includes/Parser.php

   1 <?php
   2 /**
   3  * File for Parser and related classes
   4  *
   5  * @package MediaWiki
   6  * @subpackage Parser
   7  */
   8
   9 /** */
  10 require_once( 'Sanitizer.php' );
  11 require_once( 'HttpFunctions.php' );
  12
  13 /**
  14  * Update this version number when the ParserOutput format
  15  * changes in an incompatible way, so the parser cache
  16  * can automatically discard old data.
  17  */
  18 define( 'MW_PARSER_VERSION', '1.6.0' );
  19
  20 /**
  21  * Variable substitution O(N^2) attack
  22  *
  23  * Without countermeasures, it would be possible to attack the parser by saving
  24  * a page filled with a large number of inclusions of large pages. The size of
  25  * the generated page would be proportional to the square of the input size.
  26  * Hence, we limit the number of inclusions of any given page, thus bringing any
  27  * attack back to O(N).
  28  */
  29
  30 define( 'MAX_INCLUDE_REPEAT', 100 );
  31 define( 'MAX_INCLUDE_SIZE', 1000000 ); // 1 Million
  32
  33 define( 'RLH_FOR_UPDATE', 1 );
  34
  35 # Allowed values for $mOutputType
  36 define( 'OT_HTML', 1 );
  37 define( 'OT_WIKI', 2 );
  38 define( 'OT_MSG' , 3 );
  39
  40 # string parameter for extractTags which will cause it
  41 # to strip HTML comments in addition to regular
  42 # <XML>-style tags. This should not be anything we
  43 # may want to use in wikisyntax
  44 define( 'STRIP_COMMENTS', 'HTMLCommentStrip' );
  45
  46 # Constants needed for external link processing
  47 define( 'HTTP_PROTOCOLS', 'http:\/\/|https:\/\/' );
  48 # Everything except bracket, space, or control characters
  49 define( 'EXT_LINK_URL_CLASS', '[^][<>"\\x00-\\x20\\x7F]' );
  50 # Including space
  51 define( 'EXT_LINK_TEXT_CLASS', '[^\]\\x00-\\x1F\\x7F]' );
  52 define( 'EXT_IMAGE_FNAME_CLASS', '[A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]' );
  53 define( 'EXT_IMAGE_EXTENSIONS', 'gif|png|jpg|jpeg' );
  54 define( 'EXT_LINK_BRACKETED',  '/\[(\b(' . wfUrlProtocols() . ')'.EXT_LINK_URL_CLASS.'+) *('.EXT_LINK_TEXT_CLASS.'*?)\]/S' );
  55 define( 'EXT_IMAGE_REGEX',
  56         '/^('.HTTP_PROTOCOLS.')'.  # Protocol
  57         '('.EXT_LINK_URL_CLASS.'+)\\/'.  # Hostname and path
  58         '('.EXT_IMAGE_FNAME_CLASS.'+)\\.((?i)'.EXT_IMAGE_EXTENSIONS.')$/S' # Filename
  59 );
  60
  61 /**
  62  * PHP Parser
  63  *
  64  * Processes wiki markup
  65  *
  66  * <pre>
  67  * There are three main entry points into the Parser class:
  68  * parse()
  69  *   produces HTML output
  70  * preSaveTransform().
  71  *   produces altered wiki markup.
  72  * transformMsg()
  73  *   performs brace substitution on MediaWiki messages
  74  *
  75  * Globals used:
  76  *    objects:   $wgLang, $wgContLang
  77  *
  78  * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
  79  *
  80  * settings:
  81  *  $wgUseTex*, $wgUseDynamicDates*, $wgInterwikiMagic*,
  82  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  83  *  $wgLocaltimezone, $wgAllowSpecialInclusion*
  84  *
  85  *  * only within ParserOptions
  86  * </pre>
  87  *
  88  * @package MediaWiki
  89  */
  90 class Parser
  91 {
  92         /**#@+
  93          * @access private
  94          */
  95         # Persistent:
  96         var $mTagHooks;
  97
  98         # Cleared with clearState():
  99         var $mOutput, $mAutonumber, $mDTopen, $mStripState = array();
 100         var $mVariables, $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 101         var $mInterwikiLinkHolders, $mLinkHolders, $mUniqPrefix;
 102         var $mTemplates,        // cache of already loaded templates, avoids
 103                                 // multiple SQL queries for the same string
 104             $mTemplatePath;     // stores an unsorted hash of all the templates already loaded
 105                                 // in this path. Used for loop detection.
 106
 107         # Temporary
 108         # These are variables reset at least once per parse regardless of $clearState
 109         var $mOptions,      // ParserOptions object
 110                 $mTitle,        // Title context, used for self-link rendering and similar things
 111                 $mOutputType,   // Output type, one of the OT_xxx constants
 112                 $mRevisionId;   // ID to display in {{REVISIONID}} tags
 113
 114         /**#@-*/
 115
 116         /**
 117          * Constructor
 118          *
 119          * @access public
 120          */
 121         function Parser() {
 122                 $this->mTagHooks = array();
 123                 $this->clearState();
 124         }
 125
 126         /**
 127          * Clear Parser state
 128          *
 129          * @access private
 130          */
 131         function clearState() {
 132                 $this->mOutput = new ParserOutput;
 133                 $this->mAutonumber = 0;
 134                 $this->mLastSection = '';
 135                 $this->mDTopen = false;
 136                 $this->mVariables = false;
 137                 $this->mIncludeCount = array();
 138                 $this->mStripState = array();
 139                 $this->mArgStack = array();
 140                 $this->mInPre = false;
 141                 $this->mInterwikiLinkHolders = array(
 142                         'texts' => array(),
 143                         'titles' => array()
 144                 );
 145                 $this->mLinkHolders = array(
 146                         'namespaces' => array(),
 147                         'dbkeys' => array(),
 148                         'queries' => array(),
 149                         'texts' => array(),
 150                         'titles' => array()
 151                 );
 152                 $this->mRevisionId = null;
 153                 $this->mUniqPrefix = 'UNIQ' . Parser::getRandomString();
 154
 155                 # Clear these on every parse, bug 4549
 156                 $this->mTemplates = array();
 157                 $this->mTemplatePath = array();
 158
 159                 wfRunHooks( 'ParserClearState', array( &$this ) );
 160         }
 161
 162         /**
 163          * Accessor for mUniqPrefix.
 164          *
 165          * @access public
 166          */
 167         function UniqPrefix() {
 168                 return $this->mUniqPrefix;
 169         }
 170
 171         /**
 172          * Convert wikitext to HTML
 173          * Do not call this function recursively.
 174          *
 175          * @access private
 176          * @param string $text Text we want to parse
 177          * @param Title &$title A title object
 178          * @param array $options
 179          * @param boolean $linestart
 180          * @param boolean $clearState
 181          * @param int $revid number to pass in {{REVISIONID}}
 182          * @return ParserOutput a ParserOutput
 183          */
 184         function parse( $text, &$title, $options, $linestart = true, $clearState = true, $revid = null ) {
 185                 /**
 186                  * First pass--just handle <nowiki> sections, pass the rest off
 187                  * to internalParse() which does all the real work.
 188                  */
 189
 190                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
 191                 $fname = 'Parser::parse';
 192                 wfProfileIn( $fname );
 193
 194                 if ( $clearState ) {
 195                         $this->clearState();
 196                 }
 197
 198                 $this->mOptions = $options;
 199                 $this->mTitle =& $title;
 200                 $this->mRevisionId = $revid;
 201                 $this->mOutputType = OT_HTML;
 202
 203                 $this->mStripState = NULL;
 204
 205                 //$text = $this->strip( $text, $this->mStripState );
 206                 // VOODOO MAGIC FIX! Sometimes the above segfaults in PHP5.
 207                 $x =& $this->mStripState;
 208
 209                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$x ) );
 210                 $text = $this->strip( $text, $x );
 211                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$x ) );
 212
 213                 # Hook to suspend the parser in this state
 214                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$x ) ) ) {
 215                         wfProfileOut( $fname );
 216                         return $text ;
 217                 }
 218
 219                 $text = $this->internalParse( $text );
 220
 221                 $text = $this->unstrip( $text, $this->mStripState );
 222
 223                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 224                 $fixtags = array(
 225                         # french spaces, last one Guillemet-left
 226                         # only if there is something before the space
 227                         '/(.) (?=\\?|:|;|!|\\302\\273)/' => '\\1&nbsp;\\2',
 228                         # french spaces, Guillemet-right
 229                         '/(\\302\\253) /' => '\\1&nbsp;',
 230                         '/<center *>(.*)<\\/center *>/i' => '<div class="center">\\1</div>',
 231                 );
 232                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 233
 234                 # only once and last
 235                 $text = $this->doBlockLevels( $text, $linestart );
 236
 237                 $this->replaceLinkHolders( $text );
 238
 239                 # the position of the parserConvert() call should not be changed. it
 240                 # assumes that the links are all replaced and the only thing left
 241                 # is the <nowiki> mark.
 242                 # Side-effects: this calls $this->mOutput->setTitleText()
 243                 $text = $wgContLang->parserConvert( $text, $this );
 244
 245                 $text = $this->unstripNoWiki( $text, $this->mStripState );
 246
 247                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 248
 249                 $text = Sanitizer::normalizeCharReferences( $text );
 250
 251                 if (($wgUseTidy and $this->mOptions->mTidy) or $wgAlwaysUseTidy) {
 252                         $text = Parser::tidy($text);
 253                 } else {
 254                         # attempt to sanitize at least some nesting problems
 255                         # (bug #2702 and quite a few others)
 256                         $tidyregs = array(
 257                                 # ''Something [http://www.cool.com cool''] -->
 258                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 259                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 260                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 261                                 # fix up an anchor inside another anchor, only
 262                                 # at least for a single single nested link (bug 3695)
 263                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 264                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 265                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 266                                 # contains a div, so fix it up here; replace
 267                                 # div with escaped text
 268                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 269                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 270                                 # remove empty italic or bold tag pairs, some
 271                                 # introduced by rules above
 272                                 '/<([bi])><\/\\1>/' => ''
 273                         );
 274
 275                         $text = preg_replace(
 276                                 array_keys( $tidyregs ),
 277                                 array_values( $tidyregs ),
 278                                 $text );
 279                 }
 280
 281                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 282
 283                 $this->mOutput->setText( $text );
 284                 wfProfileOut( $fname );
 285
 286                 return $this->mOutput;
 287         }
 288
 289         /**
 290          * Get a random string
 291          *
 292          * @access private
 293          * @static
 294          */
 295         function getRandomString() {
 296                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 297         }
 298
 299         function &getTitle() { return $this->mTitle; }
 300         function getOptions() { return $this->mOptions; }
 301
 302         /**
 303          * Replaces all occurrences of <$tag>content</$tag> in the text
 304          * with a random marker and returns the new text. the output parameter
 305          * $content will be an associative array filled with data on the form
 306          * $unique_marker => content.
 307          *
 308          * If $content is already set, the additional entries will be appended
 309          * If $tag is set to STRIP_COMMENTS, the function will extract
 310          * <!-- HTML comments -->
 311          *
 312          * @access private
 313          * @static
 314          */
 315         function extractTagsAndParams($tag, $text, &$content, &$tags, &$params, $uniq_prefix = ''){
 316                 $rnd = $uniq_prefix . '-' . $tag . Parser::getRandomString();
 317                 if ( !$content ) {
 318                         $content = array( );
 319                 }
 320                 $n = 1;
 321                 $stripped = '';
 322
 323                 if ( !$tags ) {
 324                         $tags = array( );
 325                 }
 326
 327                 if ( !$params ) {
 328                         $params = array( );
 329                 }
 330
 331                 if( $tag == STRIP_COMMENTS ) {
 332                         $start = '/<!--()()/';
 333                         $end   = '/-->/';
 334                 } else {
 335                         $start = "/<$tag(\\s+[^\\/>]*|\\s*)(\\/?)>/i";
 336                         $end   = "/<\\/$tag\\s*>/i";
 337                 }
 338
 339                 while ( '' != $text ) {
 340                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 341                         $stripped .= $p[0];
 342                         if( count( $p ) < 4 ) {
 343                                 break;
 344                         }
 345                         $attributes = $p[1];
 346                         $empty      = $p[2];
 347                         $inside     = $p[3];
 348
 349                         $marker = $rnd . sprintf('%08X', $n++);
 350                         $stripped .= $marker;
 351
 352                         $tags[$marker] = "<$tag$attributes$empty>";
 353                         $params[$marker] = Sanitizer::decodeTagAttributes( $attributes );
 354
 355                         if ( $empty === '/' ) {
 356                                 // Empty element tag, <tag />
 357                                 $content[$marker] = null;
 358                                 $text = $inside;
 359                         } else {
 360                                 $q = preg_split( $end, $inside, 2 );
 361                                 $content[$marker] = $q[0];
 362                                 if( count( $q ) < 2 ) {
 363                                         # No end tag -- let it run out to the end of the text.
 364                                         break;
 365                                 } else {
 366                                         $text = $q[1];
 367                                 }
 368                         }
 369                 }
 370                 return $stripped;
 371         }
 372
 373         /**
 374          * Wrapper function for extractTagsAndParams
 375          * for cases where $tags and $params isn't needed
 376          * i.e. where tags will never have params, like <nowiki>
 377          *
 378          * @access private
 379          * @static
 380          */
 381         function extractTags( $tag, $text, &$content, $uniq_prefix = '' ) {
 382                 $dummy_tags = array();
 383                 $dummy_params = array();
 384
 385                 return Parser::extractTagsAndParams( $tag, $text, $content,
 386                         $dummy_tags, $dummy_params, $uniq_prefix );
 387         }
 388
 389         /**
 390          * Strips and renders nowiki, pre, math, hiero
 391          * If $render is set, performs necessary rendering operations on plugins
 392          * Returns the text, and fills an array with data needed in unstrip()
 393          * If the $state is already a valid strip state, it adds to the state
 394          *
 395          * @param bool $stripcomments when set, HTML comments <!-- like this -->
 396          *  will be stripped in addition to other tags. This is important
 397          *  for section editing, where these comments cause confusion when
 398          *  counting the sections in the wikisource
 399          *
 400          * @access private
 401          */
 402         function strip( $text, &$state, $stripcomments = false ) {
 403                 $render = ($this->mOutputType == OT_HTML);
 404                 $html_content = array();
 405                 $nowiki_content = array();
 406                 $math_content = array();
 407                 $pre_content = array();
 408                 $comment_content = array();
 409                 $ext_content = array();
 410                 $ext_tags = array();
 411                 $ext_params = array();
 412                 $gallery_content = array();
 413
 414                 # Replace any instances of the placeholders
 415                 $uniq_prefix = $this->mUniqPrefix;
 416                 #$text = str_replace( $uniq_prefix, wfHtmlEscapeFirst( $uniq_prefix ), $text );
 417
 418                 # html
 419                 global $wgRawHtml;
 420                 if( $wgRawHtml ) {
 421                         $text = Parser::extractTags('html', $text, $html_content, $uniq_prefix);
 422                         foreach( $html_content as $marker => $content ) {
 423                                 if ($render ) {
 424                                         # Raw and unchecked for validity.
 425                                         $html_content[$marker] = $content;
 426                                 } else {
 427                                         $html_content[$marker] = '<html>'.$content.'</html>';
 428                                 }
 429                         }
 430                 }
 431
 432                 # nowiki
 433                 $text = Parser::extractTags('nowiki', $text, $nowiki_content, $uniq_prefix);
 434                 foreach( $nowiki_content as $marker => $content ) {
 435                         if( $render ){
 436                                 $nowiki_content[$marker] = wfEscapeHTMLTagsOnly( $content );
 437                         } else {
 438                                 $nowiki_content[$marker] = '<nowiki>'.$content.'</nowiki>';
 439                         }
 440                 }
 441
 442                 # math
 443                 if( $this->mOptions->getUseTeX() ) {
 444                         $text = Parser::extractTags('math', $text, $math_content, $uniq_prefix);
 445                         foreach( $math_content as $marker => $content ){
 446                                 if( $render ) {
 447                                         $math_content[$marker] = renderMath( $content );
 448                                 } else {
 449                                         $math_content[$marker] = '<math>'.$content.'</math>';
 450                                 }
 451                         }
 452                 }
 453
 454                 # pre
 455                 $text = Parser::extractTags('pre', $text, $pre_content, $uniq_prefix);
 456                 foreach( $pre_content as $marker => $content ){
 457                         if( $render ){
 458                                 $pre_content[$marker] = '<pre>' . wfEscapeHTMLTagsOnly( $content ) . '</pre>';
 459                         } else {
 460                                 $pre_content[$marker] = '<pre>'.$content.'</pre>';
 461                         }
 462                 }
 463
 464                 # gallery
 465                 $text = Parser::extractTags('gallery', $text, $gallery_content, $uniq_prefix);
 466                 foreach( $gallery_content as $marker => $content ) {
 467                         require_once( 'ImageGallery.php' );
 468                         if ( $render ) {
 469                                 $gallery_content[$marker] = $this->renderImageGallery( $content );
 470                         } else {
 471                                 $gallery_content[$marker] = '<gallery>'.$content.'</gallery>';
 472                         }
 473                 }
 474
 475                 # Comments
 476                 if($stripcomments) {
 477                         $text = Parser::extractTags(STRIP_COMMENTS, $text, $comment_content, $uniq_prefix);
 478                         foreach( $comment_content as $marker => $content ){
 479                                 $comment_content[$marker] = '<!--'.$content.'-->';
 480                         }
 481                 }
 482
 483                 # Extensions
 484                 foreach ( $this->mTagHooks as $tag => $callback ) {
 485                         $ext_content[$tag] = array();
 486                         $text = Parser::extractTagsAndParams( $tag, $text, $ext_content[$tag],
 487                                 $ext_tags[$tag], $ext_params[$tag], $uniq_prefix );
 488                         foreach( $ext_content[$tag] as $marker => $content ) {
 489                                 $full_tag = $ext_tags[$tag][$marker];
 490                                 $params = $ext_params[$tag][$marker];
 491                                 if ( $render )
 492                                         $ext_content[$tag][$marker] = call_user_func_array( $callback, array( $content, $params, &$this ) );
 493                                 else {
 494                                         if ( is_null( $content ) ) {
 495                                                 // Empty element tag
 496                                                 $ext_content[$tag][$marker] = $full_tag;
 497                                         } else {
 498                                                 $ext_content[$tag][$marker] = "$full_tag$content</$tag>";
 499                                         }
 500                                 }
 501                         }
 502                 }
 503
 504                 # Merge state with the pre-existing state, if there is one
 505                 if ( $state ) {
 506                         $state['html'] = $state['html'] + $html_content;
 507                         $state['nowiki'] = $state['nowiki'] + $nowiki_content;
 508                         $state['math'] = $state['math'] + $math_content;
 509                         $state['pre'] = $state['pre'] + $pre_content;
 510                         $state['gallery'] = $state['gallery'] + $gallery_content;
 511                         $state['comment'] = $state['comment'] + $comment_content;
 512
 513                         foreach( $ext_content as $tag => $array ) {
 514                                 if ( array_key_exists( $tag, $state ) ) {
 515                                         $state[$tag] = $state[$tag] + $array;
 516                                 }
 517                         }
 518                 } else {
 519                         $state = array(
 520                           'html' => $html_content,
 521                           'nowiki' => $nowiki_content,
 522                           'math' => $math_content,
 523                           'pre' => $pre_content,
 524                           'gallery' => $gallery_content,
 525                           'comment' => $comment_content,
 526                         ) + $ext_content;
 527                 }
 528                 return $text;
 529         }
 530
 531         /**
 532          * restores pre, math, and hiero removed by strip()
 533          *
 534          * always call unstripNoWiki() after this one
 535          * @access private
 536          */
 537         function unstrip( $text, &$state ) {
 538                 if ( !is_array( $state ) ) {
 539                         return $text;
 540                 }
 541
 542                 # Must expand in reverse order, otherwise nested tags will be corrupted
 543                 foreach( array_reverse( $state, true ) as $tag => $contentDict ) {
 544                         if( $tag != 'nowiki' && $tag != 'html' ) {
 545                                 foreach( array_reverse( $contentDict, true ) as $uniq => $content ) {
 546                                         $text = str_replace( $uniq, $content, $text );
 547                                 }
 548                         }
 549                 }
 550
 551                 return $text;
 552         }
 553
 554         /**
 555          * always call this after unstrip() to preserve the order
 556          *
 557          * @access private
 558          */
 559         function unstripNoWiki( $text, &$state ) {
 560                 if ( !is_array( $state ) ) {
 561                         return $text;
 562                 }
 563
 564                 # Must expand in reverse order, otherwise nested tags will be corrupted
 565                 for ( $content = end($state['nowiki']); $content !== false; $content = prev( $state['nowiki'] ) ) {
 566                         $text = str_replace( key( $state['nowiki'] ), $content, $text );
 567                 }
 568
 569                 global $wgRawHtml;
 570                 if ($wgRawHtml) {
 571                         for ( $content = end($state['html']); $content !== false; $content = prev( $state['html'] ) ) {
 572                                 $text = str_replace( key( $state['html'] ), $content, $text );
 573                         }
 574                 }
 575
 576                 return $text;
 577         }
 578
 579         /**
 580          * Add an item to the strip state
 581          * Returns the unique tag which must be inserted into the stripped text
 582          * The tag will be replaced with the original text in unstrip()
 583          *
 584          * @access private
 585          */
 586         function insertStripItem( $text, &$state ) {
 587                 $rnd = $this->mUniqPrefix . '-item' . Parser::getRandomString();
 588                 if ( !$state ) {
 589                         $state = array(
 590                           'html' => array(),
 591                           'nowiki' => array(),
 592                           'math' => array(),
 593                           'pre' => array(),
 594                           'comment' => array(),
 595                           'gallery' => array(),
 596                         );
 597                 }
 598                 $state['item'][$rnd] = $text;
 599                 return $rnd;
 600         }
 601
 602         /**
 603          * Interface with html tidy, used if $wgUseTidy = true.
