefixLength ) {
1836                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
1837                                         --$lastPrefixLength;
1838                                 }
1839                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
1840                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
1841                                 }
1842                                 while ( $prefixLength > $commonPrefixLength ) {
1843                                         $char = substr( $pref, $commonPrefixLength, 1 );
1844                                         $output .= $this->openList( $char );
1845
1846                                         if ( ';' == $char ) {
1847                                                 # FIXME: This is dupe of code above
1848                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
1849                                                         $t = $t2;
1850                                                         $output .= $term . $this->nextItem( ':' );
1851                                                 }
1852                                         }
1853                                         ++$commonPrefixLength;
1854                                 }
1855                                 $lastPrefix = $pref2;
1856                         }
1857                         if( 0 == $prefixLength ) {
1858                                 wfProfileIn( "$fname-paragraph" );
1859                                 # No prefix (not in list)--go to paragraph mode
1860                                 // XXX: use a stack for nestable elements like span, table and div
1861                                 $openmatch = preg_match('/(<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
1862                                 $closematch = preg_match(
1863                                         '/(<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
1864                                         '<td|<th|<div|<\\/div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul)/iS', $t );
1865                                 if ( $openmatch or $closematch ) {
1866                                         $paragraphStack = false;
1867                                         $output .= $this->closeParagraph();
1868                                         if ( $preOpenMatch and !$preCloseMatch ) {
1869                                                 $this->mInPre = true;
1870                                         }
1871                                         if ( $closematch ) {
1872                                                 $inBlockElem = false;
1873                                         } else {
1874                                                 $inBlockElem = true;
1875                                         }
1876                                 } else if ( !$inBlockElem && !$this->mInPre ) {
1877                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
1878                                                 // pre
1879                                                 if ($this->mLastSection != 'pre') {
1880                                                         $paragraphStack = false;
1881                                                         $output .= $this->closeParagraph().'<pre>';
1882                                                         $this->mLastSection = 'pre';
1883                                                 }
1884                                                 $t = substr( $t, 1 );
1885                                         } else {
1886                                                 // paragraph
1887                                                 if ( '' == trim($t) ) {
1888                                                         if ( $paragraphStack ) {
1889                                                                 $output .= $paragraphStack.'<br />';
1890                                                                 $paragraphStack = false;
1891                                                                 $this->mLastSection = 'p';
1892                                                         } else {
1893                                                                 if ($this->mLastSection != 'p' ) {
1894                                                                         $output .= $this->closeParagraph();
1895                                                                         $this->mLastSection = '';
1896                                                                         $paragraphStack = '<p>';
1897                                                                 } else {
1898                                                                         $paragraphStack = '</p><p>';
1899                                                                 }
1900                                                         }
1901                                                 } else {
1902                                                         if ( $paragraphStack ) {
1903                                                                 $output .= $paragraphStack;
1904                                                                 $paragraphStack = false;
1905                                                                 $this->mLastSection = 'p';
1906                                                         } else if ($this->mLastSection != 'p') {
1907                                                                 $output .= $this->closeParagraph().'<p>';
1908                                                                 $this->mLastSection = 'p';
1909                                                         }
1910                                                 }
1911                                         }
1912                                 }
1913                                 wfProfileOut( "$fname-paragraph" );
1914                         }
1915                         // somewhere above we forget to get out of pre block (bug 785)
1916                         if($preCloseMatch && $this->mInPre) {
1917                                 $this->mInPre = false;
1918                         }
1919                         if ($paragraphStack === false) {
1920                                 $output .= $t."\n";
1921                         }
1922                 }
1923                 while ( $prefixLength ) {
1924                         $output .= $this->closeList( $pref2{$prefixLength-1} );
1925                         --$prefixLength;
1926                 }
1927                 if ( '' != $this->mLastSection ) {
1928                         $output .= '</' . $this->mLastSection . '>';
1929                         $this->mLastSection = '';
1930                 }
1931
1932                 wfProfileOut( $fname );
1933                 return $output;
1934         }
1935
1936         /**
1937          * Split up a string on ':', ignoring any occurences inside
1938          * <a>..</a> or <span>...</span>
1939          * @param string $str the string to split
1940          * @param string &$before set to everything before the ':'
1941          * @param string &$after set to everything after the ':'
1942          * return string the position of the ':', or false if none found
1943          */
1944         function findColonNoLinks($str, &$before, &$after) {
1945                 # I wonder if we should make this count all tags, not just <a>
1946                 # and <span>. That would prevent us from matching a ':' that
1947                 # comes in the middle of italics other such formatting....
1948                 # -- Wil
1949                 $fname = 'Parser::findColonNoLinks';
1950                 wfProfileIn( $fname );
1951                 $pos = 0;
1952                 do {
1953                         $colon = strpos($str, ':', $pos);
1954
1955                         if ($colon !== false) {
1956                                 $before = substr($str, 0, $colon);
1957                                 $after = substr($str, $colon + 1);
1958
1959                                 # Skip any ':' within <a> or <span> pairs
1960                                 $a = substr_count($before, '<a');
1961                                 $s = substr_count($before, '<span');
1962                                 $ca = substr_count($before, '</a>');
1963                                 $cs = substr_count($before, '</span>');
1964
1965                                 if ($a <= $ca and $s <= $cs) {
1966                                         # Tags are balanced before ':'; ok
1967                                         break;
1968                                 }
1969                                 $pos = $colon + 1;
1970                         }
1971                 } while ($colon !== false);
1972                 wfProfileOut( $fname );
1973                 return $colon;
1974         }
1975
1976         /**
1977          * Return value of a magic variable (like PAGENAME)
1978          *
1979          * @access private
1980          */
1981         function getVariableValue( $index ) {
1982                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
1983
1984                 /**
1985                  * Some of these require message or data lookups and can be
1986                  * expensive to check many times.
1987                  */
1988                 static $varCache = array();
1989                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$varCache ) ) )
1990                         if ( isset( $varCache[$index] ) )
1991                                 return $varCache[$index];
1992
1993                 $ts = time();
1994                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
1995
1996                 switch ( $index ) {
1997                         case MAG_CURRENTMONTH:
1998                                 return $varCache[$index] = $wgContLang->formatNum( date( 'm', $ts ) );
1999                         case MAG_CURRENTMONTHNAME:
2000                                 return $varCache[$index] = $wgContLang->getMonthName( date( 'n', $ts ) );
2001                         case MAG_CURRENTMONTHNAMEGEN:
2002                                 return $varCache[$index] = $wgContLang->getMonthNameGen( date( 'n', $ts ) );
2003                         case MAG_CURRENTMONTHABBREV:
2004                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( date( 'n', $ts ) );
2005                         case MAG_CURRENTDAY:
2006                                 return $varCache[$index] = $wgContLang->formatNum( date( 'j', $ts ) );
2007                         case MAG_CURRENTDAY2:
2008                                 return $varCache[$index] = $wgContLang->formatNum( date( 'd', $ts ) );
2009                         case MAG_PAGENAME:
2010                                 return $this->mTitle->getText();
2011                         case MAG_PAGENAMEE:
2012                                 return $this->mTitle->getPartialURL();
2013                         case MAG_FULLPAGENAME:
2014                                 return $this->mTitle->getPrefixedText();
2015                         case MAG_FULLPAGENAMEE:
2016                                 return $this->mTitle->getPrefixedURL();
2017                         case MAG_REVISIONID:
2018                                 return $this->mRevisionId;
2019                         case MAG_NAMESPACE:
2020                                 return $wgContLang->getNsText( $this->mTitle->getNamespace() );
2021                         case MAG_NAMESPACEE:
2022                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2023                         case MAG_CURRENTDAYNAME:
2024                                 return $varCache[$index] = $wgContLang->getWeekdayName( date( 'w', $ts ) + 1 );
2025                         case MAG_CURRENTYEAR:
2026                                 return $varCache[$index] = $wgContLang->formatNum( date( 'Y', $ts ), true );
2027                         case MAG_CURRENTTIME:
2028                                 return $varCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2029                         case MAG_CURRENTWEEK:
2030                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2031                                 // int to remove the padding
2032                                 return $varCache[$index] = $wgContLang->formatNum( (int)date( 'W', $ts ) );
2033                         case MAG_CURRENTDOW:
2034                                 return $varCache[$index] = $wgContLang->formatNum( date( 'w', $ts ) );
2035                         case MAG_NUMBEROFARTICLES:
2036                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfArticles() );
2037                         case MAG_NUMBEROFFILES:
2038                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfFiles() );
2039                         case MAG_SITENAME:
2040                                 return $wgSitename;
2041                         case MAG_SERVER:
2042                                 return $wgServer;
2043                         case MAG_SERVERNAME:
2044                                 return $wgServerName;
2045                         case MAG_SCRIPTPATH:
2046                                 return $wgScriptPath;
2047                         default:
2048                                 $ret = null;
2049                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$varCache, &$index, &$ret ) ) )
2050                                         return $ret;
2051                                 else
2052                                         return null;
2053                 }
2054         }
2055
2056         /**
2057          * initialise the magic variables (like CURRENTMONTHNAME)
2058          *
2059          * @access private
2060          */
2061         function initialiseVariables() {
2062                 $fname = 'Parser::initialiseVariables';
2063                 wfProfileIn( $fname );
2064                 global $wgVariableIDs;
2065                 $this->mVariables = array();
2066                 foreach ( $wgVariableIDs as $id ) {
2067                         $mw =& MagicWord::get( $id );
2068                         $mw->addToArray( $this->mVariables, $id );
2069                 }
2070                 wfProfileOut( $fname );
2071         }
2072
2073         /**
2074          * parse any parentheses in format ((title|part|part))
2075          * and call callbacks to get a replacement text for any found piece
2076          *
2077          * @param string $text The text to parse
2078          * @param array $callbacks rules in form:
2079          *     '{' => array(                            # opening parentheses
2080          *                                      'end' => '}',   # closing parentheses
2081          *                                      'cb' => array(2 => callback,    # replacement callback to call if {{..}} is found
2082          *                                                                4 => callback         # replacement callback to call if {{{{..}}}} is found
2083          *                                                                )
2084          *                                      )
2085          * @access private
2086          */
2087         function replace_callback ($text, $callbacks) {
2088                 $openingBraceStack = array();   # this array will hold a stack of parentheses which are not closed yet
2089                 $lastOpeningBrace = -1;                 # last not closed parentheses
2090
2091                 for ($i = 0; $i < strlen($text); $i++) {
2092                         # check for any opening brace
2093                         $rule = null;
2094                         $nextPos = -1;
2095                         foreach ($callbacks as $key => $value) {
2096                                 $pos = strpos ($text, $key, $i);
2097                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)) {
2098                                         $rule = $value;
2099                                         $nextPos = $pos;
