includes/Parser.php

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