 604          * If tidy isn't able to correct the markup, the original will be
 605          * returned in all its glory with a warning comment appended.
 606          *
 607          * Either the external tidy program or the in-process tidy extension
 608          * will be used depending on availability. Override the default
 609          * $wgTidyInternal setting to disable the internal if it's not working.
 610          *
 611          * @param string $text Hideous HTML input
 612          * @return string Corrected HTML output
 613          * @access public
 614          * @static
 615          */
 616         function tidy( $text ) {
 617                 global $wgTidyInternal;
 618                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 619 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 620 '<head><title>test</title></head><body>'.$text.'</body></html>';
 621                 if( $wgTidyInternal ) {
 622                         $correctedtext = Parser::internalTidy( $wrappedtext );
 623                 } else {
 624                         $correctedtext = Parser::externalTidy( $wrappedtext );
 625                 }
 626                 if( is_null( $correctedtext ) ) {
 627                         wfDebug( "Tidy error detected!\n" );
 628                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 629                 }
 630                 return $correctedtext;
 631         }
 632
 633         /**
 634          * Spawn an external HTML tidy process and get corrected markup back from it.
 635          *
 636          * @access private
 637          * @static
 638          */
 639         function externalTidy( $text ) {
 640                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 641                 $fname = 'Parser::externalTidy';
 642                 wfProfileIn( $fname );
 643
 644                 $cleansource = '';
 645                 $opts = ' -utf8';
 646
 647                 $descriptorspec = array(
 648                         0 => array('pipe', 'r'),
 649                         1 => array('pipe', 'w'),
 650                         2 => array('file', '/dev/null', 'a')
 651                 );
 652                 $pipes = array();
 653                 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 654                 if (is_resource($process)) {
 655                         // Theoretically, this style of communication could cause a deadlock
 656                         // here. If the stdout buffer fills up, then writes to stdin could
 657                         // block. This doesn't appear to happen with tidy, because tidy only
 658                         // writes to stdout after it's finished reading from stdin. Search
 659                         // for tidyParseStdin and tidySaveStdout in console/tidy.c
 660                         fwrite($pipes[0], $text);
 661                         fclose($pipes[0]);
 662                         while (!feof($pipes[1])) {
 663                                 $cleansource .= fgets($pipes[1], 1024);
 664                         }
 665                         fclose($pipes[1]);
 666                         proc_close($process);
 667                 }
 668
 669                 wfProfileOut( $fname );
 670
 671                 if( $cleansource == '' && $text != '') {
 672                         // Some kind of error happened, so we couldn't get the corrected text.
 673                         // Just give up; we'll use the source text and append a warning.
 674                         return null;
 675                 } else {
 676                         return $cleansource;
 677                 }
 678         }
 679
 680         /**
 681          * Use the HTML tidy PECL extension to use the tidy library in-process,
 682          * saving the overhead of spawning a new process. Currently written to
 683          * the PHP 4.3.x version of the extension, may not work on PHP 5.
 684          *
 685          * 'pear install tidy' should be able to compile the extension module.
 686          *
 687          * @access private
 688          * @static
 689          */
 690         function internalTidy( $text ) {
 691                 global $wgTidyConf;
 692                 $fname = 'Parser::internalTidy';
 693                 wfProfileIn( $fname );
 694
 695                 tidy_load_config( $wgTidyConf );
 696                 tidy_set_encoding( 'utf8' );
 697                 tidy_parse_string( $text );
 698                 tidy_clean_repair();
 699                 if( tidy_get_status() == 2 ) {
 700                         // 2 is magic number for fatal error
 701                         // http://www.php.net/manual/en/function.tidy-get-status.php
 702                         $cleansource = null;
 703                 } else {
 704                         $cleansource = tidy_get_output();
 705                 }
 706                 wfProfileOut( $fname );
 707                 return $cleansource;
 708         }
 709
 710         /**
 711          * parse the wiki syntax used to render tables
 712          *
 713          * @access private
 714          */
 715         function doTableStuff ( $t ) {
 716                 $fname = 'Parser::doTableStuff';
 717                 wfProfileIn( $fname );
 718
 719                 $t = explode ( "\n" , $t ) ;
 720                 $td = array () ; # Is currently a td tag open?
 721                 $ltd = array () ; # Was it TD or TH?
 722                 $tr = array () ; # Is currently a tr tag open?
 723                 $ltr = array () ; # tr attributes
 724                 $has_opened_tr = array(); # Did this table open a <tr> element?
 725                 $indent_level = 0; # indent level of the table
 726                 foreach ( $t AS $k => $x )
 727                 {
 728                         $x = trim ( $x ) ;
 729                         $fc = substr ( $x , 0 , 1 ) ;
 730                         if ( preg_match( '/^(:*)\{\|(.*)$/', $x, $matches ) ) {
 731                                 $indent_level = strlen( $matches[1] );
 732
 733                                 $attributes = $this->unstripForHTML( $matches[2] );
 734
 735                                 $t[$k] = str_repeat( '<dl><dd>', $indent_level ) .
 736                                         '<table' . Sanitizer::fixTagAttributes ( $attributes, 'table' ) . '>' ;
 737                                 array_push ( $td , false ) ;
 738                                 array_push ( $ltd , '' ) ;
 739                                 array_push ( $tr , false ) ;
 740                                 array_push ( $ltr , '' ) ;
 741                                 array_push ( $has_opened_tr, false );
 742                         }
 743                         else if ( count ( $td ) == 0 ) { } # Don't do any of the following
 744                         else if ( '|}' == substr ( $x , 0 , 2 ) ) {
 745                                 $z = "</table>" . substr ( $x , 2);
 746                                 $l = array_pop ( $ltd ) ;
 747                                 if ( !array_pop ( $has_opened_tr ) ) $z = "<tr><td></td></tr>" . $z ;
 748                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 749                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 750                                 array_pop ( $ltr ) ;
 751                                 $t[$k] = $z . str_repeat( '</dd></dl>', $indent_level );
 752                         }
 753                         else if ( '|-' == substr ( $x , 0 , 2 ) ) { # Allows for |---------------
 754                                 $x = substr ( $x , 1 ) ;
 755                                 while ( $x != '' && substr ( $x , 0 , 1 ) == '-' ) $x = substr ( $x , 1 ) ;
 756                                 $z = '' ;
 757                                 $l = array_pop ( $ltd ) ;
 758                                 array_pop ( $has_opened_tr );
 759                                 array_push ( $has_opened_tr , true ) ;
 760                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 761                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 762                                 array_pop ( $ltr ) ;
 763                                 $t[$k] = $z ;
 764                                 array_push ( $tr , false ) ;
 765                                 array_push ( $td , false ) ;
 766                                 array_push ( $ltd , '' ) ;
 767                                 $attributes = $this->unstripForHTML( $x );
 768                                 array_push ( $ltr , Sanitizer::fixTagAttributes ( $attributes, 'tr' ) ) ;
 769                         }
 770                         else if ( '|' == $fc || '!' == $fc || '|+' == substr ( $x , 0 , 2 ) ) { # Caption
 771                                 # $x is a table row
 772                                 if ( '|+' == substr ( $x , 0 , 2 ) ) {
 773                                         $fc = '+' ;
 774                                         $x = substr ( $x , 1 ) ;
 775                                 }
 776                                 $after = substr ( $x , 1 ) ;
 777                                 if ( $fc == '!' ) $after = str_replace ( '!!' , '||' , $after ) ;
 778                                 $after = explode ( '||' , $after ) ;
 779                                 $t[$k] = '' ;
 780
 781                                 # Loop through each table cell
 782                                 foreach ( $after AS $theline )
 783                                 {
 784                                         $z = '' ;
 785                                         if ( $fc != '+' )
 786                                         {
 787                                                 $tra = array_pop ( $ltr ) ;
 788                                                 if ( !array_pop ( $tr ) ) $z = '<tr'.$tra.">\n" ;
 789                                                 array_push ( $tr , true ) ;
 790                                                 array_push ( $ltr , '' ) ;
 791                                                 array_pop ( $has_opened_tr );
 792                                                 array_push ( $has_opened_tr , true ) ;
 793                                         }
 794
 795                                         $l = array_pop ( $ltd ) ;
 796                                         if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 797                                         if ( $fc == '|' ) $l = 'td' ;
 798                                         else if ( $fc == '!' ) $l = 'th' ;
 799                                         else if ( $fc == '+' ) $l = 'caption' ;
 800                                         else $l = '' ;
 801                                         array_push ( $ltd , $l ) ;
 802
 803                                         # Cell parameters
 804                                         $y = explode ( '|' , $theline , 2 ) ;
 805                                         # Note that a '|' inside an invalid link should not
 806                                         # be mistaken as delimiting cell parameters
 807                                         if ( strpos( $y[0], '[[' ) !== false ) {
 808                                                 $y = array ($theline);
 809                                         }
 810                                         if ( count ( $y ) == 1 )
 811                                                 $y = "{$z}<{$l}>{$y[0]}" ;
 812                                         else {
 813                                                 $attributes = $this->unstripForHTML( $y[0] );
 814                                                 $y = "{$z}<{$l}".Sanitizer::fixTagAttributes($attributes, $l).">{$y[1]}" ;
 815                                         }
 816                                         $t[$k] .= $y ;
 817                                         array_push ( $td , true ) ;
 818                                 }
 819                         }
 820                 }
 821
 822                 # Closing open td, tr && table
 823                 while ( count ( $td ) > 0 )
 824                 {
 825                         $l = array_pop ( $ltd ) ;
 826                         if ( array_pop ( $td ) ) $t[] = '</td>' ;
 827                         if ( array_pop ( $tr ) ) $t[] = '</tr>' ;
 828                         if ( !array_pop ( $has_opened_tr ) ) $t[] = "<tr><td></td></tr>" ;
 829                         $t[] = '</table>' ;
 830                 }
 831
 832                 $t = implode ( "\n" , $t ) ;
 833                 # special case: don't return empty table
 834                 if($t == "<table>\n<tr><td></td></tr>\n</table>")
 835                         $t = '';
 836                 wfProfileOut( $fname );
 837                 return $t ;
 838         }
 839
 840         /**
 841          * Helper function for parse() that transforms wiki markup into
 842          * HTML. Only called for $mOutputType == OT_HTML.
 843          *
 844          * @access private
 845          */
 846         function internalParse( $text ) {
 847                 $args = array();
 848                 $isMain = true;
 849                 $fname = 'Parser::internalParse';
 850                 wfProfileIn( $fname );
 851
 852                 # Remove <noinclude> tags and <includeonly> sections
 853                 $text = strtr( $text, array( '<onlyinclude>' => '' , '</onlyinclude>' => '' ) );
 854                 $text = strtr( $text, array( '<noinclude>' => '', '</noinclude>' => '') );
 855                 $text = preg_replace( '/<includeonly>.*?<\/includeonly>/s', '', $text );
 856
 857                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ) );
 858                 $text = $this->replaceVariables( $text, $args );
 859
 860                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 861
 862                 $text = $this->doHeadings( $text );
 863                 if($this->mOptions->getUseDynamicDates()) {
 864                         $df =& DateFormatter::getInstance();
 865                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 866                 }
 867                 $text = $this->doAllQuotes( $text );
 868                 $text = $this->replaceInternalLinks( $text );
 869                 $text = $this->replaceExternalLinks( $text );
 870
 871                 # replaceInternalLinks may sometimes leave behind
 872                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 873                 $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
 874
 875                 $text = $this->doMagicLinks( $text );
 876                 $text = $this->doTableStuff( $text );
 877                 $text = $this->formatHeadings( $text, $isMain );
 878
 879                 wfProfileOut( $fname );
 880                 return $text;
 881         }
 882
 883         /**
 884          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 885          * magic external links.
 886          *
 887          * @access private
 888          */
 889         function &doMagicLinks( &$text ) {
 890                 $text = $this->magicISBN( $text );
 891                 $text = $this->magicRFC( $text, 'RFC ', 'rfcurl' );
 892                 $text = $this->magicRFC( $text, 'PMID ', 'pubmedurl' );
 893                 return $text;
 894         }
 895
 896         /**
 897          * Parse headers and return html
 898          *
 899          * @access private
 900          */
 901         function doHeadings( $text ) {
 902                 $fname = 'Parser::doHeadings';
 903                 wfProfileIn( $fname );
 904                 for ( $i = 6; $i >= 1; --$i ) {
 905                         $h = str_repeat( '=', $i );
 906                         $text = preg_replace( "/^{$h}(.+){$h}(\\s|$)/m",
 907                           "<h{$i}>\\1</h{$i}>\\2", $text );
 908                 }
 909                 wfProfileOut( $fname );
 910                 return $text;
 911         }
 912
 913         /**
 914          * Replace single quotes with HTML markup
 915          * @access private
 916          * @return string the altered text
 917          */
 918         function doAllQuotes( $text ) {
 919                 $fname = 'Parser::doAllQuotes';
 920                 wfProfileIn( $fname );
 921                 $outtext = '';
 922                 $lines = explode( "\n", $text );
 923                 foreach ( $lines as $line ) {
 924                         $outtext .= $this->doQuotes ( $line ) . "\n";
 925                 }
 926                 $outtext = substr($outtext, 0,-1);
 927                 wfProfileOut( $fname );
 928                 return $outtext;
 929         }
 930
 931         /**
 932          * Helper function for doAllQuotes()
 933          * @access private
 934          */
 935         function doQuotes( $text ) {
 936                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
 937                 if ( count( $arr ) == 1 )
 938                         return $text;
 939                 else
 940                 {
 941                         # First, do some preliminary work. This may shift some apostrophes from
 942                         # being mark-up to being text. It also counts the number of occurrences
 943                         # of bold and italics mark-ups.
 944                         $i = 0;
 945                         $numbold = 0;
 946                         $numitalics = 0;
 947                         foreach ( $arr as $r )
 948                         {
 949                                 if ( ( $i % 2 ) == 1 )
 950                                 {
 951                                         # If there are ever four apostrophes, assume the first is supposed to
 952                                         # be text, and the remaining three constitute mark-up for bold text.
 953                                         if ( strlen( $arr[$i] ) == 4 )
 954                                         {
 955                                                 $arr[$i-1] .= "'";
 956                                                 $arr[$i] = "'''";
 957                                         }
 958                                         # If there are more than 5 apostrophes in a row, assume they're all
 959                                         # text except for the last 5.
 960                                         else if ( strlen( $arr[$i] ) > 5 )
 961                                         {
 962                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
 963                                                 $arr[$i] = "'''''";
 964                                         }
 965                                         # Count the number of occurrences of bold and italics mark-ups.
 966                                         # We are not counting sequences of five apostrophes.
 967                                         if ( strlen( $arr[$i] ) == 2 ) $numitalics++;  else
 968                                         if ( strlen( $arr[$i] ) == 3 ) $numbold++;     else
 969                                         if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
 970                                 }
 971                                 $i++;
 972                         }
 973
 974                         # If there is an odd number of both bold and italics, it is likely
 975                         # that one of the bold ones was meant to be an apostrophe followed
 976                         # by italics. Which one we cannot know for certain, but it is more
 977                         # likely to be one that has a single-letter word before it.
 978                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
 979                         {
 980                                 $i = 0;
 981                                 $firstsingleletterword = -1;
 982                                 $firstmultiletterword = -1;
 983                                 $firstspace = -1;
 984                                 foreach ( $arr as $r )
 985                                 {
 986                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
 987                                         {
 988                                                 $x1 = substr ($arr[$i-1], -1);
 989                                                 $x2 = substr ($arr[$i-1], -2, 1);
 990                                                 if ($x1 == ' ') {
 991                                                         if ($firstspace == -1) $firstspace = $i;
 992                                                 } else if ($x2 == ' ') {
 993                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
 994                                                 } else {
 995                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
 996                                                 }
 997                                         }
 998                                         $i++;
 999                                 }
1000
1001                                 # If there is a single-letter word, use it!
1002                                 if ($firstsingleletterword > -1)
1003                                 {
1004                                         $arr [ $firstsingleletterword ] = "''";
1005                                         $arr [ $firstsingleletterword-1 ] .= "'";
1006                                 }
1007                                 # If not, but there's a multi-letter word, use that one.
1008                                 else if ($firstmultiletterword > -1)
1009                                 {
1010                                         $arr [ $firstmultiletterword ] = "''";
1011                                         $arr [ $firstmultiletterword-1 ] .= "'";
1012                                 }
1013                                 # ... otherwise use the first one that has neither.