2100                                 }
2101                         }
2102
2103                         if ($lastOpeningBrace >= 0) {
2104                                 $pos = strpos ($text, $openingBraceStack[$lastOpeningBrace]['braceEnd'], $i);
2105
2106                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)){
2107                                         $rule = null;
2108                                         $nextPos = $pos;
2109                                 }
2110
2111                                 $pos = strpos ($text, '|', $i);
2112
2113                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)){
2114                                         $rule = null;
2115                                         $nextPos = $pos;
2116                                 }
2117                         }
2118
2119                         if ($nextPos == -1)
2120                                 break;
2121
2122                         $i = $nextPos;
2123
2124                         # found openning brace, lets add it to parentheses stack
2125                         if (null != $rule) {
2126                                 $piece = array('brace' => $text[$i],
2127                                                            'braceEnd' => $rule['end'],
2128                                                            'count' => 1,
2129                                                            'title' => '',
2130                                                            'parts' => null);
2131
2132                                 # count openning brace characters
2133                                 while ($i+1 < strlen($text) && $text[$i+1] == $piece['brace']) {
2134                                         $piece['count']++;
2135                                         $i++;
2136                                 }
2137
2138                                 $piece['startAt'] = $i+1;
2139                                 $piece['partStart'] = $i+1;
2140
2141                                 # we need to add to stack only if openning brace count is enough for any given rule
2142                                 foreach ($rule['cb'] as $cnt => $fn) {
2143                                         if ($piece['count'] >= $cnt) {
2144                                                 $lastOpeningBrace ++;
2145                                                 $openingBraceStack[$lastOpeningBrace] = $piece;
2146                                                 break;
2147                                         }
2148                                 }
2149
2150                                 continue;
2151                         }
2152                         else if ($lastOpeningBrace >= 0) {
2153                                 # first check if it is a closing brace
2154                                 if ($openingBraceStack[$lastOpeningBrace]['braceEnd'] == $text[$i]) {
2155                                         # lets check if it is enough characters for closing brace
2156                                         $count = 1;
2157                                         while ($i+$count < strlen($text) && $text[$i+$count] == $text[$i])
2158                                                 $count++;
2159
2160                                         # if there are more closing parentheses than opening ones, we parse less
2161                                         if ($openingBraceStack[$lastOpeningBrace]['count'] < $count)
2162                                                 $count = $openingBraceStack[$lastOpeningBrace]['count'];
2163
2164                                         # check for maximum matching characters (if there are 5 closing characters, we will probably need only 3 - depending on the rules)
2165                                         $matchingCount = 0;
2166                                         $matchingCallback = null;
2167                                         foreach ($callbacks[$openingBraceStack[$lastOpeningBrace]['brace']]['cb'] as $cnt => $fn) {
2168                                                 if ($count >= $cnt && $matchingCount < $cnt) {
2169                                                         $matchingCount = $cnt;
2170                                                         $matchingCallback = $fn;
2171                                                 }
2172                                         }
2173
2174                                         if ($matchingCount == 0) {
2175                                                 $i += $count - 1;
2176                                                 continue;
2177                                         }
2178
2179                                         # lets set a title or last part (if '|' was found)
2180                                         if (null === $openingBraceStack[$lastOpeningBrace]['parts'])
2181                                                 $openingBraceStack[$lastOpeningBrace]['title'] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2182                                         else
2183                                                 $openingBraceStack[$lastOpeningBrace]['parts'][] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2184
2185                                         $pieceStart = $openingBraceStack[$lastOpeningBrace]['startAt'] - $matchingCount;
2186                                         $pieceEnd = $i + $matchingCount;
2187
2188                                         if( is_callable( $matchingCallback ) ) {
2189                                                 $cbArgs = array (
2190                                                                                  'text' => substr($text, $pieceStart, $pieceEnd - $pieceStart),
2191                                                                                  'title' => trim($openingBraceStack[$lastOpeningBrace]['title']),
2192                                                                                  'parts' => $openingBraceStack[$lastOpeningBrace]['parts'],
2193                                                                                  'lineStart' => (($pieceStart > 0) && ($text[$pieceStart-1] == '\n')),
2194                                                                                  );
2195                                                 # finally we can call a user callback and replace piece of text
2196                                                 $replaceWith = call_user_func( $matchingCallback, $cbArgs );
2197                                                 $text = substr($text, 0, $pieceStart) . $replaceWith . substr($text, $pieceEnd);
2198                                                 $i = $pieceStart + strlen($replaceWith) - 1;
2199                                         }
2200                                         else {
2201                                                 # null value for callback means that parentheses should be parsed, but not replaced
2202                                                 $i += $matchingCount - 1;
2203                                         }
2204
2205                                         # reset last openning parentheses, but keep it in case there are unused characters
2206                                         $piece = array('brace' => $openingBraceStack[$lastOpeningBrace]['brace'],
2207                                                                    'braceEnd' => $openingBraceStack[$lastOpeningBrace]['braceEnd'],
2208                                                                    'count' => $openingBraceStack[$lastOpeningBrace]['count'],
2209                                                                    'title' => '',
2210                                                                    'parts' => null,
2211                                                                    'startAt' => $openingBraceStack[$lastOpeningBrace]['startAt']);
2212                                         $openingBraceStack[$lastOpeningBrace--] = null;
2213
2214                                         if ($matchingCount < $piece['count']) {
2215                                                 $piece['count'] -= $matchingCount;
2216                                                 $piece['startAt'] -= $matchingCount;
2217                                                 $piece['partStart'] = $piece['startAt'];
2218                                                 # do we still qualify for any callback with remaining count?
2219                                                 foreach ($callbacks[$piece['brace']]['cb'] as $cnt => $fn) {
2220                                                         if ($piece['count'] >= $cnt) {
2221                                                                 $lastOpeningBrace ++;
2222                                                                 $openingBraceStack[$lastOpeningBrace] = $piece;
2223                                                                 break;
2224                                                         }
2225                                                 }
2226                                         }
2227                                         continue;
2228                                 }
2229
2230                                 # lets set a title if it is a first separator, or next part otherwise
2231                                 if ($text[$i] == '|') {
2232                                         if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2233                                                 $openingBraceStack[$lastOpeningBrace]['title'] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2234                                                 $openingBraceStack[$lastOpeningBrace]['parts'] = array();
2235                                         }
2236                                         else
2237                                                 $openingBraceStack[$lastOpeningBrace]['parts'][] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2238
2239                                         $openingBraceStack[$lastOpeningBrace]['partStart'] = $i + 1;
2240                                 }
2241                         }
2242                 }
2243
2244                 return $text;
2245         }
2246
2247         /**
2248          * Replace magic variables, templates, and template arguments
2249          * with the appropriate text. Templates are substituted recursively,
2250          * taking care to avoid infinite loops.
2251          *
2252          * Note that the substitution depends on value of $mOutputType:
2253          *  OT_WIKI: only {{subst:}} templates
2254          *  OT_MSG: only magic variables
2255          *  OT_HTML: all templates and magic variables
2256          *
2257          * @param string $tex The text to transform
2258          * @param array $args Key-value pairs representing template parameters to substitute
2259          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2260          * @access private
2261          */
2262         function replaceVariables( $text, $args = array(), $argsOnly = false ) {
2263                 # Prevent too big inclusions
2264                 if( strlen( $text ) > MAX_INCLUDE_SIZE ) {
2265                         return $text;
2266                 }
2267
2268                 $fname = 'Parser::replaceVariables';
2269                 wfProfileIn( $fname );
2270
2271                 # This function is called recursively. To keep track of arguments we need a stack:
2272                 array_push( $this->mArgStack, $args );
2273
2274                 $braceCallbacks = array();
2275                 if ( !$argsOnly ) {
2276                         $braceCallbacks[2] = array( &$this, 'braceSubstitution' );
2277                 }
2278                 if ( $this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI ) {
2279                         $braceCallbacks[3] = array( &$this, 'argSubstitution' );
2280                 }
2281                 $callbacks = array();
2282                 $callbacks['{'] = array('end' => '}', 'cb' => $braceCallbacks);
2283                 $callbacks['['] = array('end' => ']', 'cb' => array(2=>null));
2284                 $text = $this->replace_callback ($text, $callbacks);
2285
2286                 array_pop( $this->mArgStack );
2287
2288                 wfProfileOut( $fname );
2289                 return $text;
2290         }
2291
2292         /**
2293          * Replace magic variables
2294          * @access private
2295          */
2296         function variableSubstitution( $matches ) {
2297                 $fname = 'Parser::variableSubstitution';
2298                 $varname = $matches[1];
2299                 wfProfileIn( $fname );
2300                 if ( !$this->mVariables ) {
2301                         $this->initialiseVariables();
2302                 }
2303                 $skip = false;
2304                 if ( $this->mOutputType == OT_WIKI ) {
2305                         # Do only magic variables prefixed by SUBST
2306                         $mwSubst =& MagicWord::get( MAG_SUBST );
2307                         if (!$mwSubst->matchStartAndRemove( $varname ))
2308                                 $skip = true;
2309                         # Note that if we don't substitute the variable below,
2310                         # we don't remove the {{subst:}} magic word, in case
2311                         # it is a template rather than a magic variable.
2312                 }
2313                 if ( !$skip && array_key_exists( $varname, $this->mVariables ) ) {
2314                         $id = $this->mVariables[$varname];
2315                         $text = $this->getVariableValue( $id );
2316                         $this->mOutput->mContainsOldMagic = true;
2317                 } else {
2318                         $text = $matches[0];
2319                 }
2320                 wfProfileOut( $fname );
2321                 return $text;
2322         }
2323
2324         # Split template arguments
2325         function getTemplateArgs( $argsString ) {
2326                 if ( $argsString === '' ) {
2327                         return array();
2328                 }
2329
2330                 $args = explode( '|', substr( $argsString, 1 ) );
2331
2332                 # If any of the arguments contains a '[[' but no ']]', it needs to be
2333                 # merged with the next arg because the '|' character between belongs
2334                 # to the link syntax and not the template parameter syntax.
2335                 $argc = count($args);
2336
2337                 for ( $i = 0; $i < $argc-1; $i++ ) {
2338                         if ( substr_count ( $args[$i], '[[' ) != substr_count ( $args[$i], ']]' ) ) {
2339                                 $args[$i] .= '|'.$args[$i+1];
2340                                 array_splice($args, $i+1, 1);
2341                                 $i--;
2342                                 $argc--;
2343                         }
2344                 }
2345
2346                 return $args;
2347         }
2348
2349         /**
2350          * Return the text of a template, after recursively
2351          * replacing any variables or templates within the template.
2352          *
2353          * @param array $piece The parts of the template
2354          *  $piece['text']: matched text
2355          *  $piece['title']: the title, i.e. the part before the |
2356          *  $piece['parts']: the parameter array
2357          * @return string the text of the template
2358          * @access private
2359          */
2360         function braceSubstitution( $piece ) {
2361                 global $wgContLang;
2362                 $fname = 'Parser::braceSubstitution';
2363                 wfProfileIn( $fname );
2364
2365                 # Flags
2366                 $found = false;             # $text has been filled
2367                 $nowiki = false;            # wiki markup in $text should be escaped
2368                 $noparse = false;           # Unsafe HTML tags should not be stripped, etc.
2369                 $noargs = false;            # Don't replace triple-brace arguments in $text
2370                 $replaceHeadings = false;   # Make the edit section links go to the template not the article
2371                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2372                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2373
2374                 # Title object, where $text came from
2375                 $title = NULL;
2376
2377                 $linestart = '';
2378
2379                 # $part1 is the bit before the first |, and must contain only title characters
2380                 # $args is a list of arguments, starting from index 0, not including $part1
2381
2382                 $part1 = $piece['title'];
2383                 # If the third subpattern matched anything, it will start with |
2384
2385                 if (null == $piece['parts']) {
2386                         $replaceWith = $this->variableSubstitution (array ($piece['text'], $piece['title']));
2387                         if ($replaceWith != $piece['text']) {
2388                                 $text = $replaceWith;
2389                                 $found = true;
2390                                 $noparse = true;
2391                                 $noargs = true;
2392                         }
2393                 }
2394
2395                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2396                 $argc = count( $args );
2397
2398                 # SUBST
2399                 if ( !$found ) {
2400                         $mwSubst =& MagicWord::get( MAG_SUBST );
2401                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor ($this->mOutputType == OT_WIKI) ) {
2402                                 # One of two possibilities is true:
2403                                 # 1) Found SUBST but not in the PST phase
2404                                 # 2) Didn't find SUBST and in the PST phase
2405                                 # In either case, return without further processing
2406                                 $text = $piece['text'];
2407                                 $found = true;
2408                                 $noparse = true;
2409                                 $noargs = true;
2410                         }
2411                 }
2412
2413                 # MSG, MSGNW, INT and RAW
2414                 if ( !$found ) {
2415                         # Check for MSGNW:
2416                         $mwMsgnw =& MagicWord::get( MAG_MSGNW );
2417                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2418                                 $nowiki = true;
2419                         } else {
2420                                 # Remove obsolete MSG:
2421                                 $mwMsg =& MagicWord::get( MAG_MSG );
2422                                 $mwMsg->matchStartAndRemove( $part1 );
2423                         }
2424
2425                         # Check for RAW:
2426                         $mwRaw =& MagicWord::get( MAG_RAW );
2427                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2428                                 $forceRawInterwiki = true;
2429                         }
2430
2431                         # Check if it is an internal message
2432                         $mwInt =& MagicWord::get( MAG_INT );
2433                         if ( $mwInt->matchStartAndRemove( $part1 ) ) {
2434                                 if ( $this->incrementIncludeCount( 'int:'.$part1 ) ) {
2435                                         $text = $linestart . wfMsgReal( $part1, $args, true );
2436                                         $found = true;
2437                                 }
2438                         }
2439                 }
2440
2441                 # NS
2442                 if ( !$found ) {
2443                         # Check for NS: (namespace expansion)
2444                         $mwNs = MagicWord::get( MAG_NS );
2445                         if ( $mwNs->matchStartAndRemove( $part1 ) ) {
2446                                 if ( intval( $part1 ) || $part1 == "0" ) {
2447                                         $text = $linestart . $wgContLang->getNsText( intval( $part1 ) );
2448                                         $found = true;
2449                                 } else {
2450                                         $index = Namespace::getCanonicalIndex( strtolower( $part1 ) );
2451                                         if ( !is_null( $index ) ) {
2452                                                 $text = $linestart . $wgContLang->getNsText( $index );
2453                                                 $found = true;
2454                                         }
2455                                 }
2456                         }
2457                 }
2458
2459                 # LCFIRST, UCFIRST, LC and UC
2460                 if ( !$found ) {
2461                         $lcfirst =& MagicWord::get( MAG_LCFIRST );
2462                         $ucfirst =& MagicWord::get( MAG_UCFIRST );
2463                         $lc =& MagicWord::get( MAG_LC );
2464                         $uc =& MagicWord::get( MAG_UC );
2465                         if ( $lcfirst->matchStartAndRemove( $part1 ) ) {
2466                                 $text = $linestart . $wgContLang->lcfirst( $part1 );
2467                                 $found = true;
2468                         } else if ( $ucfirst->matchStartAndRemove( $part1 ) ) {
2469                                 $text = $linestart . $wgContLang->ucfirst( $part1 );
2470                                 $found = true;
2471                         } else if ( $lc->matchStartAndRemove( $part1 ) ) {
2472                                 $text = $linestart . $wgContLang->lc( $part1 );
2473                                 $found = true;
2474                         } else if ( $uc->matchStartAndRemove( $part1 ) ) {
2475                                  $text = $linestart . $wgContLang->uc( $part1 );
2476                                  $found = true;
2477                         }
2478                 }
2479
2480                 # LOCALURL and FULLURL
2481                 if ( !$found ) {
2482                         $mwLocal =& MagicWord::get( MAG_LOCALURL );
2483                         $mwLocalE =& MagicWord::get( MAG_LOCALURLE );
2484                         $mwFull =& MagicWord::get( MAG_FULLURL );
2485                         $mwFullE =& MagicWord::get( MAG_FULLURLE );
2486
2487
2488                         if ( $mwLocal->matchStartAndRemove( $part1 ) ) {
2489                                 $func = 'getLocalURL';
2490                         } elseif ( $mwLocalE->matchStartAndRemove( $part1 ) ) {
2491                                 $func = 'escapeLocalURL';
2492                         } elseif ( $mwFull->matchStartAndRemove( $part1 ) ) {
2493                                 $func = 'getFullURL';
2494                         } elseif ( $mwFullE->matchStartAndRemove( $part1 ) ) {
2495                                 $func = 'escapeFullURL';
2496                         } else {
2497                                 $func = false;
2498                         }
2499
2500                         if ( $func !== false ) {
2501                                 $title = Title::newFromText( $part1 );
2502                                 if ( !is_null( $title ) ) {
2503                                         if ( $argc > 0 ) {
2504                                                 $text = $linestart . $title->$func( $args[0] );
2505                                         } else {
2506                                                 $text = $linestart . $title->$func();
2507                                         }
2508                                         $found = true;
2509                                 }
2510                         }
2511                 }
2512
2513                 # GRAMMAR
2514                 if ( !$found && $argc == 1 ) {
2515                         $mwGrammar =& MagicWord::get( MAG_GRAMMAR );
2516                         if ( $mwGrammar->matchStartAndRemove( $part1 ) ) {
2517                                 $text = $linestart . $wgContLang->convertGrammar( $args[0], $part1 );
2518                                 $found = true;
2519                         }
2520                 }
2521
2522                 # PLURAL
2523                 if ( !$found && $argc >= 2 ) {
2524                         $mwPluralForm =& MagicWord::get( MAG_PLURAL );
2525                         if ( $mwPluralForm->matchStartAndRemove( $part1 ) ) {
2526                                 if ($argc==2) {$args[2]=$args[1];}
2527                                 $text = $linestart . $wgContLang->convertPlural( $part1, $args[0], $args[1], $args[2]);
2528                                 $found = true;
2529                         }
2530                 }
2531
2532                 # Template table test
2533
2534                 # Did we encounter this template already? If yes, it is in the cache
2535                 # and we need to check for loops.
2536                 if ( !$found && isset( $this->mTemplates[$piece['title']] ) ) {
2537                         $found = true;
2538
2539                         # Infinite loop test
2540                         if ( isset( $this->mTemplatePath[$part1] ) ) {
2541                                 $noparse = true;
2542                                 $noargs = true;
2543                                 $found = true;
2544                                 $text = $linestart .
2545                                         '{{' . $part1 . '}}' .
2546                                         '<!-- WARNING: template loop detected -->';
2547                                 wfDebug( "$fname: template loop broken at '$part1'\n" );
2548                         } else {
2549                                 # set $text to cached message.
2550                                 $text = $linestart . $this->mTemplates[$piece['title']];
2551                         }
2552                 }
2553
2554                 # Load from database
2555                 $lastPathLevel = $this->mTemplatePath;
2556                 if ( !$found ) {
2557                         $ns = NS_TEMPLATE;
2558                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage='' );
2559                         if ($subpage !== '') {
2560                                 $ns = $this->mTitle->getNamespace();
2561                         }
2562                         $title = Title::newFromText( $part1, $ns );
2563
2564                         if ( !is_null( $title ) ) {
2565                                 if ( !$title->isExternal() ) {
2566                                         # Check for excessive inclusion
2567                                         $dbk = $title->getPrefixedDBkey();
2568                                         if ( $this->incrementIncludeCount( $dbk ) ) {
2569                                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() ) {
2570                                                         # Capture special page output
2571                                                         $text = SpecialPage::capturePath( $title );
2572                                                         if ( is_string( $text ) ) {
2573                                                                 $found = true;
2574                                                                 $noparse = true;
2575                                                                 $noargs = true;
2576                                                                 $isHTML = true;
2577                                                                 $this->disableCache();
2578                                                         }
2579                                                 } else {
2580                                                         $articleContent = $this->fetchTemplate( $title );
2581                                                         if ( $articleContent !== false ) {
2582                                                                 $found = true;
2583                                                                 $text = $articleContent;
2584                                                                 $replaceHeadings = true;
2585                                                         }
2586                                                 }
2587                                         }
2588
2589                                         # If the title is valid but undisplayable, make a link to it
2590                                         if ( $this->mOutputType == OT_HTML && !$found ) {
2591                                                 $text = '[['.$title->getPrefixedText().']]';
2592                                                 $found = true;
2593                                         }
2594                                 } elseif ( $title->isTrans() ) {
2595                                         // Interwiki transclusion
2596                                         if ( $this->mOutputType == OT_HTML && !$forceRawInterwiki ) {
2597                                                 $text = $this->interwikiTransclude( $title, 'render' );
2598                                                 $isHTML = true;
2599                                                 $noparse = true;
2600                                         } else {
2601                                                 $text = $this->interwikiTransclude( $title, 'raw' );
2602                                                 $replaceHeadings = true;
2603                                         }
2604                                         $found = true;
2605                                 }
2606
2607                                 # Template cache array insertion
2608                                 # Use the original $piece['title'] not the mangled $part1, so that
2609                                 # modifiers such as RAW: produce separate cache entries
2610                                 if( $found ) {
2611                                         $this->mTemplates[$piece['title']] = $text;
2612                                         $text = $linestart . $text;
2613                                 }
2614                         }
2615                 }
2616
2617                 # Recursive parsing, escaping and link table handling
2618                 # Only for HTML output
2619                 if ( $nowiki && $found && $this->mOutputType == OT_HTML ) {
2620                         $text = wfEscapeWikiText( $text );
2621                 } elseif ( ($this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI) && $found ) {
2622                         if ( !$noargs ) {
2623                                 # Clean up argument array
2624                                 $assocArgs = array();
2625                                 $index = 1;
2626                                 foreach( $args as $arg ) {
2627                                         $eqpos = strpos( $arg, '=' );
2628                                         if ( $eqpos === false ) {
2629                                                 $assocArgs[$index++] = $arg;
2630                                         } else {
2631                                                 $name = trim( substr( $arg, 0, $eqpos ) );
2632                                                 $value = trim( substr( $arg, $eqpos+1 ) );
2633                                                 if ( $value === false ) {
2634                                                         $value = '';
2635                                                 }
2636                                                 if ( $name !== false ) {
2637                                                         $assocArgs[$name] = $value;
2638                                                 }
2639                                         }
2640                                 }
2641
2642                                 # Add a new element to the templace recursion path
2643                                 $this->mTemplatePath[$part1] = 1;
2644                         }
2645
2646                         if ( !$noparse ) {
2647                                 # If there are any <onlyinclude> tags, only include them
2648                                 if ( in_string( '<onlyinclude>', $text ) && in_string( '</onlyinclude>', $text ) ) {
2649                                         preg_match_all( '/<onlyinclude>(.*?)\n?<\/onlyinclude>/s', $text, $m );
2650                                         $text = '';
2651                                         foreach ($m[1] as $piece)
2652                                                 $text .= $piece;
2653                                 }
2654                                 # Remove <noinclude> sections and <includeonly> tags
2655                                 $text = preg_replace( '/<noinclude>.*?<\/noinclude>/s', '', $text );
2656                                 $text = strtr( $text, array( '<includeonly>' => '' , '</includeonly>' => '' ) );
2657
2658                                 if( $this->mOutputType == OT_HTML ) {
2659                                         # Strip <nowiki>, <pre>, etc.