1014                                 # (notice that it is possible for all three to be -1 if, for example,
1015                                 # there is only one pentuple-apostrophe in the line)
1016                                 else if ($firstspace > -1)
1017                                 {
1018                                         $arr [ $firstspace ] = "''";
1019                                         $arr [ $firstspace-1 ] .= "'";
1020                                 }
1021                         }
1022
1023                         # Now let's actually convert our apostrophic mush to HTML!
1024                         $output = '';
1025                         $buffer = '';
1026                         $state = '';
1027                         $i = 0;
1028                         foreach ($arr as $r)
1029                         {
1030                                 if (($i % 2) == 0)
1031                                 {
1032                                         if ($state == 'both')
1033                                                 $buffer .= $r;
1034                                         else
1035                                                 $output .= $r;
1036                                 }
1037                                 else
1038                                 {
1039                                         if (strlen ($r) == 2)
1040                                         {
1041                                                 if ($state == 'i')
1042                                                 { $output .= '</i>'; $state = ''; }
1043                                                 else if ($state == 'bi')
1044                                                 { $output .= '</i>'; $state = 'b'; }
1045                                                 else if ($state == 'ib')
1046                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1047                                                 else if ($state == 'both')
1048                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1049                                                 else # $state can be 'b' or ''
1050                                                 { $output .= '<i>'; $state .= 'i'; }
1051                                         }
1052                                         else if (strlen ($r) == 3)
1053                                         {
1054                                                 if ($state == 'b')
1055                                                 { $output .= '</b>'; $state = ''; }
1056                                                 else if ($state == 'bi')
1057                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1058                                                 else if ($state == 'ib')
1059                                                 { $output .= '</b>'; $state = 'i'; }
1060                                                 else if ($state == 'both')
1061                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1062                                                 else # $state can be 'i' or ''
1063                                                 { $output .= '<b>'; $state .= 'b'; }
1064                                         }
1065                                         else if (strlen ($r) == 5)
1066                                         {
1067                                                 if ($state == 'b')
1068                                                 { $output .= '</b><i>'; $state = 'i'; }
1069                                                 else if ($state == 'i')
1070                                                 { $output .= '</i><b>'; $state = 'b'; }
1071                                                 else if ($state == 'bi')
1072                                                 { $output .= '</i></b>'; $state = ''; }
1073                                                 else if ($state == 'ib')
1074                                                 { $output .= '</b></i>'; $state = ''; }
1075                                                 else if ($state == 'both')
1076                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1077                                                 else # ($state == '')
1078                                                 { $buffer = ''; $state = 'both'; }
1079                                         }
1080                                 }
1081                                 $i++;
1082                         }
1083                         # Now close all remaining tags.  Notice that the order is important.
1084                         if ($state == 'b' || $state == 'ib')
1085                                 $output .= '</b>';
1086                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
1087                                 $output .= '</i>';
1088                         if ($state == 'bi')
1089                                 $output .= '</b>';
1090                         if ($state == 'both')
1091                                 $output .= '<b><i>'.$buffer.'</i></b>';
1092                         return $output;
1093                 }
1094         }
1095
1096         /**
1097          * Replace external links
1098          *
1099          * Note: this is all very hackish and the order of execution matters a lot.
1100          * Make sure to run maintenance/parserTests.php if you change this code.
1101          *
1102          * @access private
1103          */
1104         function replaceExternalLinks( $text ) {
1105                 global $wgContLang;
1106                 $fname = 'Parser::replaceExternalLinks';
1107                 wfProfileIn( $fname );
1108
1109                 $sk =& $this->mOptions->getSkin();
1110
1111                 $bits = preg_split( EXT_LINK_BRACKETED, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1112
1113                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
1114
1115                 $i = 0;
1116                 while ( $i<count( $bits ) ) {
1117                         $url = $bits[$i++];
1118                         $protocol = $bits[$i++];
1119                         $text = $bits[$i++];
1120                         $trail = $bits[$i++];
1121
1122                         # The characters '<' and '>' (which were escaped by
1123                         # removeHTMLtags()) should not be included in
1124                         # URLs, per RFC 2396.
1125                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1126                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1127                                 $url = substr($url, 0, $m2[0][1]);
1128                         }
1129
1130                         # If the link text is an image URL, replace it with an <img> tag
1131                         # This happened by accident in the original parser, but some people used it extensively
1132                         $img = $this->maybeMakeExternalImage( $text );
1133                         if ( $img !== false ) {
1134                                 $text = $img;
1135                         }
1136
1137                         $dtrail = '';
1138
1139                         # Set linktype for CSS - if URL==text, link is essentially free
1140                         $linktype = ($text == $url) ? 'free' : 'text';
1141
1142                         # No link text, e.g. [http://domain.tld/some.link]
1143                         if ( $text == '' ) {
1144                                 # Autonumber if allowed
1145                                 if ( strpos( HTTP_PROTOCOLS, str_replace('/','\/', $protocol) ) !== false ) {
1146                                         $text = '[' . ++$this->mAutonumber . ']';
1147                                         $linktype = 'autonumber';
1148                                 } else {
1149                                         # Otherwise just use the URL
1150                                         $text = htmlspecialchars( $url );
1151                                         $linktype = 'free';
1152                                 }
1153                         } else {
1154                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1155                                 # Check for trail
1156                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1157                         }
1158
1159                         $text = $wgContLang->markNoConversion($text);
1160
1161                         # Replace &amp; from obsolete syntax with &.
1162                         # All HTML entities will be escaped by makeExternalLink()
1163                         $url = str_replace( '&amp;', '&', $url );
1164
1165                         # Process the trail (i.e. everything after this link up until start of the next link),
1166                         # replacing any non-bracketed links
1167                         $trail = $this->replaceFreeExternalLinks( $trail );
1168
1169                         # Use the encoded URL
1170                         # This means that users can paste URLs directly into the text
1171                         # Funny characters like &ouml; aren't valid in URLs anyway
1172                         # This was changed in August 2004
1173                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype ) . $dtrail . $trail;
1174
1175                         # Register link in the output object.
1176                         # Replace unnecessary URL escape codes with the referenced character
1177                         # This prevents spammers from hiding links from the filters
1178                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1179                         $this->mOutput->addExternalLink( $pasteurized );
1180                 }
1181
1182                 wfProfileOut( $fname );
1183                 return $s;
1184         }
1185
1186         /**
1187          * Replace anything that looks like a URL with a link
1188          * @access private
1189          */
1190         function replaceFreeExternalLinks( $text ) {
1191                 global $wgContLang;
1192                 $fname = 'Parser::replaceFreeExternalLinks';
1193                 wfProfileIn( $fname );
1194
1195                 $bits = preg_split( '/(\b(?:' . wfUrlProtocols() . '))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1196                 $s = array_shift( $bits );
1197                 $i = 0;
1198
1199                 $sk =& $this->mOptions->getSkin();
1200
1201                 while ( $i < count( $bits ) ){
1202                         $protocol = $bits[$i++];
1203                         $remainder = $bits[$i++];
1204
1205                         if ( preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1206                                 # Found some characters after the protocol that look promising
1207                                 $url = $protocol . $m[1];
1208                                 $trail = $m[2];
1209
1210                                 # special case: handle urls as url args:
1211                                 # http://www.example.com/foo?=http://www.example.com/bar
1212                                 if(strlen($trail) == 0 &&
1213                                         isset($bits[$i]) &&
1214                                         preg_match('/^'. wfUrlProtocols() . '$/S', $bits[$i]) &&
1215                                         preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $bits[$i + 1], $m ))
1216                                 {
1217                                         # add protocol, arg
1218                                         $url .= $bits[$i] . $bits[$i + 1]; # protocol, url as arg to previous link
1219                                         $i += 2;
1220                                         $trail = $m[2];
1221                                 }
1222
1223                                 # The characters '<' and '>' (which were escaped by
1224                                 # removeHTMLtags()) should not be included in
1225                                 # URLs, per RFC 2396.
1226                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1227                                         $trail = substr($url, $m2[0][1]) . $trail;
1228                                         $url = substr($url, 0, $m2[0][1]);
1229                                 }
1230
1231                                 # Move trailing punctuation to $trail
1232                                 $sep = ',;\.:!?';
1233                                 # If there is no left bracket, then consider right brackets fair game too
1234                                 if ( strpos( $url, '(' ) === false ) {
1235                                         $sep .= ')';
1236                                 }
1237
1238                                 $numSepChars = strspn( strrev( $url ), $sep );
1239                                 if ( $numSepChars ) {
1240                                         $trail = substr( $url, -$numSepChars ) . $trail;
1241                                         $url = substr( $url, 0, -$numSepChars );
1242                                 }
1243
1244                                 # Replace &amp; from obsolete syntax with &.
1245                                 # All HTML entities will be escaped by makeExternalLink()
1246                                 # or maybeMakeExternalImage()
1247                                 $url = str_replace( '&amp;', '&', $url );
1248
1249                                 # Is this an external image?
1250                                 $text = $this->maybeMakeExternalImage( $url );
1251                                 if ( $text === false ) {
1252                                         # Not an image, make a link
1253                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free' );
1254                                         # Register it in the output object...
1255                                         # Replace unnecessary URL escape codes with their equivalent characters
1256                                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1257                                         $this->mOutput->addExternalLink( $pasteurized );
1258                                 }
1259                                 $s .= $text . $trail;
1260                         } else {
1261                                 $s .= $protocol . $remainder;
1262                         }
1263                 }
1264                 wfProfileOut( $fname );
1265                 return $s;
1266         }
1267
1268         /**
1269          * Replace unusual URL escape codes with their equivalent characters
1270          * @param string
1271          * @return string
1272          * @static
1273          * @fixme This can merge genuinely required bits in the path or query string,
1274          *        breaking legit URLs. A proper fix would treat the various parts of
1275          *        the URL differently; as a workaround, just use the output for
1276          *        statistical records, not for actual linking/output.
1277          */
1278         function replaceUnusualEscapes( $url ) {
1279                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1280                         array( 'Parser', 'replaceUnusualEscapesCallback' ), $url );
1281         }
1282
1283         /**
1284          * Callback function used in replaceUnusualEscapes().
1285          * Replaces unusual URL escape codes with their equivalent character
1286          * @static
1287          * @access private
1288          */
1289         function replaceUnusualEscapesCallback( $matches ) {
1290                 $char = urldecode( $matches[0] );
1291                 $ord = ord( $char );
1292                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1293                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1294                         // No, shouldn't be escaped
1295                         return $char;
1296                 } else {
1297                         // Yes, leave it escaped
1298                         return $matches[0];
1299                 }
1300         }
1301
1302         /**
1303          * make an image if it's allowed, either through the global
1304          * option or through the exception
1305          * @access private
1306          */
1307         function maybeMakeExternalImage( $url ) {
1308                 $sk =& $this->mOptions->getSkin();
1309                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1310                 $imagesexception = !empty($imagesfrom);
1311                 $text = false;
1312                 if ( $this->mOptions->getAllowExternalImages()
1313                      || ( $imagesexception && strpos( $url, $imagesfrom ) === 0 ) ) {
1314                         if ( preg_match( EXT_IMAGE_REGEX, $url ) ) {
1315                                 # Image found
1316                                 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
1317                         }
1318                 }
1319                 return $text;
1320         }
1321
1322         /**
1323          * Process [[ ]] wikilinks
1324          *
1325          * @access private
1326          */
1327         function replaceInternalLinks( $s ) {
1328                 global $wgContLang;
1329                 static $fname = 'Parser::replaceInternalLinks' ;
1330
1331                 wfProfileIn( $fname );
1332
1333                 wfProfileIn( $fname.'-setup' );
1334                 static $tc = FALSE;
1335                 # the % is needed to support urlencoded titles as well
1336                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1337
1338                 $sk =& $this->mOptions->getSkin();
1339
1340                 #split the entire text string on occurences of [[
1341                 $a = explode( '[[', ' ' . $s );
1342                 #get the first element (all text up to first [[), and remove the space we added
1343                 $s = array_shift( $a );
1344                 $s = substr( $s, 1 );
1345
1346                 # Match a link having the form [[namespace:link|alternate]]trail
1347                 static $e1 = FALSE;
1348                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1349                 # Match cases where there is no "]]", which might still be images
1350                 static $e1_img = FALSE;
1351                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1352                 # Match the end of a line for a word that's not followed by whitespace,
1353                 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1354                 $e2 = wfMsgForContent( 'linkprefix' );
1355
1356                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1357
1358                 if( is_null( $this->mTitle ) ) {
1359                         wfDebugDieBacktrace( 'nooo' );
1360                 }
1361                 $nottalk = !$this->mTitle->isTalkPage();
1362
1363                 if ( $useLinkPrefixExtension ) {
1364                         if ( preg_match( $e2, $s, $m ) ) {
1365                                 $first_prefix = $m[2];
1366                         } else {
1367                                 $first_prefix = false;
1368                         }
1369                 } else {
1370                         $prefix = '';
1371                 }
1372
1373                 $selflink = $this->mTitle->getPrefixedText();
1374                 wfProfileOut( $fname.'-setup' );
1375
1376                 $checkVariantLink = sizeof($wgContLang->getVariants())>1;
1377                 $useSubpages = $this->areSubpagesAllowed();
1378
1379                 # Loop for each link
1380                 for ($k = 0; isset( $a[$k] ); $k++) {
1381                         $line = $a[$k];
1382                         if ( $useLinkPrefixExtension ) {
1383                                 wfProfileIn( $fname.'-prefixhandling' );
1384                                 if ( preg_match( $e2, $s, $m ) ) {
1385                                         $prefix = $m[2];
1386                                         $s = $m[1];
1387                                 } else {
1388                                         $prefix='';
1389                                 }
1390                                 # first link
1391                                 if($first_prefix) {
1392                                         $prefix = $first_prefix;
1393                                         $first_prefix = false;
1394                                 }
1395                                 wfProfileOut( $fname.'-prefixhandling' );
1396                         }
1397
1398                         $might_be_img = false;
1399
1400                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1401                                 $text = $m[2];
1402                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1403                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1404                                 # the real problem is with the $e1 regex
1405                                 # See bug 1300.
1406                                 #
1407                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1408                                 # and no image is in sight. See bug 2095.
1409                                 #
1410                                 if( $text !== '' &&
1411                                         preg_match( "/^\](.*)/s", $m[3], $n ) &&
1412                                         strpos($text, '[') !== false
1413                                 )
1414                                 {
1415                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1416                                         $m[3] = $n[1];
1417                                 }
1418                                 # fix up urlencoded title texts
1419                                 if(preg_match('/%/', $m[1] )) $m[1] = urldecode($m[1]);
1420                                 $trail = $m[3];
1421                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1422                                 $might_be_img = true;
1423                                 $text = $m[2];
1424                                 if(preg_match('/%/', $m[1] )) $m[1] = urldecode($m[1]);
1425                                 $trail = "";
1426                         } else { # Invalid form; output directly
1427                                 $s .= $prefix . '[[' . $line ;
1428                                 continue;
1429                         }
1430
1431                         # Don't allow internal links to pages containing
1432                         # PROTO: where PROTO is a valid URL protocol; these
1433                         # should be external links.
1434                         if (preg_match('/^(\b(?:' . wfUrlProtocols() . '))/', $m[1])) {
1435                                 $s .= $prefix . '[[' . $line ;
1436                                 continue;
1437                         }
1438
1439                         # Make subpage if necessary
1440                         if( $useSubpages ) {
1441                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1442                         } else {
1443                                 $link = $m[1];
1444                         }
1445
1446                         $noforce = (substr($m[1], 0, 1) != ':');
1447                         if (!$noforce) {
1448                                 # Strip off leading ':'
1449                                 $link = substr($link, 1);
1450                         }
1451
1452                         $nt = Title::newFromText( $this->unstripNoWiki($link, $this->mStripState) );
1453                         if( !$nt ) {
1454                                 $s .= $prefix . '[[' . $line;
1455                                 continue;
1456                         }
1457
1458                         #check other language variants of the link
1459                         #if the article does not exist
1460                         if( $checkVariantLink
1461                             && $nt->getArticleID() == 0 ) {
1462                                 $wgContLang->findVariantLink($link, $nt);
1463                         }
1464
1465                         $ns = $nt->getNamespace();
1466                         $iw = $nt->getInterWiki();
1467
1468                         if ($might_be_img) { # if this is actually an invalid link
1469                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1470                                         $found = false;
1471                                         while (isset ($a[$k+1]) ) {
1472                                                 #look at the next 'line' to see if we can close it there
1473                                                 $spliced = array_splice( $a, $k + 1, 1 );
1474                                                 $next_line = array_shift( $spliced );
1475                                                 if( preg_match("/^(.*?]].*?)]](.*)$/sD", $next_line, $m) ) {
1476                                                 # the first ]] closes the inner link, the second the image
1477                                                         $found = true;
1478                                                         $text .= '[[' . $m[1];
1479                                                         $trail = $m[2];
1480                                                         break;
1481                                                 } elseif( preg_match("/^.*?]].*$/sD", $next_line, $m) ) {
1482                                                         #if there's exactly one ]] that's fine, we'll keep looking
1483                                                         $text .= '[[' . $m[0];
1484                                                 } else {
1485                                                         #if $next_line is invalid too, we need look no further
1486                                                         $text .= '[[' . $next_line;
1487                                                         break;
1488                                                 }
1489                                         }
1490                                         if ( !$found ) {
1491                                                 # we couldn't find the end of this imageLink, so output it raw
1492                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1493                                                 $text = $this->replaceInternalLinks($text);
1494                                                 $s .= $prefix . '[[' . $link . '|' . $text;
1495                                                 # note: no $trail, because without an end, there *is* no trail
1496                                                 continue;
1497                                         }
1498                                 } else { #it's not an image, so output it raw
1499                                         $s .= $prefix . '[[' . $link . '|' . $text;
1500                                         # note: no $trail, because without an end, there *is* no trail
1501                                         continue;
1502                                 }
1503                         }
1504
1505                         $wasblank = ( '' == $text );
1506                         if( $wasblank ) $text = $link;
1507
1508
1509                         # Link not escaped by : , create the various objects
1510                         if( $noforce ) {
1511
1512                                 # Interwikis
1513                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1514                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1515                                         $s = rtrim($s . "\n");
1516                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1517                                         continue;
1518                                 }
1519
1520                                 if ( $ns == NS_IMAGE ) {
1521                                         wfProfileIn( "$fname-image" );
1522                                         if ( !wfIsBadImage( $nt->getDBkey() ) ) {
1523                                                 # recursively parse links inside the image caption
1524                                                 # actually, this will parse them in any other parameters, too,
1525                                                 # but it might be hard to fix that, and it doesn't matter ATM
1526                                                 $text = $this->replaceExternalLinks($text);
1527                                                 $text = $this->replaceInternalLinks($text);
1528
1529                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1530                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text ) ) . $trail;
1531                                                 $this->mOutput->addImage( $nt->getDBkey() );
1532
1533                                                 wfProfileOut( "$fname-image" );
1534                                                 continue;
1535                                         }
1536                                         wfProfileOut( "$fname-image" );
1537
1538                                 }
1539
1540                                 if ( $ns == NS_CATEGORY ) {
1541                                         wfProfileIn( "$fname-category" );
1542                                         $s = rtrim($s . "\n"); # bug 87
1543
1544                                         if ( $wasblank ) {
1545                                                 if ( $this->mTitle->getNamespace() == NS_CATEGORY ) {
1546                                                         $sortkey = $this->mTitle->getText();
1547                                                 } else {
1548                                                         $sortkey = $this->mTitle->getPrefixedText();
1549                                                 }
1550                                         } else {
1551                                                 $sortkey = $text;
1552                                         }
1553                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1554                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1555                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1556
1557                                         /**
1558                                          * Strip the whitespace Category links produce, see bug 87
1559                                          * @todo We might want to use trim($tmp, "\n") here.