2660                                         $text = $this->strip( $text, $this->mStripState );
2661                                         $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'replaceVariables' ), $assocArgs );
2662                                 }
2663                                 $text = $this->replaceVariables( $text, $assocArgs );
2664
2665                                 # If the template begins with a table or block-level
2666                                 # element, it should be treated as beginning a new line.
2667                                 if (!$piece['lineStart'] && preg_match('/^({\\||:|;|#|\*)/', $text)) {
2668                                         $text = "\n" . $text;
2669                                 }
2670                         } elseif ( !$noargs ) {
2671                                 # $noparse and !$noargs
2672                                 # Just replace the arguments, not any double-brace items
2673                                 # This is used for rendered interwiki transclusion
2674                                 $text = $this->replaceVariables( $text, $assocArgs, true );
2675                         }
2676                 }
2677                 # Prune lower levels off the recursion check path
2678                 $this->mTemplatePath = $lastPathLevel;
2679
2680                 if ( !$found ) {
2681                         wfProfileOut( $fname );
2682                         return $piece['text'];
2683                 } else {
2684                         if ( $isHTML ) {
2685                                 # Replace raw HTML by a placeholder
2686                                 # Add a blank line preceding, to prevent it from mucking up
2687                                 # immediately preceding headings
2688                                 $text = "\n\n" . $this->insertStripItem( $text, $this->mStripState );
2689                         } else {
2690                                 # replace ==section headers==
2691                                 # XXX this needs to go away once we have a better parser.
2692                                 if ( $this->mOutputType != OT_WIKI && $replaceHeadings ) {
2693                                         if( !is_null( $title ) )
2694                                                 $encodedname = base64_encode($title->getPrefixedDBkey());
2695                                         else
2696                                                 $encodedname = base64_encode("");
2697                                         $m = preg_split('/(^={1,6}.*?={1,6}\s*?$)/m', $text, -1,
2698                                                 PREG_SPLIT_DELIM_CAPTURE);
2699                                         $text = '';
2700                                         $nsec = 0;
2701                                         for( $i = 0; $i < count($m); $i += 2 ) {
2702                                                 $text .= $m[$i];
2703                                                 if (!isset($m[$i + 1]) || $m[$i + 1] == "") continue;
2704                                                 $hl = $m[$i + 1];
2705                                                 if( strstr($hl, "<!--MWTEMPLATESECTION") ) {
2706                                                         $text .= $hl;
2707                                                         continue;
2708                                                 }
2709                                                 preg_match('/^(={1,6})(.*?)(={1,6})\s*?$/m', $hl, $m2);
2710                                                 $text .= $m2[1] . $m2[2] . "<!--MWTEMPLATESECTION="
2711                                                         . $encodedname . "&" . base64_encode("$nsec") . "-->" . $m2[3];
2712
2713                                                 $nsec++;
2714                                         }
2715                                 }
2716                         }
2717                 }
2718
2719                 # Prune lower levels off the recursion check path
2720                 $this->mTemplatePath = $lastPathLevel;
2721
2722                 if ( !$found ) {
2723                         wfProfileOut( $fname );
2724                         return $piece['text'];
2725                 } else {
2726                         wfProfileOut( $fname );
2727                         return $text;
2728                 }
2729         }
2730
2731         /**
2732          * Fetch the unparsed text of a template and register a reference to it.
2733          */
2734         function fetchTemplate( $title ) {
2735                 $text = false;
2736                 // Loop to fetch the article, with up to 1 redirect
2737                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
2738                         $rev = Revision::newFromTitle( $title );
2739                         $this->mOutput->addTemplate( $title, $title->getArticleID() );
2740                         if ( !$rev ) {
2741                                 break;
2742                         }
2743                         $text = $rev->getText();
2744                         if ( $text === false ) {
2745                                 break;
2746                         }
2747                         // Redirect?
2748                         $title = Title::newFromRedirect( $text );
2749                 }
2750                 return $text;
2751         }
2752
2753         /**
2754          * Transclude an interwiki link.
2755          */
2756         function interwikiTransclude( $title, $action ) {
2757                 global $wgEnableScaryTranscluding, $wgCanonicalNamespaceNames;
2758
2759                 if (!$wgEnableScaryTranscluding)
2760                         return wfMsg('scarytranscludedisabled');
2761
2762                 // The namespace will actually only be 0 or 10, depending on whether there was a leading :
2763                 // But we'll handle it generally anyway
2764                 if ( $title->getNamespace() ) {
2765                         // Use the canonical namespace, which should work anywhere
2766                         $articleName = $wgCanonicalNamespaceNames[$title->getNamespace()] . ':' . $title->getDBkey();
2767                 } else {
2768                         $articleName = $title->getDBkey();
2769                 }
2770
2771                 $url = str_replace('$1', urlencode($articleName), Title::getInterwikiLink($title->getInterwiki()));
2772                 $url .= "?action=$action";
2773                 if (strlen($url) > 255)
2774                         return wfMsg('scarytranscludetoolong');
2775                 return $this->fetchScaryTemplateMaybeFromCache($url);
2776         }
2777
2778         function fetchScaryTemplateMaybeFromCache($url) {
2779                 global $wgTranscludeCacheExpiry;
2780                 $dbr =& wfGetDB(DB_SLAVE);
2781                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
2782                                 array('tc_url' => $url));
2783                 if ($obj) {
2784                         $time = $obj->tc_time;
2785                         $text = $obj->tc_contents;
2786                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
2787                                 return $text;
2788                         }
2789                 }
2790
2791                 $text = wfGetHTTP($url);
2792                 if (!$text)
2793                         return wfMsg('scarytranscludefailed', $url);
2794
2795                 $dbw =& wfGetDB(DB_MASTER);
2796                 $dbw->replace('transcache', array('tc_url'), array(
2797                         'tc_url' => $url,
2798                         'tc_time' => time(),
2799                         'tc_contents' => $text));
2800                 return $text;
2801         }
2802
2803
2804         /**
2805          * Triple brace replacement -- used for template arguments
2806          * @access private
2807          */
2808         function argSubstitution( $matches ) {
2809                 $arg = trim( $matches['title'] );
2810                 $text = $matches['text'];
2811                 $inputArgs = end( $this->mArgStack );
2812
2813                 if ( array_key_exists( $arg, $inputArgs ) ) {
2814                         $text = $inputArgs[$arg];
2815                 } else if ($this->mOutputType == OT_HTML && null != $matches['parts'] && count($matches['parts']) > 0) {
2816                         $text = $matches['parts'][0];
2817                 }
2818
2819                 return $text;
2820         }
2821
2822         /**
2823          * Returns true if the function is allowed to include this entity
2824          * @access private
2825          */
2826         function incrementIncludeCount( $dbk ) {
2827                 if ( !array_key_exists( $dbk, $this->mIncludeCount ) ) {
2828                         $this->mIncludeCount[$dbk] = 0;
2829                 }
2830                 if ( ++$this->mIncludeCount[$dbk] <= MAX_INCLUDE_REPEAT ) {
2831                         return true;
2832                 } else {
2833                         return false;
2834                 }
2835         }
2836
2837         /**
2838          * This function accomplishes several tasks:
2839          * 1) Auto-number headings if that option is enabled
2840          * 2) Add an [edit] link to sections for logged in users who have enabled the option
2841          * 3) Add a Table of contents on the top for users who have enabled the option
2842          * 4) Auto-anchor headings
2843          *
2844          * It loops through all headlines, collects the necessary data, then splits up the
2845          * string and re-inserts the newly formatted headlines.
2846          *
2847          * @param string $text
2848          * @param boolean $isMain
2849          * @access private
2850          */
2851         function formatHeadings( $text, $isMain=true ) {
2852                 global $wgMaxTocLevel, $wgContLang;
2853
2854                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
2855                 $doShowToc = true;
2856                 $forceTocHere = false;
2857                 if( !$this->mTitle->userCanEdit() ) {
2858                         $showEditLink = 0;
2859                 } else {
2860                         $showEditLink = $this->mOptions->getEditSection();
2861                 }
2862
2863                 # Inhibit editsection links if requested in the page
2864                 $esw =& MagicWord::get( MAG_NOEDITSECTION );
2865                 if( $esw->matchAndRemove( $text ) ) {
2866                         $showEditLink = 0;
2867                 }
2868                 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
2869                 # do not add TOC
2870                 $mw =& MagicWord::get( MAG_NOTOC );
2871                 if( $mw->matchAndRemove( $text ) ) {
2872                         $doShowToc = false;
2873                 }
2874
2875                 # Get all headlines for numbering them and adding funky stuff like [edit]
2876                 # links - this is for later, but we need the number of headlines right now
2877                 $numMatches = preg_match_all( '/<H([1-6])(.*?'.'>)(.*?)<\/H[1-6] *>/i', $text, $matches );
2878
2879                 # if there are fewer than 4 headlines in the article, do not show TOC
2880                 if( $numMatches < 4 ) {
2881                         $doShowToc = false;
2882                 }
2883
2884                 # if the string __TOC__ (not case-sensitive) occurs in the HTML,
2885                 # override above conditions and always show TOC at that place
2886
2887                 $mw =& MagicWord::get( MAG_TOC );
2888                 if($mw->match( $text ) ) {
2889                         $doShowToc = true;
2890                         $forceTocHere = true;
2891                 } else {
2892                         # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
2893                         # override above conditions and always show TOC above first header
2894                         $mw =& MagicWord::get( MAG_FORCETOC );
2895                         if ($mw->matchAndRemove( $text ) ) {
2896                                 $doShowToc = true;
2897                         }
2898                 }
2899
2900                 # Never ever show TOC if no headers
2901                 if( $numMatches < 1 ) {
2902                         $doShowToc = false;
2903                 }
2904
2905                 # We need this to perform operations on the HTML
2906                 $sk =& $this->mOptions->getSkin();
2907
2908                 # headline counter
2909                 $headlineCount = 0;
2910                 $sectionCount = 0; # headlineCount excluding template sections
2911
2912                 # Ugh .. the TOC should have neat indentation levels which can be
2913                 # passed to the skin functions. These are determined here
2914                 $toc = '';
2915                 $full = '';
2916                 $head = array();
2917                 $sublevelCount = array();
2918                 $levelCount = array();
2919                 $toclevel = 0;
2920                 $level = 0;
2921                 $prevlevel = 0;
2922                 $toclevel = 0;
2923                 $prevtoclevel = 0;
2924
2925                 foreach( $matches[3] as $headline ) {
2926                         $istemplate = 0;
2927                         $templatetitle = '';
2928                         $templatesection = 0;
2929                         $numbering = '';
2930
2931                         if (preg_match("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", $headline, $mat)) {
2932                                 $istemplate = 1;
2933                                 $templatetitle = base64_decode($mat[1]);
2934                                 $templatesection = 1 + (int)base64_decode($mat[2]);
2935                                 $headline = preg_replace("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", "", $headline);
2936                         }
2937
2938                         if( $toclevel ) {
2939                                 $prevlevel = $level;
2940                                 $prevtoclevel = $toclevel;
2941                         }
2942                         $level = $matches[1][$headlineCount];
2943
2944                         if( $doNumberHeadings || $doShowToc ) {
2945
2946                                 if ( $level > $prevlevel ) {
2947                                         # Increase TOC level
2948                                         $toclevel++;
2949                                         $sublevelCount[$toclevel] = 0;
2950                                         $toc .= $sk->tocIndent();
2951                                 }
2952                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
2953                                         # Decrease TOC level, find level to jump to
2954
2955                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
2956                                                 # Can only go down to level 1
2957                                                 $toclevel = 1;
2958                                         } else {
2959                                                 for ($i = $toclevel; $i > 0; $i--) {
2960                                                         if ( $levelCount[$i] == $level ) {
2961                                                                 # Found last matching level
2962                                                                 $toclevel = $i;
2963                                                                 break;
2964                                                         }
2965                                                         elseif ( $levelCount[$i] < $level ) {
2966                                                                 # Found first matching level below current level
2967                                                                 $toclevel = $i + 1;
2968                                                                 break;
2969                                                         }
2970                                                 }
2971                                         }
2972
2973                                         $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
2974                                 }
2975                                 else {
2976                                         # No change in level, end TOC line
2977                                         $toc .= $sk->tocLineEnd();
2978                                 }
2979
2980                                 $levelCount[$toclevel] = $level;
2981
2982                                 # count number of headlines for each level
2983                                 @$sublevelCount[$toclevel]++;
2984                                 $dot = 0;
2985                                 for( $i = 1; $i <= $toclevel; $i++ ) {
2986                                         if( !empty( $sublevelCount[$i] ) ) {
2987                                                 if( $dot ) {
2988                                                         $numbering .= '.';
2989                                                 }
2990                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
2991                                                 $dot = 1;
2992                                         }
2993                                 }
2994                         }
2995
2996                         # The canonized header is a version of the header text safe to use for links
2997                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
2998                         $canonized_headline = $this->unstrip( $headline, $this->mStripState );
2999                         $canonized_headline = $this->unstripNoWiki( $canonized_headline, $this->mStripState );
3000
3001                         # Remove link placeholders by the link text.