1560                                          */
1561                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1562
1563                                         wfProfileOut( "$fname-category" );
1564                                         continue;
1565                                 }
1566                         }
1567
1568                         if( ( $nt->getPrefixedText() === $selflink ) &&
1569                             ( $nt->getFragment() === '' ) ) {
1570                                 # Self-links are handled specially; generally de-link and change to bold.
1571                                 $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1572                                 continue;
1573                         }
1574
1575                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1576                         if( $ns == NS_MEDIA ) {
1577                                 $link = $sk->makeMediaLinkObj( $nt, $text );
1578                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1579                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1580                                 $this->mOutput->addImage( $nt->getDBkey() );
1581                                 continue;
1582                         } elseif( $ns == NS_SPECIAL ) {
1583                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1584                                 continue;
1585                         } elseif( $ns == NS_IMAGE ) {
1586                                 $img = Image::newFromTitle( $nt );
1587                                 if( $img->exists() ) {
1588                                         // Force a blue link if the file exists; may be a remote
1589                                         // upload on the shared repository, and we want to see its
1590                                         // auto-generated page.
1591                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1592                                         continue;
1593                                 }
1594                         }
1595                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1596                 }
1597                 wfProfileOut( $fname );
1598                 return $s;
1599         }
1600
1601         /**
1602          * Make a link placeholder. The text returned can be later resolved to a real link with
1603          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1604          * parsing of interwiki links, and secondly to allow all extistence checks and
1605          * article length checks (for stub links) to be bundled into a single query.
1606          *
1607          */
1608         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1609                 if ( ! is_object($nt) ) {
1610                         # Fail gracefully
1611                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1612                 } else {
1613                         # Separate the link trail from the rest of the link
1614                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1615
1616                         if ( $nt->isExternal() ) {
1617                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1618                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1619                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1620                         } else {
1621                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1622                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1623                                 $this->mLinkHolders['queries'][] = $query;
1624                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1625                                 $this->mLinkHolders['titles'][] = $nt;
1626
1627                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1628                         }
1629                 }
1630                 return $retVal;
1631         }
1632
1633         /**
1634          * Render a forced-blue link inline; protect against double expansion of
1635          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1636          * Since this little disaster has to split off the trail text to avoid
1637          * breaking URLs in the following text without breaking trails on the
1638          * wiki links, it's been made into a horrible function.
1639          *
1640          * @param Title $nt
1641          * @param string $text
1642          * @param string $query
1643          * @param string $trail
1644          * @param string $prefix
1645          * @return string HTML-wikitext mix oh yuck
1646          */
1647         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1648                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1649                 $sk =& $this->mOptions->getSkin();
1650                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1651                 return $this->armorLinks( $link ) . $trail;
1652         }
1653
1654         /**
1655          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1656          * going to go through further parsing steps before inline URL expansion.
1657          *
1658          * In particular this is important when using action=render, which causes
1659          * full URLs to be included.
1660          *
1661          * Oh man I hate our multi-layer parser!
1662          *
1663          * @param string more-or-less HTML
1664          * @return string less-or-more HTML with NOPARSE bits
1665          */
1666         function armorLinks( $text ) {
1667                 return preg_replace( "/\b(" . wfUrlProtocols() . ')/',
1668                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1669         }
1670
1671         /**
1672          * Return true if subpage links should be expanded on this page.
1673          * @return bool
1674          */
1675         function areSubpagesAllowed() {
1676                 # Some namespaces don't allow subpages
1677                 global $wgNamespacesWithSubpages;
1678                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1679         }
1680
1681         /**
1682          * Handle link to subpage if necessary
1683          * @param string $target the source of the link
1684          * @param string &$text the link text, modified as necessary
1685          * @return string the full name of the link
1686          * @access private
1687          */
1688         function maybeDoSubpageLink($target, &$text) {
1689                 # Valid link forms:
1690                 # Foobar -- normal
1691                 # :Foobar -- override special treatment of prefix (images, language links)
1692                 # /Foobar -- convert to CurrentPage/Foobar
1693                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1694                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1695                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1696
1697                 $fname = 'Parser::maybeDoSubpageLink';
1698                 wfProfileIn( $fname );
1699                 $ret = $target; # default return value is no change
1700
1701                 # Some namespaces don't allow subpages,
1702                 # so only perform processing if subpages are allowed
1703                 if( $this->areSubpagesAllowed() ) {
1704                         # Look at the first character
1705                         if( $target != '' && $target{0} == '/' ) {
1706                                 # / at end means we don't want the slash to be shown
1707                                 if( substr( $target, -1, 1 ) == '/' ) {
1708                                         $target = substr( $target, 1, -1 );
1709                                         $noslash = $target;
1710                                 } else {
1711                                         $noslash = substr( $target, 1 );
1712                                 }
1713
1714                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash);
1715                                 if( '' === $text ) {
1716                                         $text = $target;
1717                                 } # this might be changed for ugliness reasons
1718                         } else {
1719                                 # check for .. subpage backlinks
1720                                 $dotdotcount = 0;
1721                                 $nodotdot = $target;
1722                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1723                                         ++$dotdotcount;
1724                                         $nodotdot = substr( $nodotdot, 3 );
1725                                 }
1726                                 if($dotdotcount > 0) {
1727                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1728                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1729                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1730                                                 # / at the end means don't show full path
1731                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1732                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1733                                                         if( '' === $text ) {
1734                                                                 $text = $nodotdot;
1735                                                         }
1736                                                 }
1737                                                 $nodotdot = trim( $nodotdot );
1738                                                 if( $nodotdot != '' ) {
1739                                                         $ret .= '/' . $nodotdot;
1740                                                 }
1741                                         }
1742                                 }
1743                         }
1744                 }
1745
1746                 wfProfileOut( $fname );
1747                 return $ret;
1748         }
1749
1750         /**#@+
1751          * Used by doBlockLevels()
1752          * @access private
1753          */
1754         /* private */ function closeParagraph() {
1755                 $result = '';
1756                 if ( '' != $this->mLastSection ) {
1757                         $result = '</' . $this->mLastSection  . ">\n";
1758                 }
1759                 $this->mInPre = false;
1760                 $this->mLastSection = '';
1761                 return $result;
1762         }
1763         # getCommon() returns the length of the longest common substring
1764         # of both arguments, starting at the beginning of both.
1765         #
1766         /* private */ function getCommon( $st1, $st2 ) {
1767                 $fl = strlen( $st1 );
1768                 $shorter = strlen( $st2 );
1769                 if ( $fl < $shorter ) { $shorter = $fl; }
1770
1771                 for ( $i = 0; $i < $shorter; ++$i ) {
1772                         if ( $st1{$i} != $st2{$i} ) { break; }
1773                 }
1774                 return $i;
1775         }
1776         # These next three functions open, continue, and close the list
1777         # element appropriate to the prefix character passed into them.
1778         #
1779         /* private */ function openList( $char ) {
1780                 $result = $this->closeParagraph();
1781
1782                 if ( '*' == $char ) { $result .= '<ul><li>'; }
1783                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
1784                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
1785                 else if ( ';' == $char ) {
1786                         $result .= '<dl><dt>';
1787                         $this->mDTopen = true;
1788                 }
1789                 else { $result = '<!-- ERR 1 -->'; }
1790
1791                 return $result;
1792         }
1793
1794         /* private */ function nextItem( $char ) {
1795                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
1796                 else if ( ':' == $char || ';' == $char ) {
1797                         $close = '</dd>';
1798                         if ( $this->mDTopen ) { $close = '</dt>'; }
1799                         if ( ';' == $char ) {
1800                                 $this->mDTopen = true;
1801                                 return $close . '<dt>';
1802                         } else {
1803                                 $this->mDTopen = false;
1804                                 return $close . '<dd>';
1805                         }
1806                 }
1807                 return '<!-- ERR 2 -->';
1808         }
1809
1810         /* private */ function closeList( $char ) {
1811                 if ( '*' == $char ) { $text = '</li></ul>'; }
1812                 else if ( '#' == $char ) { $text = '</li></ol>'; }
1813                 else if ( ':' == $char ) {
1814                         if ( $this->mDTopen ) {
1815                                 $this->mDTopen = false;
1816                                 $text = '</dt></dl>';
1817                         } else {
1818                                 $text = '</dd></dl>';
1819                         }
1820                 }
1821                 else {  return '<!-- ERR 3 -->'; }
1822                 return $text."\n";
1823         }
1824         /**#@-*/
1825
1826         /**
1827          * Make lists from lines starting with ':', '*', '#', etc.
1828          *
1829          * @access private
1830          * @return string the lists rendered as HTML
1831          */
1832         function doBlockLevels( $text, $linestart ) {
1833                 $fname = 'Parser::doBlockLevels';
1834                 wfProfileIn( $fname );
1835
1836                 # Parsing through the text line by line.  The main thing
1837                 # happening here is handling of block-level elements p, pre,
1838                 # and making lists from lines starting with * # : etc.
1839                 #
1840                 $textLines = explode( "\n", $text );
1841
1842                 $lastPrefix = $output = '';
1843                 $this->mDTopen = $inBlockElem = false;
1844                 $prefixLength = 0;
1845                 $paragraphStack = false;
1846
1847                 if ( !$linestart ) {
1848                         $output .= array_shift( $textLines );
1849                 }
1850                 foreach ( $textLines as $oLine ) {
1851                         $lastPrefixLength = strlen( $lastPrefix );
1852                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
1853                         $preOpenMatch = preg_match('/<pre/i', $oLine );
1854                         if ( !$this->mInPre ) {
1855                                 # Multiple prefixes may abut each other for nested lists.
1856                                 $prefixLength = strspn( $oLine, '*#:;' );
1857                                 $pref = substr( $oLine, 0, $prefixLength );
1858
1859                                 # eh?
1860                                 $pref2 = str_replace( ';', ':', $pref );
1861                                 $t = substr( $oLine, $prefixLength );
1862                                 $this->mInPre = !empty($preOpenMatch);
1863                         } else {
1864                                 # Don't interpret any other prefixes in preformatted text
1865                                 $prefixLength = 0;
1866                                 $pref = $pref2 = '';
1867                                 $t = $oLine;
1868                         }
1869
1870                         # List generation
1871                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
1872                                 # Same as the last item, so no need to deal with nesting or opening stuff
1873                                 $output .= $this->nextItem( substr( $pref, -1 ) );
1874                                 $paragraphStack = false;
1875
1876                                 if ( substr( $pref, -1 ) == ';') {
1877                                         # The one nasty exception: definition lists work like this:
1878                                         # ; title : definition text
1879                                         # So we check for : in the remainder text to split up the
1880                                         # title and definition, without b0rking links.
1881                                         $term = $t2 = '';
1882                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
1883                                                 $t = $t2;
1884                                                 $output .= $term . $this->nextItem( ':' );
1885                                         }
1886                                 }
1887                         } elseif( $prefixLength || $lastPrefixLength ) {
1888                                 # Either open or close a level...
1889                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
1890                                 $paragraphStack = false;
1891
1892                                 while( $commonPrefixLength < $lastPrefixLength ) {
1893                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
1894                                         --$lastPrefixLength;
1895                                 }
1896                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
1897                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
1898                                 }
1899                                 while ( $prefixLength > $commonPrefixLength ) {
1900                                         $char = substr( $pref, $commonPrefixLength, 1 );
1901                                         $output .= $this->openList( $char );
1902
1903                                         if ( ';' == $char ) {
1904                                                 # FIXME: This is dupe of code above
1905                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
1906                                                         $t = $t2;
1907                                                         $output .= $term . $this->nextItem( ':' );
1908                                                 }
1909                                         }
1910                                         ++$commonPrefixLength;
1911                                 }
1912                                 $lastPrefix = $pref2;
1913                         }
1914                         if( 0 == $prefixLength ) {
1915                                 wfProfileIn( "$fname-paragraph" );
1916                                 # No prefix (not in list)--go to paragraph mode
1917                                 // XXX: use a stack for nestable elements like span, table and div
1918                                 $openmatch = preg_match('/(<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
1919                                 $closematch = preg_match(
1920                                         '/(<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
1921                                         '<td|<th|<div|<\\/div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul)/iS', $t );
1922                                 if ( $openmatch or $closematch ) {
1923                                         $paragraphStack = false;
1924                                         $output .= $this->closeParagraph();
1925                                         if ( $preOpenMatch and !$preCloseMatch ) {
1926                                                 $this->mInPre = true;
1927                                         }
1928                                         if ( $closematch ) {
1929                                                 $inBlockElem = false;
1930                                         } else {
1931                                                 $inBlockElem = true;
1932                                         }
1933                                 } else if ( !$inBlockElem && !$this->mInPre ) {
1934                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
1935                                                 // pre
1936                                                 if ($this->mLastSection != 'pre') {
1937                                                         $paragraphStack = false;
1938                                                         $output .= $this->closeParagraph().'<pre>';
1939                                                         $this->mLastSection = 'pre';
1940                                                 }
1941                                                 $t = substr( $t, 1 );
1942                                         } else {
1943                                                 // paragraph
1944                                                 if ( '' == trim($t) ) {
1945                                                         if ( $paragraphStack ) {
1946                                                                 $output .= $paragraphStack.'<br />';
1947                                                                 $paragraphStack = false;
1948                                                                 $this->mLastSection = 'p';
1949                                                         } else {
1950                                                                 if ($this->mLastSection != 'p' ) {
1951                                                                         $output .= $this->closeParagraph();
1952                                                                         $this->mLastSection = '';
1953                                                                         $paragraphStack = '<p>';
1954                                                                 } else {
1955                                                                         $paragraphStack = '</p><p>';
1956                                                                 }
1957                                                         }
1958                                                 } else {
1959                                                         if ( $paragraphStack ) {
1960                                                                 $output .= $paragraphStack;
1961                                                                 $paragraphStack = false;
1962                                                                 $this->mLastSection = 'p';
1963                                                         } else if ($this->mLastSection != 'p') {
1964                                                                 $output .= $this->closeParagraph().'<p>';
1965                                                                 $this->mLastSection = 'p';
1966                                                         }
1967                                                 }
1968                                         }
1969                                 }
1970                                 wfProfileOut( "$fname-paragraph" );
1971                         }
1972                         // somewhere above we forget to get out of pre block (bug 785)
1973                         if($preCloseMatch && $this->mInPre) {
1974                                 $this->mInPre = false;
1975                         }
1976                         if ($paragraphStack === false) {
1977                                 $output .= $t."\n";
1978                         }
1979                 }
1980                 while ( $prefixLength ) {
1981                         $output .= $this->closeList( $pref2{$prefixLength-1} );
1982                         --$prefixLength;
1983                 }
1984                 if ( '' != $this->mLastSection ) {
1985                         $output .= '</' . $this->mLastSection . '>';
1986                         $this->mLastSection = '';
1987                 }
1988
1989                 wfProfileOut( $fname );
1990                 return $output;
1991         }
1992
1993         /**
1994          * Split up a string on ':', ignoring any occurences inside
1995          * <a>..</a> or <span>...</span>
1996          * @param string $str the string to split
1997          * @param string &$before set to everything before the ':'
1998          * @param string &$after set to everything after the ':'
1999          * return string the position of the ':', or false if none found
2000          */
2001         function findColonNoLinks($str, &$before, &$after) {
2002                 # I wonder if we should make this count all tags, not just <a>
2003                 # and <span>. That would prevent us from matching a ':' that
2004                 # comes in the middle of italics other such formatting....
2005                 # -- Wil
2006                 $fname = 'Parser::findColonNoLinks';
2007                 wfProfileIn( $fname );
2008                 $pos = 0;
2009                 do {
2010                         $colon = strpos($str, ':', $pos);
2011
2012                         if ($colon !== false) {
2013                                 $before = substr($str, 0, $colon);
2014                                 $after = substr($str, $colon + 1);
2015
2016                                 # Skip any ':' within <a> or <span> pairs
2017                                 $a = substr_count($before, '<a');
2018                                 $s = substr_count($before, '<span');
2019                                 $ca = substr_count($before, '</a>');
2020                                 $cs = substr_count($before, '</span>');
2021
2022                                 if ($a <= $ca and $s <= $cs) {
2023                                         # Tags are balanced before ':'; ok
2024                                         break;
2025                                 }
2026                                 $pos = $colon + 1;
2027                         }
2028                 } while ($colon !== false);
2029                 wfProfileOut( $fname );
2030                 return $colon;
2031         }
2032
2033         /**
2034          * Return value of a magic variable (like PAGENAME)
2035          *
2036          * @access private
2037          */
2038         function getVariableValue( $index ) {
2039                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2040
2041                 /**
2042                  * Some of these require message or data lookups and can be
2043                  * expensive to check many times.
2044                  */
2045                 static $varCache = array();
2046                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$varCache ) ) )
2047                         if ( isset( $varCache[$index] ) )
2048                                 return $varCache[$index];
2049
2050                 $ts = time();
2051                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2052
2053                 switch ( $index ) {
2054                         case MAG_CURRENTMONTH:
2055                                 return $varCache[$index] = $wgContLang->formatNum( date( 'm', $ts ) );
2056                         case MAG_CURRENTMONTHNAME:
2057                                 return $varCache[$index] = $wgContLang->getMonthName( date( 'n', $ts ) );
2058                         case MAG_CURRENTMONTHNAMEGEN:
2059                                 return $varCache[$index] = $wgContLang->getMonthNameGen( date( 'n', $ts ) );
2060                         case MAG_CURRENTMONTHABBREV:
2061                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( date( 'n', $ts ) );
2062                         case MAG_CURRENTDAY:
2063                                 return $varCache[$index] = $wgContLang->formatNum( date( 'j', $ts ) );
2064                         case MAG_CURRENTDAY2:
2065                                 return $varCache[$index] = $wgContLang->formatNum( date( 'd', $ts ) );
2066                         case MAG_PAGENAME:
2067                                 return $this->mTitle->getText();
2068                         case MAG_PAGENAMEE:
2069                                 return $this->mTitle->getPartialURL();
2070                         case MAG_FULLPAGENAME:
2071                                 return $this->mTitle->getPrefixedText();
2072                         case MAG_FULLPAGENAMEE:
2073                                 return $this->mTitle->getPrefixedURL();
2074                         case MAG_SUBPAGENAME:
2075                                 return $this->mTitle->getSubpageText();
2076                         case MAG_REVISIONID:
2077                                 return $this->mRevisionId;
2078                         case MAG_NAMESPACE:
2079                                 return $wgContLang->getNsText( $this->mTitle->getNamespace() );
2080                         case MAG_NAMESPACEE:
2081                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2082                         case MAG_CURRENTDAYNAME:
2083                                 return $varCache[$index] = $wgContLang->getWeekdayName( date( 'w', $ts ) + 1 );
2084                         case MAG_CURRENTYEAR:
2085                                 return $varCache[$index] = $wgContLang->formatNum( date( 'Y', $ts ), true );
2086                         case MAG_CURRENTTIME:
2087                                 return $varCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2088                         case MAG_CURRENTWEEK:
2089                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2090                                 // int to remove the padding
2091                                 return $varCache[$index] = $wgContLang->formatNum( (int)date( 'W', $ts ) );
2092                         case MAG_CURRENTDOW:
2093                                 return $varCache[$index] = $wgContLang->formatNum( date( 'w', $ts ) );
2094                         case MAG_NUMBEROFARTICLES:
2095                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfArticles() );
2096                         case MAG_NUMBEROFFILES:
2097                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfFiles() );
2098                         case MAG_SITENAME:
2099                                 return $wgSitename;
2100                         case MAG_SERVER:
2101                                 return $wgServer;
2102                         case MAG_SERVERNAME:
2103                                 return $wgServerName;
2104                         case MAG_SCRIPTPATH:
2105                                 return $wgScriptPath;
2106                         default:
2107                                 $ret = null;
2108                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$varCache, &$index, &$ret ) ) )
2109                                         return $ret;
2110                                 else
2111                                         return null;
2112                 }
2113         }
2114
2115         /**
2116          * initialise the magic variables (like CURRENTMONTHNAME)
2117          *
2118          * @access private
2119          */
2120         function initialiseVariables() {
2121                 $fname = 'Parser::initialiseVariables';
2122                 wfProfileIn( $fname );
2123                 global $wgVariableIDs;
2124                 $this->mVariables = array();
2125                 foreach ( $wgVariableIDs as $id ) {
2126                         $mw =& MagicWord::get( $id );
2127                         $mw->addToArray( $this->mVariables, $id );
2128                 }
2129                 wfProfileOut( $fname );
2130         }
2131
2132         /**
2133          * parse any parentheses in format ((title|part|part))
2134          * and call callbacks to get a replacement text for any found piece
2135          *
2136          * @param string $text The text to parse
2137          * @param array $callbacks rules in form:
2138          *     '{' => array(                            # opening parentheses
2139          *                                      'end' => '}',   # closing parentheses
2140          *                                      'cb' => array(2 => callback,    # replacement callback to call if {{..}} is found
2141          *                                                                4 => callback         # replacement callback to call if {{{{..}}}} is found
2142          *                                                                )
2143          *                                      )
2144          * @access private
2145          */
2146         function replace_callback ($text, $callbacks) {
2147                 $openingBraceStack = array();   # this array will hold a stack of parentheses which are not closed yet
2148                 $lastOpeningBrace = -1;                 # last not closed parentheses
2149
2150                 for ($i = 0; $i < strlen($text); $i++) {
2151                         # check for any opening brace
2152                         $rule = null;
2153                         $nextPos = -1;
2154                         foreach ($callbacks as $key => $value) {
2155                                 $pos = strpos ($text, $key, $i);
2156                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)) {
2157                                         $rule = $value;
2158                                         $nextPos = $pos;
2159                                 }
2160                         }
2161
2162                         if ($lastOpeningBrace >= 0) {
2163                                 $pos = strpos ($text, $openingBraceStack[$lastOpeningBrace]['braceEnd'], $i);
2164
2165                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)){
2166                                         $rule = null;
2167                                         $nextPos = $pos;
2168                                 }
2169
2170                                 $pos = strpos ($text, '|', $i);
2171
2172                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)){
2173                                         $rule = null;
2174                                         $nextPos = $pos;
2175                                 }
2176                         }
2177
2178                         if ($nextPos == -1)
2179                                 break;
2180
2181                         $i = $nextPos;
2182
2183                         # found openning brace, lets add it to parentheses stack
2184                         if (null != $rule) {
2185                                 $piece = array('brace' => $text[$i],
2186                                                            'braceEnd' => $rule['end'],
2187                                                            'count' => 1,
2188                                                            'title' => '',
2189                                                            'parts' => null);
2190
2191                                 # count openning brace characters
2192                                 while ($i+1 < strlen($text) && $text[$i+1] == $piece['brace']) {
2193                                         $piece['count']++;
2194                                         $i++;
2195                                 }
2196
2197                                 $piece['startAt'] = $i+1;
2198                                 $piece['partStart'] = $i+1;
2199
2200                                 # we need to add to stack only if openning brace count is enough for any given rule
2201                                 foreach ($rule['cb'] as $cnt => $fn) {
2202                                         if ($piece['count'] >= $cnt) {
2203                                                 $lastOpeningBrace ++;
2204                                                 $openingBraceStack[$lastOpeningBrace] = $piece;
2205                                                 break;
2206                                         }
2207                                 }
2208
2209                                 continue;
2210                         }
2211                         else if ($lastOpeningBrace >= 0) {
2212                                 # first check if it is a closing brace
2213                                 if ($openingBraceStack[$lastOpeningBrace]['braceEnd'] == $text[$i]) {
2214                                         # lets check if it is enough characters for closing brace
2215                                         $count = 1;
2216                                         while ($i+$count < strlen($text) && $text[$i+$count] == $text[$i])
2217                                                 $count++;
2218
2219                                         # if there are more closing parentheses than opening ones, we parse less
2220                                         if ($openingBraceStack[$lastOpeningBrace]['count'] < $count)
2221                                                 $count = $openingBraceStack[$lastOpeningBrace]['count'];
2222
2223                                         # check for maximum matching characters (if there are 5 closing characters, we will probably need only 3 - depending on the rules)
2224                                         $matchingCount = 0;
2225                                         $matchingCallback = null;
2226                                         foreach ($callbacks[$openingBraceStack[$lastOpeningBrace]['brace']]['cb'] as $cnt => $fn) {
2227                                                 if ($count >= $cnt && $matchingCount < $cnt) {
2228                                                         $matchingCount = $cnt;
2229                                                         $matchingCallback = $fn;
2230                                                 }
2231                                         }
2232
2233                                         if ($matchingCount == 0) {
2234                                                 $i += $count - 1;
2235                                                 continue;
2236                                         }
2237
2238                                         # lets set a title or last part (if '|' was found)
2239                                         if (null === $openingBraceStack[$lastOpeningBrace]['parts'])
2240                                                 $openingBraceStack[$lastOpeningBrace]['title'] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2241                                         else
2242                                                 $openingBraceStack[$lastOpeningBrace]['parts'][] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2243
2244                                         $pieceStart = $openingBraceStack[$lastOpeningBrace]['startAt'] - $matchingCount;
2245                                         $pieceEnd = $i + $matchingCount;
2246
2247                                         if( is_callable( $matchingCallback ) ) {
2248                                                 $cbArgs = array (
2249                                                                                  'text' => substr($text, $pieceStart, $pieceEnd - $pieceStart),
2250                                                                                  'title' => trim($openingBraceStack[$lastOpeningBrace]['title']),
2251                                                                                  'parts' => $openingBraceStack[$lastOpeningBrace]['parts'],
2252                                                                                  'lineStart' => (($pieceStart > 0) && ($text[$pieceStart-1] == '\n')),
2253                                                                                  );
2254                                                 # finally we can call a user callback and replace piece of text
2255                                                 $replaceWith = call_user_func( $matchingCallback, $cbArgs );
2256                                                 $text = substr($text, 0, $pieceStart) . $replaceWith . substr($text, $pieceEnd);
2257                                                 $i = $pieceStart + strlen($replaceWith) - 1;
2258                                         }
2259                                         else {
2260                                                 # null value for callback means that parentheses should be parsed, but not replaced
2261                                                 $i += $matchingCount - 1;
2262                                         }
2263
2264                                         # reset last openning parentheses, but keep it in case there are unused characters
2265                                         $piece = array('brace' => $openingBraceStack[$lastOpeningBrace]['brace'],
2266                                                                    'braceEnd' => $openingBraceStack[$lastOpeningBrace]['braceEnd'],
2267                                                                    'count' => $openingBraceStack[$lastOpeningBrace]['count'],
2268                                                                    'title' => '',
2269                                                                    'parts' => null,
2270                                                                    'startAt' => $openingBraceStack[$lastOpeningBrace]['startAt']);
2271                                         $openingBraceStack[$lastOpeningBrace--] = null;
2272
2273                                         if ($matchingCount < $piece['count']) {
2274                                                 $piece['count'] -= $matchingCount;
2275                                                 $piece['startAt'] -= $matchingCount;
2276                                                 $piece['partStart'] = $piece['startAt'];
2277                                                 # do we still qualify for any callback with remaining count?
2278                                                 foreach ($callbacks[$piece['brace']]['cb'] as $cnt => $fn) {
2279                                                         if ($piece['count'] >= $cnt) {
2280                                                                 $lastOpeningBrace ++;
2281                                                                 $openingBraceStack[$lastOpeningBrace] = $piece;
2282                                                                 break;
2283                                                         }
2284                                                 }
2285                                         }
2286                                         continue;
2287                                 }
2288
2289                                 # lets set a title if it is a first separator, or next part otherwise
2290                                 if ($text[$i] == '|') {
2291                                         if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2292                                                 $openingBraceStack[$lastOpeningBrace]['title'] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2293                                                 $openingBraceStack[$lastOpeningBrace]['parts'] = array();
2294                                         }
2295                                         else
2296                                                 $openingBraceStack[$lastOpeningBrace]['parts'][] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2297
2298                                         $openingBraceStack[$lastOpeningBrace]['partStart'] = $i + 1;
2299                                 }
2300                         }
2301                 }
2302
2303                 return $text;
2304         }
2305
2306         /**
2307          * Replace magic variables, templates, and template arguments
2308          * with the appropriate text. Templates are substituted recursively,
2309          * taking care to avoid infinite loops.
2310          *
2311          * Note that the substitution depends on value of $mOutputType:
2312          *  OT_WIKI: only {{subst:}} templates
2313          *  OT_MSG: only magic variables
2314          *  OT_HTML: all templates and magic variables
2315          *
2316          * @param string $tex The text to transform
2317          * @param array $args Key-value pairs representing template parameters to substitute
2318          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2319          * @access private
2320          */
2321         function replaceVariables( $text, $args = array(), $argsOnly = false ) {
2322                 # Prevent too big inclusions
2323                 if( strlen( $text ) > MAX_INCLUDE_SIZE ) {
2324                         return $text;
2325                 }
2326
2327                 $fname = 'Parser::replaceVariables';
2328                 wfProfileIn( $fname );
2329
2330                 # This function is called recursively. To keep track of arguments we need a stack:
2331                 array_push( $this->mArgStack, $args );
2332
2333                 $braceCallbacks = array();
2334                 if ( !$argsOnly ) {
2335                         $braceCallbacks[2] = array( &$this, 'braceSubstitution' );
2336                 }
2337                 if ( $this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI ) {
2338                         $braceCallbacks[3] = array( &$this, 'argSubstitution' );
2339                 }
2340                 $callbacks = array();
2341                 $callbacks['{'] = array('end' => '}', 'cb' => $braceCallbacks);
2342                 $callbacks['['] = array('end' => ']', 'cb' => array(2=>null));
2343                 $text = $this->replace_callback ($text, $callbacks);
2344
2345                 array_pop( $this->mArgStack );
2346
2347                 wfProfileOut( $fname );
2348                 return $text;
2349         }
2350
2351         /**
2352          * Replace magic variables
2353          * @access private
2354          */
2355         function variableSubstitution( $matches ) {
2356                 $fname = 'Parser::variableSubstitution';
2357                 $varname = $matches[1];
2358                 wfProfileIn( $fname );
2359                 if ( !$this->mVariables ) {
2360                         $this->initialiseVariables();
2361                 }
2362                 $skip = false;
2363                 if ( $this->mOutputType == OT_WIKI ) {
2364                         # Do only magic variables prefixed by SUBST
2365                         $mwSubst =& MagicWord::get( MAG_SUBST );
2366                         if (!$mwSubst->matchStartAndRemove( $varname ))
2367                                 $skip = true;
2368                         # Note that if we don't substitute the variable below,
2369                         # we don't remove the {{subst:}} magic word, in case
2370                         # it is a template rather than a magic variable.
2371                 }
2372                 if ( !$skip && array_key_exists( $varname, $this->mVariables ) ) {
2373                         $id = $this->mVariables[$varname];
2374                         $text = $this->getVariableValue( $id );
2375                         $this->mOutput->mContainsOldMagic = true;
2376                 } else {
2377                         $text = $matches[0];
2378                 }
2379                 wfProfileOut( $fname );
2380                 return $text;
2381         }
2382
2383         # Split template arguments
2384         function getTemplateArgs( $argsString ) {
2385                 if ( $argsString === '' ) {
2386                         return array();
2387                 }
2388
2389                 $args = explode( '|', substr( $argsString, 1 ) );
2390
2391                 # If any of the arguments contains a '[[' but no ']]', it needs to be
2392                 # merged with the next arg because the '|' character between belongs
2393                 # to the link syntax and not the template parameter syntax.
2394                 $argc = count($args);
2395
2396                 for ( $i = 0; $i < $argc-1; $i++ ) {
2397                         if ( substr_count ( $args[$i], '[[' ) != substr_count ( $args[$i], ']]' ) ) {
2398                                 $args[$i] .= '|'.$args[$i+1];
2399                                 array_splice($args, $i+1, 1);
2400                                 $i--;
2401                                 $argc--;
2402                         }
2403                 }
2404
2405                 return $args;
2406         }
2407
2408         /**
2409          * Return the text of a template, after recursively
2410          * replacing any variables or templates within the template.
2411          *
2412          * @param array $piece The parts of the template
2413          *  $piece['text']: matched text
2414          *  $piece['title']: the title, i.e. the part before the |
2415          *  $piece['parts']: the parameter array
2416          * @return string the text of the template
2417          * @access private
2418          */
2419         function braceSubstitution( $piece ) {
2420                 global $wgContLang;
2421                 $fname = 'Parser::braceSubstitution';
2422                 wfProfileIn( $fname );
2423
2424                 # Flags
2425                 $found = false;             # $text has been filled
2426                 $nowiki = false;            # wiki markup in $text should be escaped
2427                 $noparse = false;           # Unsafe HTML tags should not be stripped, etc.
2428                 $noargs = false;            # Don't replace triple-brace arguments in $text
2429                 $replaceHeadings = false;   # Make the edit section links go to the template not the article
2430                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2431                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2432
2433                 # Title object, where $text came from
2434                 $title = NULL;
2435
2436                 $linestart = '';
2437
2438                 # $part1 is the bit before the first |, and must contain only title characters
2439                 # $args is a list of arguments, starting from index 0, not including $part1
2440
2441                 $part1 = $piece['title'];
2442                 # If the third subpattern matched anything, it will start with |
2443
2444                 if (null == $piece['parts']) {
2445                         $replaceWith = $this->variableSubstitution (array ($piece['text'], $piece['title']));
2446                         if ($replaceWith != $piece['text']) {
2447                                 $text = $replaceWith;
2448                                 $found = true;
2449                                 $noparse = true;
2450                                 $noargs = true;
2451                         }
2452                 }
2453
2454                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2455                 $argc = count( $args );
2456
2457                 # SUBST
2458                 if ( !$found ) {
2459                         $mwSubst =& MagicWord::get( MAG_SUBST );
2460                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor ($this->mOutputType == OT_WIKI) ) {
2461                                 # One of two possibilities is true:
2462                                 # 1) Found SUBST but not in the PST phase
2463                                 # 2) Didn't find SUBST and in the PST phase
2464                                 # In either case, return without further processing
2465                                 $text = $piece['text'];
2466                                 $found = true;
2467                                 $noparse = true;
2468                                 $noargs = true;
2469                         }
2470                 }
2471
2472                 # MSG, MSGNW, INT and RAW
2473                 if ( !$found ) {
2474                         # Check for MSGNW:
2475                         $mwMsgnw =& MagicWord::get( MAG_MSGNW );
2476                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2477                                 $nowiki = true;
2478                         } else {
2479                                 # Remove obsolete MSG:
2480                                 $mwMsg =& MagicWord::get( MAG_MSG );
2481                                 $mwMsg->matchStartAndRemove( $part1 );
2482                         }
2483
2484                         # Check for RAW:
2485                         $mwRaw =& MagicWord::get( MAG_RAW );
2486                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2487                                 $forceRawInterwiki = true;
2488                         }
2489
2490                         # Check if it is an internal message
2491                         $mwInt =& MagicWord::get( MAG_INT );
2492                         if ( $mwInt->matchStartAndRemove( $part1 ) ) {
2493                                 if ( $this->incrementIncludeCount( 'int:'.$part1 ) ) {
2494                                         $text = $linestart . wfMsgReal( $part1, $args, true );
2495                                         $found = true;
2496                                 }
2497                         }
2498                 }
2499
2500                 # NS
2501                 if ( !$found ) {
2502                         # Check for NS: (namespace expansion)
2503                         $mwNs = MagicWord::get( MAG_NS );
2504                         if ( $mwNs->matchStartAndRemove( $part1 ) ) {
2505                                 if ( intval( $part1 ) || $part1 == "0" ) {
2506                                         $text = $linestart . $wgContLang->getNsText( intval( $part1 ) );
2507                                         $found = true;
2508                                 } else {
2509                                         $index = Namespace::getCanonicalIndex( strtolower( $part1 ) );
2510                                         if ( !is_null( $index ) ) {
2511                                                 $text = $linestart . $wgContLang->getNsText( $index );
2512                                                 $found = true;
2513                                         }
2514                                 }
2515                         }
2516                 }
2517
2518                 # LCFIRST, UCFIRST, LC and UC
2519                 if ( !$found ) {
2520                         $lcfirst =& MagicWord::get( MAG_LCFIRST );
2521                         $ucfirst =& MagicWord::get( MAG_UCFIRST );
2522                         $lc =& MagicWord::get( MAG_LC );
2523                         $uc =& MagicWord::get( MAG_UC );
2524                         if ( $lcfirst->matchStartAndRemove( $part1 ) ) {
2525                                 $text = $linestart . $wgContLang->lcfirst( $part1 );
2526                                 $found = true;
2527                         } else if ( $ucfirst->matchStartAndRemove( $part1 ) ) {
2528                                 $text = $linestart . $wgContLang->ucfirst( $part1 );
2529                                 $found = true;
2530                         } else if ( $lc->matchStartAndRemove( $part1 ) ) {
2531                                 $text = $linestart . $wgContLang->lc( $part1 );
2532                                 $found = true;
2533                         } else if ( $uc->matchStartAndRemove( $part1 ) ) {
2534                                  $text = $linestart . $wgContLang->uc( $part1 );
2535                                  $found = true;
2536                         }
2537                 }
2538
2539                 # LOCALURL and FULLURL
2540                 if ( !$found ) {
2541                         $mwLocal =& MagicWord::get( MAG_LOCALURL );
2542                         $mwLocalE =& MagicWord::get( MAG_LOCALURLE );
2543                         $mwFull =& MagicWord::get( MAG_FULLURL );
2544                         $mwFullE =& MagicWord::get( MAG_FULLURLE );
2545
2546
2547                         if ( $mwLocal->matchStartAndRemove( $part1 ) ) {
2548                                 $func = 'getLocalURL';
2549                         } elseif ( $mwLocalE->matchStartAndRemove( $part1 ) ) {
2550                                 $func = 'escapeLocalURL';
2551                         } elseif ( $mwFull->matchStartAndRemove( $part1 ) ) {
2552                                 $func = 'getFullURL';
2553                         } elseif ( $mwFullE->matchStartAndRemove( $part1 ) ) {
2554                                 $func = 'escapeFullURL';
2555                         } else {
2556                                 $func = false;
2557                         }
2558
2559                         if ( $func !== false ) {
2560                                 $title = Title::newFromText( $part1 );
2561                                 if ( !is_null( $title ) ) {
2562                                         if ( $argc > 0 ) {
2563                                                 $text = $linestart . $title->$func( $args[0] );
2564                                         } else {
2565                                                 $text = $linestart . $title->$func();
2566                                         }
2567                                         $found = true;
2568                                 }
2569                         }
2570                 }
2571
2572                 # GRAMMAR
2573                 if ( !$found && $argc == 1 ) {
2574                         $mwGrammar =& MagicWord::get( MAG_GRAMMAR );
2575                         if ( $mwGrammar->matchStartAndRemove( $part1 ) ) {
2576                                 $text = $linestart . $wgContLang->convertGrammar( $args[0], $part1 );
2577                                 $found = true;
2578                         }
2579                 }
2580
2581                 # PLURAL
2582                 if ( !$found && $argc >= 2 ) {
2583                         $mwPluralForm =& MagicWord::get( MAG_PLURAL );
2584                         if ( $mwPluralForm->matchStartAndRemove( $part1 ) ) {
2585                                 if ($argc==2) {$args[2]=$args[1];}
2586                                 $text = $linestart . $wgContLang->convertPlural( $part1, $args[0], $args[1], $args[2]);
2587                                 $found = true;
2588                         }
2589                 }
2590
2591                 # Template table test
2592
2593                 # Did we encounter this template already? If yes, it is in the cache
2594                 # and we need to check for loops.