3002                         #     <!--LINK number-->
3003                         # turns into
3004                         #     link text with suffix
3005                         $canonized_headline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
3006                                                             "\$this->mLinkHolders['texts'][\$1]",
3007                                                             $canonized_headline );
3008                         $canonized_headline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
3009                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
3010                                                             $canonized_headline );
3011
3012                         # strip out HTML
3013                         $canonized_headline = preg_replace( '/<.*?' . '>/','',$canonized_headline );
3014                         $tocline = trim( $canonized_headline );
3015                         $canonized_headline = Sanitizer::escapeId( $tocline );
3016                         $refers[$headlineCount] = $canonized_headline;
3017
3018                         # count how many in assoc. array so we can track dupes in anchors
3019                         @$refers[$canonized_headline]++;
3020                         $refcount[$headlineCount]=$refers[$canonized_headline];
3021
3022                         # Don't number the heading if it is the only one (looks silly)
3023                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3024                                 # the two are different if the line contains a link
3025                                 $headline=$numbering . ' ' . $headline;
3026                         }
3027
3028                         # Create the anchor for linking from the TOC to the section
3029                         $anchor = $canonized_headline;
3030                         if($refcount[$headlineCount] > 1 ) {
3031                                 $anchor .= '_' . $refcount[$headlineCount];
3032                         }
3033                         if( $doShowToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3034                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3035                         }
3036                         if( $showEditLink && ( !$istemplate || $templatetitle !== "" ) ) {
3037                                 if ( empty( $head[$headlineCount] ) ) {
3038                                         $head[$headlineCount] = '';
3039                                 }
3040                                 if( $istemplate )
3041                                         $head[$headlineCount] .= $sk->editSectionLinkForOther($templatetitle, $templatesection);
3042                                 else
3043                                         $head[$headlineCount] .= $sk->editSectionLink($this->mTitle, $sectionCount+1);
3044                         }
3045
3046                         # give headline the correct <h#> tag
3047                         @$head[$headlineCount] .= "<a name=\"$anchor\"></a><h".$level.$matches[2][$headlineCount] .$headline.'</h'.$level.'>';
3048
3049                         $headlineCount++;
3050                         if( !$istemplate )
3051                                 $sectionCount++;
3052                 }
3053
3054                 if( $doShowToc ) {
3055                         $toc .= $sk->tocUnindent( $toclevel - 1 );
3056                         $toc = $sk->tocList( $toc );
3057                 }
3058
3059                 # split up and insert constructed headlines
3060
3061                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3062                 $i = 0;
3063
3064                 foreach( $blocks as $block ) {
3065                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
3066                                 # This is the [edit] link that appears for the top block of text when
3067                                 # section editing is enabled
3068
3069                                 # Disabled because it broke block formatting
3070                                 # For example, a bullet point in the top line
3071                                 # $full .= $sk->editSectionLink(0);
3072                         }
3073                         $full .= $block;
3074                         if( $doShowToc && !$i && $isMain && !$forceTocHere) {
3075                         # Top anchor now in skin
3076                                 $full = $full.$toc;
3077                         }
3078
3079                         if( !empty( $head[$i] ) ) {
3080                                 $full .= $head[$i];
3081                         }
3082                         $i++;
3083                 }
3084                 if($forceTocHere) {
3085                         $mw =& MagicWord::get( MAG_TOC );
3086                         return $mw->replace( $toc, $full );
3087                 } else {
3088                         return $full;
3089                 }
3090         }
3091
3092         /**
3093          * Return an HTML link for the "ISBN 123456" text
3094          * @access private
3095          */
3096         function magicISBN( $text ) {
3097                 $fname = 'Parser::magicISBN';
3098                 wfProfileIn( $fname );
3099
3100                 $a = split( 'ISBN ', ' '.$text );
3101                 if ( count ( $a ) < 2 ) {
3102                         wfProfileOut( $fname );
3103                         return $text;
3104                 }
3105                 $text = substr( array_shift( $a ), 1);
3106                 $valid = '0123456789-Xx';
3107
3108                 foreach ( $a as $x ) {
3109                         $isbn = $blank = '' ;
3110                         while ( ' ' == $x{0} ) {
3111                                 $blank .= ' ';
3112                                 $x = substr( $x, 1 );
3113                         }
3114                         if ( $x == '' ) { # blank isbn
3115                                 $text .= "ISBN $blank";
3116                                 continue;
3117                         }
3118                         while ( strstr( $valid, $x{0} ) != false ) {
3119                                 $isbn .= $x{0};
3120                                 $x = substr( $x, 1 );
3121                         }
3122                         $num = str_replace( '-', '', $isbn );
3123                         $num = str_replace( ' ', '', $num );
3124                         $num = str_replace( 'x', 'X', $num );
3125
3126                         if ( '' == $num ) {
3127                                 $text .= "ISBN $blank$x";
3128                         } else {
3129                                 $titleObj = Title::makeTitle( NS_SPECIAL, 'Booksources' );
3130                                 $text .= '<a href="' .
3131                                 $titleObj->escapeLocalUrl( 'isbn='.$num ) .
3132                                         "\" class=\"internal\">ISBN $isbn</a>";
3133                                 $text .= $x;
3134                         }
3135                 }
3136                 wfProfileOut( $fname );
3137                 return $text;
3138         }
3139
3140         /**
3141          * Return an HTML link for the "RFC 1234" text
3142          *
3143          * @access private
3144          * @param string $text     Text to be processed
3145          * @param string $keyword  Magic keyword to use (default RFC)
3146          * @param string $urlmsg   Interface message to use (default rfcurl)
3147          * @return string
3148          */
3149         function magicRFC( $text, $keyword='RFC ', $urlmsg='rfcurl'  ) {
3150
3151                 $valid = '0123456789';
3152                 $internal = false;
3153
3154                 $a = split( $keyword, ' '.$text );
3155                 if ( count ( $a ) < 2 ) {
3156                         return $text;
3157                 }
3158                 $text = substr( array_shift( $a ), 1);
3159
3160                 /* Check if keyword is preceed by [[.
3161                  * This test is made here cause of the array_shift above
3162                  * that prevent the test to be done in the foreach.
3163                  */
3164                 if ( substr( $text, -2 ) == '[[' ) {
3165                         $internal = true;
3166                 }
3167
3168                 foreach ( $a as $x ) {
3169                         /* token might be empty if we have RFC RFC 1234 */
3170                         if ( $x=='' ) {
3171                                 $text.=$keyword;
3172                                 continue;
3173                                 }
3174
3175                         $id = $blank = '' ;
3176
3177                         /** remove and save whitespaces in $blank */
3178                         while ( $x{0} == ' ' ) {
3179                                 $blank .= ' ';
3180                                 $x = substr( $x, 1 );
3181                         }
3182
3183                         /** remove and save the rfc number in $id */
3184                         while ( strstr( $valid, $x{0} ) != false ) {
3185                                 $id .= $x{0};
3186                                 $x = substr( $x, 1 );
3187                         }
3188
3189                         if ( $id == '' ) {
3190                                 /* call back stripped spaces*/
3191                                 $text .= $keyword.$blank.$x;
3192                         } elseif( $internal ) {
3193                                 /* normal link */
3194                                 $text .= $keyword.$id.$x;
3195                         } else {
3196                                 /* build the external link*/
3197                                 $url = wfMsg( $urlmsg, $id);
3198                                 $sk =& $this->mOptions->getSkin();
3199                                 $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
3200                                 $text .= "<a href='{$url}'{$la}>{$keyword}{$id}</a>{$x}";
3201                         }
3202
3203                         /* Check if the next RFC keyword is preceed by [[ */
3204                         $internal = ( substr($x,-2) == '[[' );
3205                 }
3206                 return $text;
3207         }
3208
3209         /**
3210          * Transform wiki markup when saving a page by doing \r\n -> \n
3211          * conversion, substitting signatures, {{subst:}} templates, etc.
3212          *
3213          * @param string $text the text to transform
3214          * @param Title &$title the Title object for the current article
3215          * @param User &$user the User object describing the current user
3216          * @param ParserOptions $options parsing options
3217          * @param bool $clearState whether to clear the parser state first
3218          * @return string the altered wiki markup
3219          * @access public
3220          */
3221         function preSaveTransform( $text, &$title, &$user, $options, $clearState = true ) {
3222                 $this->mOptions = $options;
3223                 $this->mTitle =& $title;
3224                 $this->mOutputType = OT_WIKI;
3225
3226                 if ( $clearState ) {
3227                         $this->clearState();
3228                 }
3229
3230                 $stripState = false;
3231                 $pairs = array(
3232                         "\r\n" => "\n",
3233                 );
3234                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3235                 $text = $this->strip( $text, $stripState, true );
3236                 $text = $this->pstPass2( $text, $user );
3237                 $text = $this->unstrip( $text, $stripState );
3238                 $text = $this->unstripNoWiki( $text, $stripState );
3239                 return $text;
3240         }
3241
3242         /**
3243          * Pre-save transform helper function
3244          * @access private
3245          */
3246         function pstPass2( $text, &$user ) {
3247                 global $wgContLang, $wgLocaltimezone;
3248
3249                 /* Note: This is the timestamp saved as hardcoded wikitext to
3250                  * the database, we use $wgContLang here in order to give
3251                  * everyone the same signiture and use the default one rather
3252                  * than the one selected in each users preferences.