2595                 if ( !$found && isset( $this->mTemplates[$piece['title']] ) ) {
2596                         $found = true;
2597
2598                         # Infinite loop test
2599                         if ( isset( $this->mTemplatePath[$part1] ) ) {
2600                                 $noparse = true;
2601                                 $noargs = true;
2602                                 $found = true;
2603                                 $text = $linestart .
2604                                         '{{' . $part1 . '}}' .
2605                                         '<!-- WARNING: template loop detected -->';
2606                                 wfDebug( "$fname: template loop broken at '$part1'\n" );
2607                         } else {
2608                                 # set $text to cached message.
2609                                 $text = $linestart . $this->mTemplates[$piece['title']];
2610                         }
2611                 }
2612
2613                 # Load from database
2614                 $lastPathLevel = $this->mTemplatePath;
2615                 if ( !$found ) {
2616                         $ns = NS_TEMPLATE;
2617                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage='' );
2618                         if ($subpage !== '') {
2619                                 $ns = $this->mTitle->getNamespace();
2620                         }
2621                         $title = Title::newFromText( $part1, $ns );
2622
2623                         if ( !is_null( $title ) ) {
2624                                 if ( !$title->isExternal() ) {
2625                                         # Check for excessive inclusion
2626                                         $dbk = $title->getPrefixedDBkey();
2627                                         if ( $this->incrementIncludeCount( $dbk ) ) {
2628                                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() ) {
2629                                                         # Capture special page output
2630                                                         $text = SpecialPage::capturePath( $title );
2631                                                         if ( is_string( $text ) ) {
2632                                                                 $found = true;
2633                                                                 $noparse = true;
2634                                                                 $noargs = true;
2635                                                                 $isHTML = true;
2636                                                                 $this->disableCache();
2637                                                         }
2638                                                 } else {
2639                                                         $articleContent = $this->fetchTemplate( $title );
2640                                                         if ( $articleContent !== false ) {
2641                                                                 $found = true;
2642                                                                 $text = $articleContent;
2643                                                                 $replaceHeadings = true;
2644                                                         }
2645                                                 }
2646                                         }
2647
2648                                         # If the title is valid but undisplayable, make a link to it
2649                                         if ( $this->mOutputType == OT_HTML && !$found ) {
2650                                                 $text = '[['.$title->getPrefixedText().']]';
2651                                                 $found = true;
2652                                         }
2653                                 } elseif ( $title->isTrans() ) {
2654                                         // Interwiki transclusion
2655                                         if ( $this->mOutputType == OT_HTML && !$forceRawInterwiki ) {
2656                                                 $text = $this->interwikiTransclude( $title, 'render' );
2657                                                 $isHTML = true;
2658                                                 $noparse = true;
2659                                         } else {
2660                                                 $text = $this->interwikiTransclude( $title, 'raw' );
2661                                                 $replaceHeadings = true;
2662                                         }
2663                                         $found = true;
2664                                 }
2665
2666                                 # Template cache array insertion
2667                                 # Use the original $piece['title'] not the mangled $part1, so that
2668                                 # modifiers such as RAW: produce separate cache entries
2669                                 if( $found ) {
2670                                         $this->mTemplates[$piece['title']] = $text;
2671                                         $text = $linestart . $text;
2672                                 }
2673                         }
2674                 }
2675
2676                 # Recursive parsing, escaping and link table handling
2677                 # Only for HTML output
2678                 if ( $nowiki && $found && $this->mOutputType == OT_HTML ) {
2679                         $text = wfEscapeWikiText( $text );
2680                 } elseif ( ($this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI) && $found ) {
2681                         if ( !$noargs ) {
2682                                 # Clean up argument array
2683                                 $assocArgs = array();
2684                                 $index = 1;
2685                                 foreach( $args as $arg ) {
2686                                         $eqpos = strpos( $arg, '=' );
2687                                         if ( $eqpos === false ) {
2688                                                 $assocArgs[$index++] = $arg;
2689                                         } else {
2690                                                 $name = trim( substr( $arg, 0, $eqpos ) );
2691                                                 $value = trim( substr( $arg, $eqpos+1 ) );
2692                                                 if ( $value === false ) {
2693                                                         $value = '';
2694                                                 }
2695                                                 if ( $name !== false ) {
2696                                                         $assocArgs[$name] = $value;
2697                                                 }
2698                                         }
2699                                 }
2700
2701                                 # Add a new element to the templace recursion path
2702                                 $this->mTemplatePath[$part1] = 1;
2703                         }
2704
2705                         if ( !$noparse ) {
2706                                 # If there are any <onlyinclude> tags, only include them
2707                                 if ( in_string( '<onlyinclude>', $text ) && in_string( '</onlyinclude>', $text ) ) {
2708                                         preg_match_all( '/<onlyinclude>(.*?)\n?<\/onlyinclude>/s', $text, $m );
2709                                         $text = '';
2710                                         foreach ($m[1] as $piece)
2711                                                 $text .= $piece;
2712                                 }
2713                                 # Remove <noinclude> sections and <includeonly> tags
2714                                 $text = preg_replace( '/<noinclude>.*?<\/noinclude>/s', '', $text );
2715                                 $text = strtr( $text, array( '<includeonly>' => '' , '</includeonly>' => '' ) );
2716
2717                                 if( $this->mOutputType == OT_HTML ) {
2718                                         # Strip <nowiki>, <pre>, etc.
2719                                         $text = $this->strip( $text, $this->mStripState );
2720                                         $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'replaceVariables' ), $assocArgs );
2721                                 }
2722                                 $text = $this->replaceVariables( $text, $assocArgs );
2723
2724                                 # If the template begins with a table or block-level
2725                                 # element, it should be treated as beginning a new line.
2726                                 if (!$piece['lineStart'] && preg_match('/^({\\||:|;|#|\*)/', $text)) {
2727                                         $text = "\n" . $text;
2728                                 }
2729                         } elseif ( !$noargs ) {
2730                                 # $noparse and !$noargs
2731                                 # Just replace the arguments, not any double-brace items
2732                                 # This is used for rendered interwiki transclusion
2733                                 $text = $this->replaceVariables( $text, $assocArgs, true );
2734                         }
2735                 }
2736                 # Prune lower levels off the recursion check path
2737                 $this->mTemplatePath = $lastPathLevel;
2738
2739                 if ( !$found ) {
2740                         wfProfileOut( $fname );
2741                         return $piece['text'];
2742                 } else {
2743                         if ( $isHTML ) {
2744                                 # Replace raw HTML by a placeholder
2745                                 # Add a blank line preceding, to prevent it from mucking up
2746                                 # immediately preceding headings
2747                                 $text = "\n\n" . $this->insertStripItem( $text, $this->mStripState );
2748                         } else {
2749                                 # replace ==section headers==
2750                                 # XXX this needs to go away once we have a better parser.
2751                                 if ( $this->mOutputType != OT_WIKI && $replaceHeadings ) {
2752                                         if( !is_null( $title ) )
2753                                                 $encodedname = base64_encode($title->getPrefixedDBkey());
2754                                         else
2755                                                 $encodedname = base64_encode("");
2756                                         $m = preg_split('/(^={1,6}.*?={1,6}\s*?$)/m', $text, -1,
2757                                                 PREG_SPLIT_DELIM_CAPTURE);
2758                                         $text = '';
2759                                         $nsec = 0;
2760                                         for( $i = 0; $i < count($m); $i += 2 ) {
2761                                                 $text .= $m[$i];
2762                                                 if (!isset($m[$i + 1]) || $m[$i + 1] == "") continue;
2763                                                 $hl = $m[$i + 1];
2764                                                 if( strstr($hl, "<!--MWTEMPLATESECTION") ) {
2765                                                         $text .= $hl;
2766                                                         continue;
2767                                                 }
2768                                                 preg_match('/^(={1,6})(.*?)(={1,6})\s*?$/m', $hl, $m2);
2769                                                 $text .= $m2[1] . $m2[2] . "<!--MWTEMPLATESECTION="
2770                                                         . $encodedname . "&" . base64_encode("$nsec") . "-->" . $m2[3];
2771
2772                                                 $nsec++;
2773                                         }
2774                                 }
2775                         }
2776                 }
2777
2778                 # Prune lower levels off the recursion check path
2779                 $this->mTemplatePath = $lastPathLevel;
2780
2781                 if ( !$found ) {
2782                         wfProfileOut( $fname );
2783                         return $piece['text'];
2784                 } else {
2785                         wfProfileOut( $fname );
2786                         return $text;
2787                 }
2788         }
2789
2790         /**
2791          * Fetch the unparsed text of a template and register a reference to it.
2792          */
2793         function fetchTemplate( $title ) {
2794                 $text = false;
2795                 // Loop to fetch the article, with up to 1 redirect
2796                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
2797                         $rev = Revision::newFromTitle( $title );
2798                         $this->mOutput->addTemplate( $title, $title->getArticleID() );
2799                         if ( !$rev ) {
2800                                 break;
2801                         }
2802                         $text = $rev->getText();
2803                         if ( $text === false ) {
2804                                 break;
2805                         }
2806                         // Redirect?
2807                         $title = Title::newFromRedirect( $text );
2808                 }
2809                 return $text;
2810         }
2811
2812         /**
2813          * Transclude an interwiki link.
2814          */
2815         function interwikiTransclude( $title, $action ) {
2816                 global $wgEnableScaryTranscluding, $wgCanonicalNamespaceNames;
2817
2818                 if (!$wgEnableScaryTranscluding)
2819                         return wfMsg('scarytranscludedisabled');
2820
2821                 // The namespace will actually only be 0 or 10, depending on whether there was a leading :
2822                 // But we'll handle it generally anyway
2823                 if ( $title->getNamespace() ) {
2824                         // Use the canonical namespace, which should work anywhere
2825                         $articleName = $wgCanonicalNamespaceNames[$title->getNamespace()] . ':' . $title->getDBkey();
2826                 } else {
2827                         $articleName = $title->getDBkey();
2828                 }
2829
2830                 $url = str_replace('$1', urlencode($articleName), Title::getInterwikiLink($title->getInterwiki()));
2831                 $url .= "?action=$action";
2832                 if (strlen($url) > 255)
2833                         return wfMsg('scarytranscludetoolong');
2834                 return $this->fetchScaryTemplateMaybeFromCache($url);
2835         }
2836
2837         function fetchScaryTemplateMaybeFromCache($url) {
2838                 global $wgTranscludeCacheExpiry;
2839                 $dbr =& wfGetDB(DB_SLAVE);
2840                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
2841                                 array('tc_url' => $url));
2842                 if ($obj) {
2843                         $time = $obj->tc_time;
2844                         $text = $obj->tc_contents;
2845                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
2846                                 return $text;
2847                         }
2848                 }
2849
2850                 $text = wfGetHTTP($url);
2851                 if (!$text)
2852                         return wfMsg('scarytranscludefailed', $url);
2853
2854                 $dbw =& wfGetDB(DB_MASTER);
2855                 $dbw->replace('transcache', array('tc_url'), array(
2856                         'tc_url' => $url,
2857                         'tc_time' => time(),
2858                         'tc_contents' => $text));
2859                 return $text;
2860         }
2861
2862
2863         /**
2864          * Triple brace replacement -- used for template arguments
2865          * @access private
2866          */
2867         function argSubstitution( $matches ) {
2868                 $arg = trim( $matches['title'] );
2869                 $text = $matches['text'];
2870                 $inputArgs = end( $this->mArgStack );
2871
2872                 if ( array_key_exists( $arg, $inputArgs ) ) {
2873                         $text = $inputArgs[$arg];
2874                 } else if ($this->mOutputType == OT_HTML && null != $matches['parts'] && count($matches['parts']) > 0) {
2875                         $text = $matches['parts'][0];
2876                 }
2877
2878                 return $text;
2879         }
2880
2881         /**
2882          * Returns true if the function is allowed to include this entity
2883          * @access private
2884          */
2885         function incrementIncludeCount( $dbk ) {
2886                 if ( !array_key_exists( $dbk, $this->mIncludeCount ) ) {
2887                         $this->mIncludeCount[$dbk] = 0;
2888                 }
2889                 if ( ++$this->mIncludeCount[$dbk] <= MAX_INCLUDE_REPEAT ) {
2890                         return true;
2891                 } else {
2892                         return false;
2893                 }
2894         }
2895
2896         /**
2897          * This function accomplishes several tasks:
2898          * 1) Auto-number headings if that option is enabled
2899          * 2) Add an [edit] link to sections for logged in users who have enabled the option
2900          * 3) Add a Table of contents on the top for users who have enabled the option
2901          * 4) Auto-anchor headings
2902          *
2903          * It loops through all headlines, collects the necessary data, then splits up the
2904          * string and re-inserts the newly formatted headlines.
2905          *
2906          * @param string $text
2907          * @param boolean $isMain
2908          * @access private
2909          */
2910         function formatHeadings( $text, $isMain=true ) {
2911                 global $wgMaxTocLevel, $wgContLang;
2912
2913                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
2914                 $doShowToc = true;
2915                 $forceTocHere = false;
2916                 if( !$this->mTitle->userCanEdit() ) {
2917                         $showEditLink = 0;
2918                 } else {
2919                         $showEditLink = $this->mOptions->getEditSection();
2920                 }
2921
2922                 # Inhibit editsection links if requested in the page
2923                 $esw =& MagicWord::get( MAG_NOEDITSECTION );
2924                 if( $esw->matchAndRemove( $text ) ) {
2925                         $showEditLink = 0;
2926                 }
2927                 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
2928                 # do not add TOC
2929                 $mw =& MagicWord::get( MAG_NOTOC );
2930                 if( $mw->matchAndRemove( $text ) ) {
2931                         $doShowToc = false;
2932                 }
2933
2934                 # Get all headlines for numbering them and adding funky stuff like [edit]
2935                 # links - this is for later, but we need the number of headlines right now
2936                 $numMatches = preg_match_all( '/<H([1-6])(.*?'.'>)(.*?)<\/H[1-6] *>/i', $text, $matches );
2937
2938                 # if there are fewer than 4 headlines in the article, do not show TOC
2939                 if( $numMatches < 4 ) {
2940                         $doShowToc = false;
2941                 }
2942
2943                 # if the string __TOC__ (not case-sensitive) occurs in the HTML,
2944                 # override above conditions and always show TOC at that place
2945
2946                 $mw =& MagicWord::get( MAG_TOC );
2947                 if($mw->match( $text ) ) {
2948                         $doShowToc = true;
2949                         $forceTocHere = true;
2950                 } else {
2951                         # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
2952                         # override above conditions and always show TOC above first header
2953                         $mw =& MagicWord::get( MAG_FORCETOC );
2954                         if ($mw->matchAndRemove( $text ) ) {
2955                                 $doShowToc = true;
2956                         }
2957                 }
2958
2959                 # Never ever show TOC if no headers
2960                 if( $numMatches < 1 ) {
2961                         $doShowToc = false;
2962                 }
2963
2964                 # We need this to perform operations on the HTML
2965                 $sk =& $this->mOptions->getSkin();
2966
2967                 # headline counter
2968                 $headlineCount = 0;
2969                 $sectionCount = 0; # headlineCount excluding template sections
2970
2971                 # Ugh .. the TOC should have neat indentation levels which can be
2972                 # passed to the skin functions. These are determined here
2973                 $toc = '';
2974                 $full = '';
2975                 $head = array();
2976                 $sublevelCount = array();
2977                 $levelCount = array();
2978                 $toclevel = 0;
2979                 $level = 0;
2980                 $prevlevel = 0;
2981                 $toclevel = 0;
2982                 $prevtoclevel = 0;
2983
2984                 foreach( $matches[3] as $headline ) {
2985                         $istemplate = 0;
2986                         $templatetitle = '';
2987                         $templatesection = 0;
2988                         $numbering = '';
2989
2990                         if (preg_match("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", $headline, $mat)) {
2991                                 $istemplate = 1;
2992                                 $templatetitle = base64_decode($mat[1]);
2993                                 $templatesection = 1 + (int)base64_decode($mat[2]);
2994                                 $headline = preg_replace("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", "", $headline);
2995                         }
2996
2997                         if( $toclevel ) {
2998                                 $prevlevel = $level;
2999                                 $prevtoclevel = $toclevel;
3000                         }
3001                         $level = $matches[1][$headlineCount];
3002
3003                         if( $doNumberHeadings || $doShowToc ) {
3004
3005                                 if ( $level > $prevlevel ) {
3006                                         # Increase TOC level
3007                                         $toclevel++;
3008                                         $sublevelCount[$toclevel] = 0;
3009                                         $toc .= $sk->tocIndent();
3010                                 }
3011                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3012                                         # Decrease TOC level, find level to jump to
3013
3014                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
3015                                                 # Can only go down to level 1
3016                                                 $toclevel = 1;
3017                                         } else {
3018                                                 for ($i = $toclevel; $i > 0; $i--) {
3019                                                         if ( $levelCount[$i] == $level ) {
3020                                                                 # Found last matching level
3021                                                                 $toclevel = $i;
3022                                                                 break;
3023                                                         }
3024                                                         elseif ( $levelCount[$i] < $level ) {
3025                                                                 # Found first matching level below current level
3026                                                                 $toclevel = $i + 1;
3027                                                                 break;
3028                                                         }
3029                                                 }
3030                                         }
3031
3032                                         $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3033                                 }
3034                                 else {
3035                                         # No change in level, end TOC line
3036                                         $toc .= $sk->tocLineEnd();
3037                                 }
3038
3039                                 $levelCount[$toclevel] = $level;
3040
3041                                 # count number of headlines for each level
3042                                 @$sublevelCount[$toclevel]++;
3043                                 $dot = 0;
3044                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3045                                         if( !empty( $sublevelCount[$i] ) ) {
3046                                                 if( $dot ) {
3047                                                         $numbering .= '.';
3048                                                 }
3049                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3050                                                 $dot = 1;
3051                                         }
3052                                 }
3053                         }
3054
3055                         # The canonized header is a version of the header text safe to use for links
3056                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3057                         $canonized_headline = $this->unstrip( $headline, $this->mStripState );
3058                         $canonized_headline = $this->unstripNoWiki( $canonized_headline, $this->mStripState );
3059
3060                         # Remove link placeholders by the link text.
3061                         #     <!--LINK number-->
3062                         # turns into
3063                         #     link text with suffix
3064                         $canonized_headline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
3065                                                             "\$this->mLinkHolders['texts'][\$1]",
3066                                                             $canonized_headline );
3067                         $canonized_headline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
3068                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
3069                                                             $canonized_headline );
3070
3071                         # strip out HTML
3072                         $canonized_headline = preg_replace( '/<.*?' . '>/','',$canonized_headline );
3073                         $tocline = trim( $canonized_headline );
3074                         # Save headline for section edit hint before it's escaped
3075                         $headline_hint = trim( $canonized_headline );
3076                         $canonized_headline = Sanitizer::escapeId( $tocline );
3077                         $refers[$headlineCount] = $canonized_headline;
3078
3079                         # count how many in assoc. array so we can track dupes in anchors
3080                         @$refers[$canonized_headline]++;
3081                         $refcount[$headlineCount]=$refers[$canonized_headline];
3082
3083                         # Don't number the heading if it is the only one (looks silly)
3084                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3085                                 # the two are different if the line contains a link
3086                                 $headline=$numbering . ' ' . $headline;
3087                         }
3088
3089                         # Create the anchor for linking from the TOC to the section
3090                         $anchor = $canonized_headline;
3091                         if($refcount[$headlineCount] > 1 ) {
3092                                 $anchor .= '_' . $refcount[$headlineCount];
3093                         }
3094                         if( $doShowToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3095                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3096                         }
3097                         if( $showEditLink && ( !$istemplate || $templatetitle !== "" ) ) {
3098                                 if ( empty( $head[$headlineCount] ) ) {
3099                                         $head[$headlineCount] = '';
3100                                 }
3101                                 if( $istemplate )
3102                                         $head[$headlineCount] .= $sk->editSectionLinkForOther($templatetitle, $templatesection);
3103                                 else
3104                                         $head[$headlineCount] .= $sk->editSectionLink($this->mTitle, $sectionCount+1, $headline_hint);
3105                         }
3106
3107                         # give headline the correct <h#> tag
3108                         @$head[$headlineCount] .= "<a name=\"$anchor\"></a><h".$level.$matches[2][$headlineCount] .$headline.'</h'.$level.'>';
3109
3110                         $headlineCount++;
3111                         if( !$istemplate )
3112                                 $sectionCount++;
3113                 }
3114
3115                 if( $doShowToc ) {
3116                         $toc .= $sk->tocUnindent( $toclevel - 1 );
3117                         $toc = $sk->tocList( $toc );
3118                 }
3119
3120                 # split up and insert constructed headlines
3121
3122                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3123                 $i = 0;
3124
3125                 foreach( $blocks as $block ) {
3126                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
3127                                 # This is the [edit] link that appears for the top block of text when
3128                                 # section editing is enabled
3129
3130                                 # Disabled because it broke block formatting
3131                                 # For example, a bullet point in the top line
3132                                 # $full .= $sk->editSectionLink(0);
3133                         }
3134                         $full .= $block;
3135                         if( $doShowToc && !$i && $isMain && !$forceTocHere) {
3136                         # Top anchor now in skin
3137                                 $full = $full.$toc;
3138                         }
3139
3140                         if( !empty( $head[$i] ) ) {
3141                                 $full .= $head[$i];
3142                         }
3143                         $i++;
3144                 }
3145                 if($forceTocHere) {
3146                         $mw =& MagicWord::get( MAG_TOC );
3147                         return $mw->replace( $toc, $full );
3148                 } else {
3149                         return $full;
3150                 }
3151         }
3152
3153         /**
3154          * Return an HTML link for the "ISBN 123456" text
3155          * @access private
3156          */
3157         function magicISBN( $text ) {
3158                 $fname = 'Parser::magicISBN';
3159                 wfProfileIn( $fname );
3160
3161                 $a = split( 'ISBN ', ' '.$text );
3162                 if ( count ( $a ) < 2 ) {
3163                         wfProfileOut( $fname );
3164                         return $text;
3165                 }
3166                 $text = substr( array_shift( $a ), 1);
3167                 $valid = '0123456789-Xx';
3168
3169                 foreach ( $a as $x ) {
3170                         $isbn = $blank = '' ;
3171                         while ( ' ' == $x{0} ) {
3172                                 $blank .= ' ';
3173                                 $x = substr( $x, 1 );
3174                         }
3175                         if ( $x == '' ) { # blank isbn
3176                                 $text .= "ISBN $blank";
3177                                 continue;
3178                         }
3179                         while ( strstr( $valid, $x{0} ) != false ) {
3180                                 $isbn .= $x{0};
3181                                 $x = substr( $x, 1 );
3182                         }
3183                         $num = str_replace( '-', '', $isbn );
3184                         $num = str_replace( ' ', '', $num );
3185                         $num = str_replace( 'x', 'X', $num );
3186
3187                         if ( '' == $num ) {
3188                                 $text .= "ISBN $blank$x";
3189                         } else {
3190                                 $titleObj = Title::makeTitle( NS_SPECIAL, 'Booksources' );
3191                                 $text .= '<a href="' .
3192                                 $titleObj->escapeLocalUrl( 'isbn='.$num ) .
3193                                         "\" class=\"internal\">ISBN $isbn</a>";
3194                                 $text .= $x;
3195                         }
3196                 }
3197                 wfProfileOut( $fname );
3198                 return $text;
3199         }
3200
3201         /**
3202          * Return an HTML link for the "RFC 1234" text
3203          *
3204          * @access private
3205          * @param string $text     Text to be processed
3206          * @param string $keyword  Magic keyword to use (default RFC)
3207          * @param string $urlmsg   Interface message to use (default rfcurl)
3208          * @return string
3209          */
3210         function magicRFC( $text, $keyword='RFC ', $urlmsg='rfcurl'  ) {
3211
3212                 $valid = '0123456789';
3213                 $internal = false;
3214
3215                 $a = split( $keyword, ' '.$text );
3216                 if ( count ( $a ) < 2 ) {
3217                         return $text;
3218                 }
3219                 $text = substr( array_shift( $a ), 1);
3220
3221                 /* Check if keyword is preceed by [[.
3222                  * This test is made here cause of the array_shift above
3223                  * that prevent the test to be done in the foreach.
3224                  */
3225                 if ( substr( $text, -2 ) == '[[' ) {
3226                         $internal = true;
3227                 }
3228
3229                 foreach ( $a as $x ) {
3230                         /* token might be empty if we have RFC RFC 1234 */
3231                         if ( $x=='' ) {
3232                                 $text.=$keyword;
3233                                 continue;
3234                                 }
3235
3236                         $id = $blank = '' ;
3237
3238                         /** remove and save whitespaces in $blank */
3239                         while ( $x{0} == ' ' ) {
3240                                 $blank .= ' ';
3241                                 $x = substr( $x, 1 );
3242                         }
3243
3244                         /** remove and save the rfc number in $id */
3245                         while ( strstr( $valid, $x{0} ) != false ) {
3246                                 $id .= $x{0};
3247                                 $x = substr( $x, 1 );
3248                         }
3249
3250                         if ( $id == '' ) {
3251                                 /* call back stripped spaces*/
3252                                 $text .= $keyword.$blank.$x;
3253                         } elseif( $internal ) {
3254                                 /* normal link */
3255                                 $text .= $keyword.$id.$x;
3256                         } else {
3257                                 /* build the external link*/
3258                                 $url = wfMsg( $urlmsg, $id);
3259                                 $sk =& $this->mOptions->getSkin();
3260                                 $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
3261                                 $text .= "<a href='{$url}'{$la}>{$keyword}{$id}</a>{$x}";
3262                         }
3263
3264                         /* Check if the next RFC keyword is preceed by [[ */
3265                         $internal = ( substr($x,-2) == '[[' );
3266                 }
3267                 return $text;
3268         }
3269
3270         /**
3271          * Transform wiki markup when saving a page by doing \r\n -> \n
3272          * conversion, substitting signatures, {{subst:}} templates, etc.
3273          *
3274          * @param string $text the text to transform
3275          * @param Title &$title the Title object for the current article
3276          * @param User &$user the User object describing the current user
3277          * @param ParserOptions $options parsing options
3278          * @param bool $clearState whether to clear the parser state first
3279          * @return string the altered wiki markup
3280          * @access public
3281          */
3282         function preSaveTransform( $text, &$title, &$user, $options, $clearState = true ) {
3283                 $this->mOptions = $options;
3284                 $this->mTitle =& $title;
3285                 $this->mOutputType = OT_WIKI;
3286
3287                 if ( $clearState ) {
3288                         $this->clearState();
3289                 }
3290
3291                 $stripState = false;
3292                 $pairs = array(
3293                         "\r\n" => "\n",
3294                 );
3295                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3296                 $text = $this->strip( $text, $stripState, true );
3297                 $text = $this->pstPass2( $text, $user );
3298                 $text = $this->unstrip( $text, $stripState );
3299                 $text = $this->unstripNoWiki( $text, $stripState );
3300                 return $text;
3301         }
3302
3303         /**
3304          * Pre-save transform helper function
3305          * @access private
3306          */
3307         function pstPass2( $text, &$user ) {
3308                 global $wgContLang, $wgLocaltimezone;
3309
3310                 /* Note: This is the timestamp saved as hardcoded wikitext to
3311                  * the database, we use $wgContLang here in order to give
3312                  * everyone the same signiture and use the default one rather
3313                  * than the one selected in each users preferences.
3314                  */
3315                 if ( isset( $wgLocaltimezone ) ) {
3316                         $oldtz = getenv( 'TZ' );
3317                         putenv( 'TZ='.$wgLocaltimezone );
3318                 }
3319                 $d = $wgContLang->timeanddate( date( 'YmdHis' ), false, false) .
3320                   ' (' . date( 'T' ) . ')';
3321                 if ( isset( $wgLocaltimezone ) ) {
3322                         putenv( 'TZ='.$oldtz );
3323                 }
3324
3325                 # Variable replacement
3326                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3327                 $text = $this->replaceVariables( $text );
3328
3329                 # Signatures
3330                 $sigText = $this->getUserSig( $user );
3331                 $text = strtr( $text, array(
3332                         '~~~~~' => $d,
3333                         '~~~~' => "$sigText $d",
3334                         '~~~' => $sigText
3335                 ) );
3336
3337                 # Context links: [[|name]] and [[name (context)|]]
3338                 #
3339                 global $wgLegalTitleChars;
3340                 $tc = "[$wgLegalTitleChars]";
3341                 $np = str_replace( array( '(', ')' ), array( '', '' ), $tc ); # No parens
3342
3343                 $namespacechar = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
3344                 $conpat = "/^({$np}+) \\(({$tc}+)\\)$/";
3345
3346                 $p1 = "/\[\[({$np}+) \\(({$np}+)\\)\\|]]/";             # [[page (context)|]]
3347                 $p2 = "/\[\[\\|({$tc}+)]]/";                                    # [[|page]]
3348                 $p3 = "/\[\[(:*$namespacechar+):({$np}+)\\|]]/";                # [[namespace:page|]] and [[:namespace:page|]]
3349                 $p4 = "/\[\[(:*$namespacechar+):({$np}+) \\(({$np}+)\\)\\|]]/"; # [[ns:page (cont)|]] and [[:ns:page (cont)|]]
3350                 $context = '';
3351                 $t = $this->mTitle->getText();
3352                 if ( preg_match( $conpat, $t, $m ) ) {
3353                         $context = $m[2];
3354                 }
3355                 $text = preg_replace( $p4, '[[\\1:\\2 (\\3)|\\2]]', $text );
3356                 $text = preg_replace( $p1, '[[\\1 (\\2)|\\1]]', $text );
3357                 $text = preg_replace( $p3, '[[\\1:\\2|\\2]]', $text );
3358
3359                 if ( '' == $context ) {
3360                         $text = preg_replace( $p2, '[[\\1]]', $text );
3361                 } else {
3362                         $text = preg_replace( $p2, "[[\\1 ({$context})|\\1]]", $text );
3363                 }
3364
3365                 # Trim trailing whitespace
3366                 # MAG_END (__END__) tag allows for trailing
3367                 # whitespace to be deliberately included
3368                 $text = rtrim( $text );
3369                 $mw =& MagicWord::get( MAG_END );
3370                 $mw->matchAndRemove( $text );
3371
3372                 return $text;
3373         }
3374
3375         /**
3376          * Fetch the user's signature text, if any, and normalize to
3377          * validated, ready-to-insert wikitext.
3378          *
3379          * @param User $user
3380          * @return string
3381          * @access private
3382          */
3383         function getUserSig( &$user ) {
3384                 $username = $user->getName();
3385                 $nickname = $user->getOption( 'nickname' );
3386                 $nickname = $nickname === '' ? $username : $nickname;
3387
3388                 if( $user->getBoolOption( 'fancysig' ) !== false ) {
3389                         # Sig. might contain markup; validate this
3390                         if( $this->validateSig( $nickname ) !== false ) {
3391                                 # Validated; clean up (if needed) and return it
3392                                 return( $this->cleanSig( $nickname ) );
3393                         } else {
3394                                 # Failed to validate; fall back to the default
3395                                 $nickname = $username;
3396                                 wfDebug( "Parser::getUserSig: $username has bad XML tags in signature.\n" );
3397                         }
3398                 }
3399
3400                 # If we're still here, make it a link to the user page
3401                 $userpage = $user->getUserPage();
3402                 return( '[[' . $userpage->getPrefixedText() . '|' . wfEscapeWikiText( $nickname ) . ']]' );
3403         }
3404
3405         /**
3406          * Check that the user's signature contains no bad XML
3407          *
3408          * @param string $text
3409          * @return mixed An expanded string, or false if invalid.
3410          */
3411         function validateSig( $text ) {
3412                 return( wfIsWellFormedXmlFragment( $text ) ? $text : false );
3413         }
3414
3415         /**
3416          * Clean up signature text
3417          *
3418          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures
3419          * 2) Substitute all transclusions
3420          *
3421          * @param string $text
3422          * @return string Signature text
3423          */
3424         function cleanSig( $text ) {
3425                 $substWord = MagicWord::get( MAG_SUBST );
3426                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3427                 $substText = '{{' . $substWord->getSynonym( 0 );
3428
3429                 $text = preg_replace( $substRegex, $substText, $text );
3430                 $text = preg_replace( '/~{3,5}/', '', $text );
3431                 $text = $this->replaceVariables( $text );
3432
3433                 return $text;
3434         }
3435
3436         /**
3437          * Set up some variables which are usually set up in parse()
3438          * so that an external function can call some class members with confidence
3439          * @access public
3440          */
3441         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
3442                 $this->mTitle =& $title;
3443                 $this->mOptions = $options;
3444                 $this->mOutputType = $outputType;
3445                 if ( $clearState ) {
3446                         $this->clearState();
3447                 }
3448         }
3449
3450         /**
3451          * Transform a MediaWiki message by replacing magic variables.
3452          *
3453          * @param string $text the text to transform
3454          * @param ParserOptions $options  options
3455          * @return string the text with variables substituted
3456          * @access public
3457          */
3458         function transformMsg( $text, $options ) {
3459                 global $wgTitle;
3460                 static $executing = false;
3461
3462                 $fname = "Parser::transformMsg";
3463
3464                 # Guard against infinite recursion
3465                 if ( $executing ) {
3466                         return $text;
3467                 }
3468                 $executing = true;
3469
3470                 wfProfileIn($fname);
3471
3472                 $this->mTitle = $wgTitle;
3473                 $this->mOptions = $options;
3474                 $this->mOutputType = OT_MSG;
3475                 $this->clearState();
3476                 $text = $this->replaceVariables( $text );
3477
3478                 $executing = false;
3479                 wfProfileOut($fname);
3480                 return $text;
3481         }
3482
3483         /**
3484          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
3485          * The callback should have the following form:
3486          *    function myParserHook( $text, $params, &$parser ) { ... }
3487          *
3488          * Transform and return $text. Use $parser for any required context, e.g. use
3489          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
3490          *
3491          * @access public
3492          *
3493          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
3494          * @param mixed $callback The callback function (and object) to use for the tag
3495          *
3496          * @return The old value of the mTagHooks array associated with the hook
3497          */
3498         function setHook( $tag, $callback ) {
3499                 $oldVal = @$this->mTagHooks[$tag];
3500                 $this->mTagHooks[$tag] = $callback;
3501
3502                 return $oldVal;
3503         }
3504
3505         /**
3506          * Replace <!--LINK--> link placeholders with actual links, in the buffer
3507          * Placeholders created in Skin::makeLinkObj()
3508          * Returns an array of links found, indexed by PDBK:
3509          *  0 - broken
3510          *  1 - normal link
3511          *  2 - stub
3512          * $options is a bit field, RLH_FOR_UPDATE to select for update
3513          */
3514         function replaceLinkHolders( &$text, $options = 0 ) {
3515                 global $wgUser;
3516                 global $wgOutputReplace;
3517
3518                 $fname = 'Parser::replaceLinkHolders';
3519                 wfProfileIn( $fname );
3520
3521                 $pdbks = array();
3522                 $colours = array();
3523                 $sk =& $this->mOptions->getSkin();
3524                 $linkCache =& LinkCache::singleton();
3525
3526                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
3527                         wfProfileIn( $fname.'-check' );
3528                         $dbr =& wfGetDB( DB_SLAVE );
3529                         $page = $dbr->tableName( 'page' );
3530                         $threshold = $wgUser->getOption('stubthreshold');
3531
3532                         # Sort by namespace
3533                         asort( $this->mLinkHolders['namespaces'] );
3534
3535                         # Generate query
3536                         $query = false;
3537                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3538                                 # Make title object
3539                                 $title = $this->mLinkHolders['titles'][$key];
3540
3541                                 # Skip invalid entries.
3542                                 # Result will be ugly, but prevents crash.