3253                  */
3254                 if ( isset( $wgLocaltimezone ) ) {
3255                         $oldtz = getenv( 'TZ' );
3256                         putenv( 'TZ='.$wgLocaltimezone );
3257                 }
3258                 $d = $wgContLang->timeanddate( date( 'YmdHis' ), false, false) .
3259                   ' (' . date( 'T' ) . ')';
3260                 if ( isset( $wgLocaltimezone ) ) {
3261                         putenv( 'TZ='.$oldtz );
3262                 }
3263
3264                 # Variable replacement
3265                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3266                 $text = $this->replaceVariables( $text );
3267
3268                 # Signatures
3269                 $sigText = $this->getUserSig( $user );
3270                 $text = strtr( $text, array(
3271                         '~~~~~' => $d,
3272                         '~~~~' => "$sigText $d",
3273                         '~~~' => $sigText
3274                 ) );
3275
3276                 # Context links: [[|name]] and [[name (context)|]]
3277                 #
3278                 global $wgLegalTitleChars;
3279                 $tc = "[$wgLegalTitleChars]";
3280                 $np = str_replace( array( '(', ')' ), array( '', '' ), $tc ); # No parens
3281
3282                 $namespacechar = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
3283                 $conpat = "/^({$np}+) \\(({$tc}+)\\)$/";
3284
3285                 $p1 = "/\[\[({$np}+) \\(({$np}+)\\)\\|]]/";             # [[page (context)|]]
3286                 $p2 = "/\[\[\\|({$tc}+)]]/";                                    # [[|page]]
3287                 $p3 = "/\[\[(:*$namespacechar+):({$np}+)\\|]]/";                # [[namespace:page|]] and [[:namespace:page|]]
3288                 $p4 = "/\[\[(:*$namespacechar+):({$np}+) \\(({$np}+)\\)\\|]]/"; # [[ns:page (cont)|]] and [[:ns:page (cont)|]]
3289                 $context = '';
3290                 $t = $this->mTitle->getText();
3291                 if ( preg_match( $conpat, $t, $m ) ) {
3292                         $context = $m[2];
3293                 }
3294                 $text = preg_replace( $p4, '[[\\1:\\2 (\\3)|\\2]]', $text );
3295                 $text = preg_replace( $p1, '[[\\1 (\\2)|\\1]]', $text );
3296                 $text = preg_replace( $p3, '[[\\1:\\2|\\2]]', $text );
3297
3298                 if ( '' == $context ) {
3299                         $text = preg_replace( $p2, '[[\\1]]', $text );
3300                 } else {
3301                         $text = preg_replace( $p2, "[[\\1 ({$context})|\\1]]", $text );
3302                 }
3303
3304                 # Trim trailing whitespace
3305                 # MAG_END (__END__) tag allows for trailing
3306                 # whitespace to be deliberately included
3307                 $text = rtrim( $text );
3308                 $mw =& MagicWord::get( MAG_END );
3309                 $mw->matchAndRemove( $text );
3310
3311                 return $text;
3312         }
3313
3314         /**
3315          * Fetch the user's signature text, if any, and normalize to
3316          * validated, ready-to-insert wikitext.
3317          *
3318          * @param User $user
3319          * @return string
3320          * @access private
3321          */
3322         function getUserSig( &$user ) {
3323                 global $wgContLang;
3324
3325                 $username = $user->getName();
3326                 $nickname = $user->getOption( 'nickname' );
3327                 $nickname = $nickname === '' ? $username : $nickname;
3328
3329                 if( $user->getBoolOption( 'fancysig' ) !== false ) {
3330                         # Sig. might contain markup; validate this
3331                         if( $this->validateSig( $nickname ) !== false ) {
3332                                 # Validated; clean up (if needed) and return it
3333                                 return( $this->cleanSig( $nickname ) );
3334                         } else {
3335                                 # Failed to validate; fall back to the default
3336                                 $nickname = $username;
3337                                 wfDebug( "Parser::getUserSig: $username has bad XML tags in signature.\n" );
3338                         }
3339                 }
3340
3341                 # If we're still here, make it a link to the user page
3342                 $userpage = $user->getUserPage();
3343                 return( '[[' . $userpage->getPrefixedText() . '|' . wfEscapeWikiText( $nickname ) . ']]' );
3344         }
3345
3346         /**
3347          * Check that the user's signature contains no bad XML
3348          *
3349          * @param string $text
3350          * @return mixed An expanded string, or false if invalid.
3351          */
3352         function validateSig( $text ) {
3353                 return( wfIsWellFormedXmlFragment( $text ) ? $text : false );
3354         }
3355
3356         /**
3357          * Clean up signature text
3358          *
3359          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures
3360          * 2) Substitute all transclusions
3361          *
3362          * @param string $text
3363          * @return string Signature text
3364          */
3365         function cleanSig( $text ) {
3366                 $substWord = MagicWord::get( MAG_SUBST );
3367                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3368                 $substText = '{{' . $substWord->getSynonym( 0 );
3369
3370                 $text = preg_replace( $substRegex, $substText, $text );
3371                 $text = preg_replace( '/~{3,5}/', '', $text );
3372                 $text = $this->replaceVariables( $text );
3373
3374                 return $text;
3375         }
3376
3377         /**
3378          * Set up some variables which are usually set up in parse()
3379          * so that an external function can call some class members with confidence
3380          * @access public
3381          */
3382         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
3383                 $this->mTitle =& $title;
3384                 $this->mOptions = $options;
3385                 $this->mOutputType = $outputType;
3386                 if ( $clearState ) {
3387                         $this->clearState();
3388                 }
3389         }
3390
3391         /**
3392          * Transform a MediaWiki message by replacing magic variables.
3393          *
3394          * @param string $text the text to transform
3395          * @param ParserOptions $options  options
3396          * @return string the text with variables substituted
3397          * @access public
3398          */
3399         function transformMsg( $text, $options ) {
3400                 global $wgTitle;
3401                 static $executing = false;
3402
3403                 $fname = "Parser::transformMsg";
3404
3405                 # Guard against infinite recursion
3406                 if ( $executing ) {
3407                         return $text;
3408                 }
3409                 $executing = true;
3410
3411                 wfProfileIn($fname);
3412
3413                 $this->mTitle = $wgTitle;
3414                 $this->mOptions = $options;
3415                 $this->mOutputType = OT_MSG;
3416                 $this->clearState();
3417                 $text = $this->replaceVariables( $text );
3418
3419                 $executing = false;
3420                 wfProfileOut($fname);
3421                 return $text;
3422         }
3423
3424         /**
3425          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
3426          * Callback will be called with the text within
3427          * Transform and return the text within
3428          *
3429          * @access public
3430          *
3431          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
3432          * @param mixed $callback The callback function (and object) to use for the tag
3433          *
3434          * @return The old value of the mTagHooks array associated with the hook
3435          */
3436         function setHook( $tag, $callback ) {
3437                 $oldVal = @$this->mTagHooks[$tag];
3438                 $this->mTagHooks[$tag] = $callback;
3439
3440                 return $oldVal;
3441         }
3442
3443         /**
3444          * Replace <!--LINK--> link placeholders with actual links, in the buffer
3445          * Placeholders created in Skin::makeLinkObj()
3446          * Returns an array of links found, indexed by PDBK:
3447          *  0 - broken
3448          *  1 - normal link
3449          *  2 - stub
3450          * $options is a bit field, RLH_FOR_UPDATE to select for update
3451          */
3452         function replaceLinkHolders( &$text, $options = 0 ) {
3453                 global $wgUser;
3454                 global $wgOutputReplace;
3455
3456                 $fname = 'Parser::replaceLinkHolders';
3457                 wfProfileIn( $fname );
3458
3459                 $pdbks = array();
3460                 $colours = array();
3461                 $sk =& $this->mOptions->getSkin();
3462                 $linkCache =& LinkCache::singleton();
3463
3464                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
3465                         wfProfileIn( $fname.'-check' );
3466                         $dbr =& wfGetDB( DB_SLAVE );
3467                         $page = $dbr->tableName( 'page' );
3468                         $threshold = $wgUser->getOption('stubthreshold');
3469
3470                         # Sort by namespace
3471                         asort( $this->mLinkHolders['namespaces'] );
3472
3473                         # Generate query
3474                         $query = false;
3475                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3476                                 # Make title object
3477                                 $title = $this->mLinkHolders['titles'][$key];
3478
3479                                 # Skip invalid entries.
3480                                 # Result will be ugly, but prevents crash.
3481                                 if ( is_null( $title ) ) {
3482                                         continue;
3483                                 }
3484                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
3485
3486                                 # Check if it's a static known link, e.g. interwiki
3487                                 if ( $title->isAlwaysKnown() ) {
3488                                         $colours[$pdbk] = 1;
3489                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
3490                                         $colours[$pdbk] = 1;
3491                                         $this->mOutput->addLink( $title, $id );
3492                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
3493                                         $colours[$pdbk] = 0;
3494                                 } else {
3495                                         # Not in the link cache, add it to the query
3496                                         if ( !isset( $current ) ) {
3497                                                 $current = $ns;
3498                                                 $query =  "SELECT page_id, page_namespace, page_title";
3499                                                 if ( $threshold > 0 ) {
3500                                                         $query .= ', page_len, page_is_redirect';
3501                                                 }
3502                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
3503                                         } elseif ( $current != $ns ) {
3504                                                 $current = $ns;
3505                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
3506                                         } else {
3507                                                 $query .= ', ';
3508                                         }
3509
3510                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
3511                                 }
3512                         }
3513                         if ( $query ) {
3514                                 $query .= '))';
3515                                 if ( $options & RLH_FOR_UPDATE ) {
3516                                         $query .= ' FOR UPDATE';
3517                                 }
3518
3519                                 $res = $dbr->query( $query, $fname );
3520
3521                                 # Fetch data and form into an associative array
3522                                 # non-existent = broken
3523                                 # 1 = known
3524                                 # 2 = stub
3525                                 while ( $s = $dbr->fetchObject($res) ) {
3526                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
3527                                         $pdbk = $title->getPrefixedDBkey();
3528                                         $linkCache->addGoodLinkObj( $s->page_id, $title );
3529                                         $this->mOutput->addLink( $title, $s->page_id );
3530
3531                                         if ( $threshold >  0 ) {
3532                                                 $size = $s->page_len;
3533                                                 if ( $s->page_is_redirect || $s->page_namespace != 0 || $size >= $threshold ) {
3534                                                         $colours[$pdbk] = 1;
3535                                                 } else {
3536                                                         $colours[$pdbk] = 2;
3537                                                 }
3538                                         } else {
3539                                                 $colours[$pdbk] = 1;
3540                                         }
3541                                 }
3542                         }
3543                         wfProfileOut( $fname.'-check' );
3544
3545                         # Construct search and replace arrays
3546                         wfProfileIn( $fname.'-construct' );
3547                         $wgOutputReplace = array();
3548                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3549                                 $pdbk = $pdbks[$key];
3550                                 $searchkey = "<!--LINK $key-->";
3551                                 $title = $this->mLinkHolders['titles'][$key];
3552                                 if ( empty( $colours[$pdbk] ) ) {
3553                                         $linkCache->addBadLinkObj( $title );
3554                                         $colours[$pdbk] = 0;
3555                                         $this->mOutput->addLink( $title, 0 );
3556                                         $wgOutputReplace[$searchkey] = $sk->makeBrokenLinkObj( $title,
3557                                                                         $this->mLinkHolders['texts'][$key],
3558                                                                         $this->mLinkHolders['queries'][$key] );
3559                                 } elseif ( $colours[$pdbk] == 1 ) {
3560                                         $wgOutputReplace[$searchkey] = $sk->makeKnownLinkObj( $title,
3561                                                                         $this->mLinkHolders['texts'][$key],
3562                                                                         $this->mLinkHolders['queries'][$key] );
3563                                 } elseif ( $colours[$pdbk] == 2 ) {
3564                                         $wgOutputReplace[$searchkey] = $sk->makeStubLinkObj( $title,
3565                                                                         $this->mLinkHolders['texts'][$key],
3566                                                                         $this->mLinkHolders['queries'][$key] );
3567                                 }
3568                         }
3569                         wfProfileOut( $fname.'-construct' );
3570
3571                         # Do the thing
3572                         wfProfileIn( $fname.'-replace' );
3573
3574                         $text = preg_replace_callback(
3575                                 '/(<!--LINK .*?-->)/',
3576                                 "wfOutputReplaceMatches",
3577                                 $text);
3578
3579                         wfProfileOut( $fname.'-replace' );
3580                 }
3581
3582                 # Now process interwiki link holders
3583                 # This is quite a bit simpler than internal links
3584                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
3585                         wfProfileIn( $fname.'-interwiki' );
3586                         # Make interwiki link HTML
3587                         $wgOutputReplace = array();
3588                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
3589                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
3590                                 $wgOutputReplace[$key] = $sk->makeLinkObj( $title, $link );
3591                         }
3592
3593                         $text = preg_replace_callback(
3594                                 '/<!--IWLINK (.*?)-->/',
3595                                 "wfOutputReplaceMatches",
3596                                 $text );
3597                         wfProfileOut( $fname.'-interwiki' );
3598                 }
3599
3600                 wfProfileOut( $fname );
3601                 return $colours;
3602         }
3603
3604         /**
3605          * Replace <!--LINK--> link placeholders with plain text of links
3606          * (not HTML-formatted).
3607          * @param string $text
3608          * @return string
3609          */
3610         function replaceLinkHoldersText( $text ) {
3611                 global $wgUser;
3612                 global $wgOutputReplace;
3613
3614                 $fname = 'Parser::replaceLinkHoldersText';
3615                 wfProfileIn( $fname );
3616
3617                 $text = preg_replace_callback(
3618                         '/<!--(LINK|IWLINK) (.*?)-->/',
3619                         array( &$this, 'replaceLinkHoldersTextCallback' ),
3620                         $text );
3621
3622                 wfProfileOut( $fname );
3623                 return $text;
3624         }
3625
3626         /**
3627          * @param array $matches
3628          * @return string
3629          * @access private
3630          */
3631         function replaceLinkHoldersTextCallback( $matches ) {
3632                 $type = $matches[1];
3633                 $key  = $matches[2];
3634                 if( $type == 'LINK' ) {
3635                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
3636                                 return $this->mLinkHolders['texts'][$key];
3637                         }
3638                 } elseif( $type == 'IWLINK' ) {
3639                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
3640                                 return $this->mInterwikiLinkHolders['texts'][$key];
3641                         }
3642                 }
3643                 return $matches[0];
3644         }
3645
3646         /**
3647          * Renders an image gallery from a text with one line per image.
3648          * text labels may be given by using |-style alternative text. E.g.
3649          *   Image:one.jpg|The number "1"
3650          *   Image:tree.jpg|A tree
3651          * given as text will return the HTML of a gallery with two images,
3652          * labeled 'The number "1"' and
3653          * 'A tree'.
3654          */
3655         function renderImageGallery( $text ) {
3656                 # Setup the parser
3657                 $parserOptions = new ParserOptions;
3658                 $localParser = new Parser();
3659
3660                 $ig = new ImageGallery();
3661                 $ig->setShowBytes( false );
3662                 $ig->setShowFilename( false );
3663                 $lines = explode( "\n", $text );
3664
3665                 foreach ( $lines as $line ) {
3666                         # match lines like these:
3667                         # Image:someimage.jpg|This is some image
3668                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
3669                         # Skip empty lines
3670                         if ( count( $matches ) == 0 ) {
3671                                 continue;
3672                         }
3673                         $nt =& Title::newFromText( $matches[1] );
3674                         if( is_null( $nt ) ) {
3675                                 # Bogus title. Ignore these so we don't bomb out later.
3676                                 continue;
3677                         }
3678                         if ( isset( $matches[3] ) ) {
3679                                 $label = $matches[3];
3680                         } else {
3681                                 $label = '';
3682                         }
3683
3684                         $pout = $localParser->parse( $label , $this->mTitle, $parserOptions );
3685                         $html = $pout->getText();
3686
3687                         $ig->add( new Image( $nt ), $html );
3688                         $this->mOutput->addImage( $nt->getDBkey() );
3689                 }
3690                 return $ig->toHTML();
3691         }
3692
3693         /**
3694          * Parse image options text and use it to make an image
3695          */
3696         function makeImage( &$nt, $options ) {
3697                 global $wgContLang, $wgUseImageResize, $wgUser;
3698
3699                 $align = '';
3700
3701                 # Check if the options text is of the form "options|alt text"
3702                 # Options are:
3703                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
3704                 #  * left               no resizing, just left align. label is used for alt= only
3705                 #  * right              same, but right aligned
3706                 #  * none               same, but not aligned
3707                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
3708                 #  * center             center the image
3709                 #  * framed             Keep original image size, no magnify-button.
3710
3711                 $part = explode( '|', $options);
3712
3713                 $mwThumb  =& MagicWord::get( MAG_IMG_THUMBNAIL );
3714                 $mwManualThumb =& MagicWord::get( MAG_IMG_MANUALTHUMB );
3715                 $mwLeft   =& MagicWord::get( MAG_IMG_LEFT );
3716                 $mwRight  =& MagicWord::get( MAG_IMG_RIGHT );
3717                 $mwNone   =& MagicWord::get( MAG_IMG_NONE );
3718                 $mwWidth  =& MagicWord::get( MAG_IMG_WIDTH );
3719                 $mwCenter =& MagicWord::get( MAG_IMG_CENTER );
3720                 $mwFramed =& MagicWord::get( MAG_IMG_FRAMED );
3721                 $caption = '';
3722
3723                 $width = $height = $framed = $thumb = false;
3724                 $manual_thumb = '' ;
3725
3726                 foreach( $part as $key => $val ) {
3727                         if ( $wgUseImageResize && ! is_null( $mwThumb->matchVariableStartToEnd($val) ) ) {
3728                                 $thumb=true;
3729                         } elseif ( ! is_null( $match = $mwManualThumb->matchVariableStartToEnd($val) ) ) {
3730                                 # use manually specified thumbnail
3731                                 $thumb=true;
3732                                 $manual_thumb = $match;
3733                         } elseif ( ! is_null( $mwRight->matchVariableStartToEnd($val) ) ) {
3734                                 # remember to set an alignment, don't render immediately
3735                                 $align = 'right';
3736                         } elseif ( ! is_null( $mwLeft->matchVariableStartToEnd($val) ) ) {
3737                                 # remember to set an alignment, don't render immediately
3738                                 $align = 'left';
3739                         } elseif ( ! is_null( $mwCenter->matchVariableStartToEnd($val) ) ) {
3740                                 # remember to set an alignment, don't render immediately
3741                                 $align = 'center';
3742                         } elseif ( ! is_null( $mwNone->matchVariableStartToEnd($val) ) ) {
3743                                 # remember to set an alignment, don't render immediately
3744                                 $align = 'none';
3745                         } elseif ( $wgUseImageResize && ! is_null( $match = $mwWidth->matchVariableStartToEnd($val) ) ) {
3746                                 wfDebug( "MAG_IMG_WIDTH match: $match\n" );
3747                                 # $match is the image width in pixels
3748                                 if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $match, $m ) ) {
3749                                         $width = intval( $m[1] );
3750                                         $height = intval( $m[2] );
3751                                 } else {
3752                                         $width = intval($match);
3753                                 }
3754                         } elseif ( ! is_null( $mwFramed->matchVariableStartToEnd($val) ) ) {
3755                                 $framed=true;
3756                         } else {
3757                                 $caption = $val;
3758                         }
3759                 }
3760                 # Strip bad stuff out of the alt text
3761                 $alt = $this->replaceLinkHoldersText( $caption );
3762                 $alt = Sanitizer::stripAllTags( $alt );
3763
3764                 # Linker does the rest
3765                 $sk =& $this->mOptions->getSkin();
3766                 return $sk->makeImageLinkObj( $nt, $caption, $alt, $align, $width, $height, $framed, $thumb, $manual_thumb );
3767         }
3768
3769         /**
3770          * Set a flag in the output object indicating that the content is dynamic and
3771          * shouldn't be cached.
3772          */
3773         function disableCache() {
3774                 $this->mOutput->mCacheTime = -1;
3775         }
3776
3777         /**#@+
3778          * Callback from the Sanitizer for expanding items found in HTML attribute
3779          * values, so they can be safely tested and escaped.
3780          * @param string $text
3781          * @param array $args
3782          * @return string
3783          * @access private
3784          */
3785         function attributeStripCallback( &$text, $args ) {
3786                 $text = $this->replaceVariables( $text, $args );
3787                 $text = $this->unstripForHTML( $text );
3788                 return $text;
3789         }
3790
3791         function unstripForHTML( $text ) {
3792                 $text = $this->unstrip( $text, $this->mStripState );
3793                 $text = $this->unstripNoWiki( $text, $this->mStripState );
3794                 return $text;
3795         }
3796         /**#@-*/
3797
3798         /**#@+
3799          * Accessor/mutator
3800          */
3801         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
3802         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
3803         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
3804         /**#@-*/
3805
3806         /**#@+
3807          * Accessor
3808          */
3809         function getTags() { return array_keys( $this->mTagHooks ); }
3810         /**#@-*/
3811 }
3812
3813 /**
3814  * @todo document
3815  * @package MediaWiki
3816  */
3817 class ParserOutput
3818 {
3819         var $mText,             # The output text
3820                 $mLanguageLinks,    # List of the full text of language links, in the order they appear
3821                 $mCategories,       # Map of category names to sort keys
3822                 $mContainsOldMagic, # Boolean variable indicating if the input contained variables like {{CURRENTDAY}}
3823                 $mCacheTime,        # Timestamp on this article, or -1 for uncacheable. Used in ParserCache.
3824                 $mVersion,          # Compatibility check
3825                 $mTitleText,        # title text of the chosen language variant
3826                 $mLinks,            # 2-D map of NS/DBK to ID for the links in the document. ID=zero for broken.
3827                 $mTemplates,        # 2-D map of NS/DBK to ID for the template references. ID=zero for broken.
3828                 $mImages,           # DB keys of the images used, in the array key only
3829                 $mExternalLinks;    # External link URLs, in the key only
3830
3831         function ParserOutput( $text = '', $languageLinks = array(), $categoryLinks = array(),
3832                 $containsOldMagic = false, $titletext = '' )
3833         {
3834                 $this->mText = $text;
3835                 $this->mLanguageLinks = $languageLinks;
3836                 $this->mCategories = $categoryLinks;
3837                 $this->mContainsOldMagic = $containsOldMagic;
3838                 $this->mCacheTime = '';
3839                 $this->mVersion = MW_PARSER_VERSION;
3840                 $this->mTitleText = $titletext;
3841                 $this->mLinks = array();
3842                 $this->mTemplates = array();
3843                 $this->mImages = array();
3844                 $this->mExternalLinks = array();
3845         }
3846
3847         function getText()                   { return $this->mText; }
3848         function getLanguageLinks()          { return $this->mLanguageLinks; }
3849         function getCategoryLinks()          { return array_keys( $this->mCategories ); }
3850         function &getCategories()            { return $this->mCategories; }
3851         function getCacheTime()              { return $this->mCacheTime; }
3852         function getTitleText()              { return $this->mTitleText; }
3853         function &getLinks()                 { return $this->mLinks; }
3854         function &getTemplates()             { return $this->mTemplates; }
3855         function &getImages()                { return $this->mImages; }
3856         function &getExternalLinks()         { return $this->mExternalLinks; }
3857
3858         function containsOldMagic()          { return $this->mContainsOldMagic; }
3859         function setText( $text )            { return wfSetVar( $this->mText, $text ); }
3860         function setLanguageLinks( $ll )     { return wfSetVar( $this->mLanguageLinks, $ll ); }
3861         function setCategoryLinks( $cl )     { return wfSetVar( $this->mCategories, $cl ); }
3862         function setContainsOldMagic( $com ) { return wfSetVar( $this->mContainsOldMagic, $com ); }
3863         function setCacheTime( $t )          { return wfSetVar( $this->mCacheTime, $t ); }
3864         function setTitleText( $t )          { return wfSetVar ($this->mTitleText, $t); }
3865
3866         function addCategory( $c, $sort )    { $this->mCategories[$c] = $sort; }
3867         function addImage( $name )           { $this->mImages[$name] = 1; }
3868         function addLanguageLink( $t )       { $this->mLanguageLinks[] = $t; }
3869         function addExternalLink( $url )     { $this->mExternalLinks[$url] = 1; }
3870
3871         function addLink( $title, $id ) {
3872                 $ns = $title->getNamespace();
3873                 $dbk = $title->getDBkey();
3874                 if ( !isset( $this->mLinks[$ns] ) ) {
3875                         $this->mLinks[$ns] = array();
3876                 }
3877                 $this->mLinks[$ns][$dbk] = $id;
3878         }
3879
3880         function addTemplate( $title, $id ) {
3881                 $ns = $title->getNamespace();
3882                 $dbk = $title->getDBkey();
3883                 if ( !isset( $this->mTemplates[$ns] ) ) {
3884                         $this->mTemplates[$ns] = array();
3885                 }
3886                 $this->mTemplates[$ns][$dbk] = $id;
3887         }
3888
3889         /**
3890          * @deprecated
3891          */
3892         /*
3893         function merge( $other ) {
3894                 $this->mLanguageLinks = array_merge( $this->mLanguageLinks, $other->mLanguageLinks );
3895                 $this->mCategories = array_merge( $this->mCategories, $this->mLanguageLinks );
3896                 $this->mContainsOldMagic = $this->mContainsOldMagic || $other->mContainsOldMagic;
3897         }*/
3898
3899         /**
3900          * Return true if this cached output object predates the global or
3901          * per-article cache invalidation timestamps, or if it comes from
3902          * an incompatible older version.
3903          *
3904          * @param string $touched the affected article's last touched timestamp
3905          * @return bool
3906          * @access public
3907          */
3908         function expired( $touched ) {
3909                 global $wgCacheEpoch;
3910                 return $this->getCacheTime() == -1 || // parser says it's uncacheable
3911                        $this->getCacheTime() < $touched ||
3912                        $this->getCacheTime() <= $wgCacheEpoch ||
3913                        !isset( $this->mVersion ) ||
3914                        version_compare( $this->mVersion, MW_PARSER_VERSION, "lt" );
3915         }
3916 }
3917
3918 /**
3919  * Set options of the Parser
3920  * @todo document
3921  * @package MediaWiki
3922  */
3923 class ParserOptions
3924 {
3925         # All variables are private
3926         var $mUseTeX;                    # Use texvc to expand <math> tags
3927         var $mUseDynamicDates;           # Use DateFormatter to format dates
3928         var $mInterwikiMagic;            # Interlanguage links are removed and returned in an array
3929         var $mAllowExternalImages;       # Allow external images inline
3930         var $mAllowExternalImagesFrom;   # If not, any exception?
3931         var $mSkin;                      # Reference to the preferred skin
3932         var $mDateFormat;                # Date format index
3933         var $mEditSection;               # Create "edit section" links
3934         var $mNumberHeadings;            # Automatically number headings
3935         var $mAllowSpecialInclusion;     # Allow inclusion of special pages
3936         var $mTidy;                      # Ask for tidy cleanup
3937
3938         function getUseTeX()                        { return $this->mUseTeX; }
3939         function getUseDynamicDates()               { return $this->mUseDynamicDates; }
3940         function getInterwikiMagic()                { return $this->mInterwikiMagic; }
3941         function getAllowExternalImages()           { return $this->mAllowExternalImages; }
3942         function getAllowExternalImagesFrom()       { return $this->mAllowExternalImagesFrom; }
3943         function &getSkin()                         { return $this->mSkin; }
3944         function getDateFormat()                    { return $this->mDateFormat; }
3945         function getEditSection()                   { return $this->mEditSection; }
3946         function getNumberHeadings()                { return $this->mNumberHeadings; }
3947         function getAllowSpecialInclusion()         { return $this->mAllowSpecialInclusion; }
3948         function getTidy()                          { return $this->mTidy; }
3949
3950         function setUseTeX( $x )                    { return wfSetVar( $this->mUseTeX, $x ); }
3951         function setUseDynamicDates( $x )           { return wfSetVar( $this->mUseDynamicDates, $x ); }
3952         function setInterwikiMagic( $x )            { return wfSetVar( $this->mInterwikiMagic, $x ); }
3953         function setAllowExternalImages( $x )       { return wfSetVar( $this->mAllowExternalImages, $x ); }
3954         function setAllowExternalImagesFrom( $x )   { return wfSetVar( $this->mAllowExternalImagesFrom, $x ); }
3955         function setDateFormat( $x )                { return wfSetVar( $this->mDateFormat, $x ); }
3956         function setEditSection( $x )               { return wfSetVar( $this->mEditSection, $x ); }
3957         function setNumberHeadings( $x )            { return wfSetVar( $this->mNumberHeadings, $x ); }
3958         function setAllowSpecialInclusion( $x )     { return wfSetVar( $this->mAllowSpecialInclusion, $x ); }
3959         function setTidy( $x )                      { return wfSetVar( $this->mTidy, $x); }
3960         function setSkin( &$x ) { $this->mSkin =& $x; }
3961
3962         function ParserOptions() {
3963                 global $wgUser;
3964                 $this->initialiseFromUser( $wgUser );
3965         }
3966
3967         /**
3968          * Get parser options
3969          * @static
3970          */
3971         function newFromUser( &$user ) {
3972                 $popts = new ParserOptions;
3973                 $popts->initialiseFromUser( $user );
3974                 return $popts;
3975         }
3976
3977         /** Get user options */
3978         function initialiseFromUser( &$userInput ) {
3979                 global $wgUseTeX, $wgUseDynamicDates, $wgInterwikiMagic, $wgAllowExternalImages,
3980                        $wgAllowExternalImagesFrom, $wgAllowSpecialInclusion;
3981                 $fname = 'ParserOptions::initialiseFromUser';
3982                 wfProfileIn( $fname );
3983                 if ( !$userInput ) {
3984                         $user = new User;
3985                         $user->setLoaded( true );
3986                 } else {
3987                         $user =& $userInput;
3988                 }
3989
3990                 $this->mUseTeX = $wgUseTeX;
3991                 $this->mUseDynamicDates = $wgUseDynamicDates;
3992                 $this->mInterwikiMagic = $wgInterwikiMagic;
3993                 $this->mAllowExternalImages = $wgAllowExternalImages;
3994                 $this->mAllowExternalImagesFrom = $wgAllowExternalImagesFrom;
3995                 wfProfileIn( $fname.'-skin' );
3996                 $this->mSkin =& $user->getSkin();
3997                 wfProfileOut( $fname.'-skin' );
3998                 $this->mDateFormat = $user->getOption( 'date' );
3999                 $this->mEditSection = true;
4000                 $this->mNumberHeadings = $user->getOption( 'numberheadings' );
4001                 $this->mAllowSpecialInclusion = $wgAllowSpecialInclusion;
4002                 $this->mTidy = false;
4003                 wfProfileOut( $fname );
4004         }
4005 }
4006
4007 /**
4008  * Callback function used by Parser::replaceLinkHolders()
4009  * to substitute link placeholders.
4010  */
4011 function &wfOutputReplaceMatches( $matches ) {
4012         global $wgOutputReplace;
4013         return $wgOutputReplace[$matches[1]];
4014 }
4015
4016 /**
4017  * Return the total number of articles
4018  */
4019 function wfNumberOfArticles() {
4020         global $wgNumberOfArticles;
4021
4022         wfLoadSiteStats();
4023         return $wgNumberOfArticles;
4024 }
4025
4026 /**
4027  * Return the number of files
4028  */
4029 function wfNumberOfFiles() {
4030         $fname = 'Parser::wfNumberOfFiles';
4031
4032         wfProfileIn( $fname );
4033         $dbr =& wfGetDB( DB_SLAVE );
4034         $res = $dbr->selectField('image', 'COUNT(*)', array(), $fname );
4035         wfProfileOut( $fname );
4036
4037         return $res;
4038 }
4039
4040 /**
4041  * Get various statistics from the database
4042  * @access private
4043  */
4044 function wfLoadSiteStats() {
4045         global $wgNumberOfArticles, $wgTotalViews, $wgTotalEdits;
4046         $fname = 'wfLoadSiteStats';
4047
4048         if ( -1 != $wgNumberOfArticles ) return;
4049         $dbr =& wfGetDB( DB_SLAVE );
4050         $s = $dbr->selectRow( 'site_stats',
4051                 array( 'ss_total_views', 'ss_total_edits', 'ss_good_articles' ),
4052                 array( 'ss_row_id' => 1 ), $fname
4053         );
4054
4055         if ( $s === false ) {
4056                 return;
4057         } else {
4058                 $wgTotalViews = $s->ss_total_views;
4059                 $wgTotalEdits = $s->ss_total_edits;
4060                 $wgNumberOfArticles = $s->ss_good_articles;
4061         }
4062 }
4063
4064 /**
4065  * Escape html tags
4066  * Basically replacing " > and < with HTML entities ( &quot;, &gt;, &lt;)
4067  *
4068  * @param string $in Text that might contain HTML tags
4069  * @return string Escaped string
4070  */
4071 function wfEscapeHTMLTagsOnly( $in ) {
4072         return str_replace(
4073                 array( '"', '>', '<' ),
4074                 array( '&quot;', '&gt;', '&lt;' ),
4075                 $in );
4076 }
4077
4078 ?>