3543                                 if ( is_null( $title ) ) {
3544                                         continue;
3545                                 }
3546                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
3547
3548                                 # Check if it's a static known link, e.g. interwiki
3549                                 if ( $title->isAlwaysKnown() ) {
3550                                         $colours[$pdbk] = 1;
3551                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
3552                                         $colours[$pdbk] = 1;
3553                                         $this->mOutput->addLink( $title, $id );
3554                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
3555                                         $colours[$pdbk] = 0;
3556                                 } else {
3557                                         # Not in the link cache, add it to the query
3558                                         if ( !isset( $current ) ) {
3559                                                 $current = $ns;
3560                                                 $query =  "SELECT page_id, page_namespace, page_title";
3561                                                 if ( $threshold > 0 ) {
3562                                                         $query .= ', page_len, page_is_redirect';
3563                                                 }
3564                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
3565                                         } elseif ( $current != $ns ) {
3566                                                 $current = $ns;
3567                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
3568                                         } else {
3569                                                 $query .= ', ';
3570                                         }
3571
3572                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
3573                                 }
3574                         }
3575                         if ( $query ) {
3576                                 $query .= '))';
3577                                 if ( $options & RLH_FOR_UPDATE ) {
3578                                         $query .= ' FOR UPDATE';
3579                                 }
3580
3581                                 $res = $dbr->query( $query, $fname );
3582
3583                                 # Fetch data and form into an associative array
3584                                 # non-existent = broken
3585                                 # 1 = known
3586                                 # 2 = stub
3587                                 while ( $s = $dbr->fetchObject($res) ) {
3588                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
3589                                         $pdbk = $title->getPrefixedDBkey();
3590                                         $linkCache->addGoodLinkObj( $s->page_id, $title );
3591                                         $this->mOutput->addLink( $title, $s->page_id );
3592
3593                                         if ( $threshold >  0 ) {
3594                                                 $size = $s->page_len;
3595                                                 if ( $s->page_is_redirect || $s->page_namespace != 0 || $size >= $threshold ) {
3596                                                         $colours[$pdbk] = 1;
3597                                                 } else {
3598                                                         $colours[$pdbk] = 2;
3599                                                 }
3600                                         } else {
3601                                                 $colours[$pdbk] = 1;
3602                                         }
3603                                 }
3604                         }
3605                         wfProfileOut( $fname.'-check' );
3606
3607                         # Construct search and replace arrays
3608                         wfProfileIn( $fname.'-construct' );
3609                         $wgOutputReplace = array();
3610                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3611                                 $pdbk = $pdbks[$key];
3612                                 $searchkey = "<!--LINK $key-->";
3613                                 $title = $this->mLinkHolders['titles'][$key];
3614                                 if ( empty( $colours[$pdbk] ) ) {
3615                                         $linkCache->addBadLinkObj( $title );
3616                                         $colours[$pdbk] = 0;
3617                                         $this->mOutput->addLink( $title, 0 );
3618                                         $wgOutputReplace[$searchkey] = $sk->makeBrokenLinkObj( $title,
3619                                                                         $this->mLinkHolders['texts'][$key],
3620                                                                         $this->mLinkHolders['queries'][$key] );
3621                                 } elseif ( $colours[$pdbk] == 1 ) {
3622                                         $wgOutputReplace[$searchkey] = $sk->makeKnownLinkObj( $title,
3623                                                                         $this->mLinkHolders['texts'][$key],
3624                                                                         $this->mLinkHolders['queries'][$key] );
3625                                 } elseif ( $colours[$pdbk] == 2 ) {
3626                                         $wgOutputReplace[$searchkey] = $sk->makeStubLinkObj( $title,
3627                                                                         $this->mLinkHolders['texts'][$key],
3628                                                                         $this->mLinkHolders['queries'][$key] );
3629                                 }
3630                         }
3631                         wfProfileOut( $fname.'-construct' );
3632
3633                         # Do the thing
3634                         wfProfileIn( $fname.'-replace' );
3635
3636                         $text = preg_replace_callback(
3637                                 '/(<!--LINK .*?-->)/',
3638                                 "wfOutputReplaceMatches",
3639                                 $text);
3640
3641                         wfProfileOut( $fname.'-replace' );
3642                 }
3643
3644                 # Now process interwiki link holders
3645                 # This is quite a bit simpler than internal links
3646                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
3647                         wfProfileIn( $fname.'-interwiki' );
3648                         # Make interwiki link HTML
3649                         $wgOutputReplace = array();
3650                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
3651                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
3652                                 $wgOutputReplace[$key] = $sk->makeLinkObj( $title, $link );
3653                         }
3654
3655                         $text = preg_replace_callback(
3656                                 '/<!--IWLINK (.*?)-->/',
3657                                 "wfOutputReplaceMatches",
3658                                 $text );
3659                         wfProfileOut( $fname.'-interwiki' );
3660                 }
3661
3662                 wfProfileOut( $fname );
3663                 return $colours;
3664         }
3665
3666         /**
3667          * Replace <!--LINK--> link placeholders with plain text of links
3668          * (not HTML-formatted).
3669          * @param string $text
3670          * @return string
3671          */
3672         function replaceLinkHoldersText( $text ) {
3673                 $fname = 'Parser::replaceLinkHoldersText';
3674                 wfProfileIn( $fname );
3675
3676                 $text = preg_replace_callback(
3677                         '/<!--(LINK|IWLINK) (.*?)-->/',
3678                         array( &$this, 'replaceLinkHoldersTextCallback' ),
3679                         $text );
3680
3681                 wfProfileOut( $fname );
3682                 return $text;
3683         }
3684
3685         /**
3686          * @param array $matches
3687          * @return string
3688          * @access private
3689          */
3690         function replaceLinkHoldersTextCallback( $matches ) {
3691                 $type = $matches[1];
3692                 $key  = $matches[2];
3693                 if( $type == 'LINK' ) {
3694                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
3695                                 return $this->mLinkHolders['texts'][$key];
3696                         }
3697                 } elseif( $type == 'IWLINK' ) {
3698                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
3699                                 return $this->mInterwikiLinkHolders['texts'][$key];
3700                         }
3701                 }
3702                 return $matches[0];
3703         }
3704
3705         /**
3706          * Renders an image gallery from a text with one line per image.
3707          * text labels may be given by using |-style alternative text. E.g.
3708          *   Image:one.jpg|The number "1"
3709          *   Image:tree.jpg|A tree
3710          * given as text will return the HTML of a gallery with two images,
3711          * labeled 'The number "1"' and
3712          * 'A tree'.
3713          */
3714         function renderImageGallery( $text ) {
3715                 # Setup the parser
3716                 $parserOptions = new ParserOptions;
3717                 $localParser = new Parser();
3718
3719                 $ig = new ImageGallery();
3720                 $ig->setShowBytes( false );
3721                 $ig->setShowFilename( false );
3722                 $lines = explode( "\n", $text );
3723
3724                 foreach ( $lines as $line ) {
3725                         # match lines like these:
3726                         # Image:someimage.jpg|This is some image
3727                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
3728                         # Skip empty lines
3729                         if ( count( $matches ) == 0 ) {
3730                                 continue;
3731                         }
3732                         $nt =& Title::newFromText( $matches[1] );
3733                         if( is_null( $nt ) ) {
3734                                 # Bogus title. Ignore these so we don't bomb out later.
3735                                 continue;
3736                         }
3737                         if ( isset( $matches[3] ) ) {
3738                                 $label = $matches[3];
3739                         } else {
3740                                 $label = '';
3741                         }
3742
3743                         $pout = $localParser->parse( $label , $this->mTitle, $parserOptions );
3744                         $html = $pout->getText();
3745
3746                         $ig->add( new Image( $nt ), $html );
3747                         $this->mOutput->addImage( $nt->getDBkey() );
3748                 }
3749                 return $ig->toHTML();
3750         }
3751
3752         /**
3753          * Parse image options text and use it to make an image
3754          */
3755         function makeImage( &$nt, $options ) {
3756                 global $wgUseImageResize;
3757
3758                 $align = '';
3759
3760                 # Check if the options text is of the form "options|alt text"
3761                 # Options are:
3762                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
3763                 #  * left               no resizing, just left align. label is used for alt= only
3764                 #  * right              same, but right aligned
3765                 #  * none               same, but not aligned
3766                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
3767                 #  * center             center the image
3768                 #  * framed             Keep original image size, no magnify-button.
3769
3770                 $part = explode( '|', $options);
3771
3772                 $mwThumb  =& MagicWord::get( MAG_IMG_THUMBNAIL );
3773                 $mwManualThumb =& MagicWord::get( MAG_IMG_MANUALTHUMB );
3774                 $mwLeft   =& MagicWord::get( MAG_IMG_LEFT );
3775                 $mwRight  =& MagicWord::get( MAG_IMG_RIGHT );
3776                 $mwNone   =& MagicWord::get( MAG_IMG_NONE );
3777                 $mwWidth  =& MagicWord::get( MAG_IMG_WIDTH );
3778                 $mwCenter =& MagicWord::get( MAG_IMG_CENTER );
3779                 $mwFramed =& MagicWord::get( MAG_IMG_FRAMED );
3780                 $caption = '';
3781
3782                 $width = $height = $framed = $thumb = false;
3783                 $manual_thumb = '' ;
3784
3785                 foreach( $part as $key => $val ) {
3786                         if ( $wgUseImageResize && ! is_null( $mwThumb->matchVariableStartToEnd($val) ) ) {
3787                                 $thumb=true;
3788                         } elseif ( ! is_null( $match = $mwManualThumb->matchVariableStartToEnd($val) ) ) {
3789                                 # use manually specified thumbnail
3790                                 $thumb=true;
3791                                 $manual_thumb = $match;
3792                         } elseif ( ! is_null( $mwRight->matchVariableStartToEnd($val) ) ) {
3793                                 # remember to set an alignment, don't render immediately
3794                                 $align = 'right';
3795                         } elseif ( ! is_null( $mwLeft->matchVariableStartToEnd($val) ) ) {
3796                                 # remember to set an alignment, don't render immediately
3797                                 $align = 'left';
3798                         } elseif ( ! is_null( $mwCenter->matchVariableStartToEnd($val) ) ) {
3799                                 # remember to set an alignment, don't render immediately
3800                                 $align = 'center';
3801                         } elseif ( ! is_null( $mwNone->matchVariableStartToEnd($val) ) ) {
3802                                 # remember to set an alignment, don't render immediately
3803                                 $align = 'none';
3804                         } elseif ( $wgUseImageResize && ! is_null( $match = $mwWidth->matchVariableStartToEnd($val) ) ) {
3805                                 wfDebug( "MAG_IMG_WIDTH match: $match\n" );
3806                                 # $match is the image width in pixels
3807                                 if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $match, $m ) ) {
3808                                         $width = intval( $m[1] );
3809                                         $height = intval( $m[2] );
3810                                 } else {
3811                                         $width = intval($match);
3812                                 }
3813                         } elseif ( ! is_null( $mwFramed->matchVariableStartToEnd($val) ) ) {
3814                                 $framed=true;
3815                         } else {
3816                                 $caption = $val;
3817                         }
3818                 }
3819                 # Strip bad stuff out of the alt text
3820                 $alt = $this->replaceLinkHoldersText( $caption );
3821                 $alt = Sanitizer::stripAllTags( $alt );
3822
3823                 # Linker does the rest
3824                 $sk =& $this->mOptions->getSkin();
3825                 return $sk->makeImageLinkObj( $nt, $caption, $alt, $align, $width, $height, $framed, $thumb, $manual_thumb );
3826         }
3827
3828         /**
3829          * Set a flag in the output object indicating that the content is dynamic and
3830          * shouldn't be cached.
3831          */
3832         function disableCache() {
3833                 $this->mOutput->mCacheTime = -1;
3834         }
3835
3836         /**#@+
3837          * Callback from the Sanitizer for expanding items found in HTML attribute
3838          * values, so they can be safely tested and escaped.
3839          * @param string $text
3840          * @param array $args
3841          * @return string
3842          * @access private
3843          */
3844         function attributeStripCallback( &$text, $args ) {
3845                 $text = $this->replaceVariables( $text, $args );
3846                 $text = $this->unstripForHTML( $text );
3847                 return $text;
3848         }
3849
3850         function unstripForHTML( $text ) {
3851                 $text = $this->unstrip( $text, $this->mStripState );
3852                 $text = $this->unstripNoWiki( $text, $this->mStripState );
3853                 return $text;
3854         }
3855         /**#@-*/
3856
3857         /**#@+
3858          * Accessor/mutator
3859          */
3860         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
3861         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
3862         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
3863         /**#@-*/
3864
3865         /**#@+
3866          * Accessor
3867          */
3868         function getTags() { return array_keys( $this->mTagHooks ); }
3869         /**#@-*/
3870 }
3871
3872 /**
3873  * @todo document
3874  * @package MediaWiki
3875  */
3876 class ParserOutput
3877 {
3878         var $mText,             # The output text
3879                 $mLanguageLinks,    # List of the full text of language links, in the order they appear
3880                 $mCategories,       # Map of category names to sort keys
3881                 $mContainsOldMagic, # Boolean variable indicating if the input contained variables like {{CURRENTDAY}}
3882                 $mCacheTime,        # Time when this object was generated, or -1 for uncacheable. Used in ParserCache.
3883                 $mVersion,          # Compatibility check
3884                 $mTitleText,        # title text of the chosen language variant
3885                 $mLinks,            # 2-D map of NS/DBK to ID for the links in the document. ID=zero for broken.
3886                 $mTemplates,        # 2-D map of NS/DBK to ID for the template references. ID=zero for broken.
3887                 $mImages,           # DB keys of the images used, in the array key only
3888                 $mExternalLinks;    # External link URLs, in the key only
3889
3890         function ParserOutput( $text = '', $languageLinks = array(), $categoryLinks = array(),
3891                 $containsOldMagic = false, $titletext = '' )
3892         {
3893                 $this->mText = $text;
3894                 $this->mLanguageLinks = $languageLinks;
3895                 $this->mCategories = $categoryLinks;
3896                 $this->mContainsOldMagic = $containsOldMagic;
3897                 $this->mCacheTime = '';
3898                 $this->mVersion = MW_PARSER_VERSION;
3899                 $this->mTitleText = $titletext;
3900                 $this->mLinks = array();
3901                 $this->mTemplates = array();
3902                 $this->mImages = array();
3903                 $this->mExternalLinks = array();
3904         }
3905
3906         function getText()                   { return $this->mText; }
3907         function getLanguageLinks()          { return $this->mLanguageLinks; }
3908         function getCategoryLinks()          { return array_keys( $this->mCategories ); }
3909         function &getCategories()            { return $this->mCategories; }
3910         function getCacheTime()              { return $this->mCacheTime; }
3911         function getTitleText()              { return $this->mTitleText; }
3912         function &getLinks()                 { return $this->mLinks; }
3913         function &getTemplates()             { return $this->mTemplates; }
3914         function &getImages()                { return $this->mImages; }
3915         function &getExternalLinks()         { return $this->mExternalLinks; }
3916
3917         function containsOldMagic()          { return $this->mContainsOldMagic; }
3918         function setText( $text )            { return wfSetVar( $this->mText, $text ); }
3919         function setLanguageLinks( $ll )     { return wfSetVar( $this->mLanguageLinks, $ll ); }
3920         function setCategoryLinks( $cl )     { return wfSetVar( $this->mCategories, $cl ); }
3921         function setContainsOldMagic( $com ) { return wfSetVar( $this->mContainsOldMagic, $com ); }
3922         function setCacheTime( $t )          { return wfSetVar( $this->mCacheTime, $t ); }
3923         function setTitleText( $t )          { return wfSetVar ($this->mTitleText, $t); }
3924
3925         function addCategory( $c, $sort )    { $this->mCategories[$c] = $sort; }
3926         function addImage( $name )           { $this->mImages[$name] = 1; }
3927         function addLanguageLink( $t )       { $this->mLanguageLinks[] = $t; }
3928         function addExternalLink( $url )     { $this->mExternalLinks[$url] = 1; }
3929
3930         function addLink( $title, $id ) {
3931                 $ns = $title->getNamespace();
3932                 $dbk = $title->getDBkey();
3933                 if ( !isset( $this->mLinks[$ns] ) ) {
3934                         $this->mLinks[$ns] = array();
3935                 }
3936                 $this->mLinks[$ns][$dbk] = $id;
3937         }
3938
3939         function addTemplate( $title, $id ) {
3940                 $ns = $title->getNamespace();
3941                 $dbk = $title->getDBkey();
3942                 if ( !isset( $this->mTemplates[$ns] ) ) {
3943                         $this->mTemplates[$ns] = array();
3944                 }
3945                 $this->mTemplates[$ns][$dbk] = $id;
3946         }
3947
3948         /**
3949          * @deprecated
3950          */
3951         /*
3952         function merge( $other ) {
3953                 $this->mLanguageLinks = array_merge( $this->mLanguageLinks, $other->mLanguageLinks );
3954                 $this->mCategories = array_merge( $this->mCategories, $this->mLanguageLinks );
3955                 $this->mContainsOldMagic = $this->mContainsOldMagic || $other->mContainsOldMagic;
3956         }*/
3957
3958         /**
3959          * Return true if this cached output object predates the global or
3960          * per-article cache invalidation timestamps, or if it comes from
3961          * an incompatible older version.
3962          *
3963          * @param string $touched the affected article's last touched timestamp
3964          * @return bool
3965          * @access public
3966          */
3967         function expired( $touched ) {
3968                 global $wgCacheEpoch;
3969                 return $this->getCacheTime() == -1 || // parser says it's uncacheable
3970                        $this->getCacheTime() < $touched ||
3971                        $this->getCacheTime() <= $wgCacheEpoch ||
3972                        !isset( $this->mVersion ) ||
3973                        version_compare( $this->mVersion, MW_PARSER_VERSION, "lt" );
3974         }
3975 }
3976
3977 /**
3978  * Set options of the Parser
3979  * @todo document
3980  * @package MediaWiki
3981  */
3982 class ParserOptions
3983 {
3984         # All variables are private
3985         var $mUseTeX;                    # Use texvc to expand <math> tags
3986         var $mUseDynamicDates;           # Use DateFormatter to format dates
3987         var $mInterwikiMagic;            # Interlanguage links are removed and returned in an array
3988         var $mAllowExternalImages;       # Allow external images inline
3989         var $mAllowExternalImagesFrom;   # If not, any exception?
3990         var $mSkin;                      # Reference to the preferred skin
3991         var $mDateFormat;                # Date format index
3992         var $mEditSection;               # Create "edit section" links
3993         var $mNumberHeadings;            # Automatically number headings
3994         var $mAllowSpecialInclusion;     # Allow inclusion of special pages
3995         var $mTidy;                      # Ask for tidy cleanup
3996
3997         function getUseTeX()                        { return $this->mUseTeX; }
3998         function getUseDynamicDates()               { return $this->mUseDynamicDates; }
3999         function getInterwikiMagic()                { return $this->mInterwikiMagic; }
4000         function getAllowExternalImages()           { return $this->mAllowExternalImages; }
4001         function getAllowExternalImagesFrom()       { return $this->mAllowExternalImagesFrom; }
4002         function &getSkin()                         { return $this->mSkin; }
4003         function getDateFormat()                    { return $this->mDateFormat; }
4004         function getEditSection()                   { return $this->mEditSection; }
4005         function getNumberHeadings()                { return $this->mNumberHeadings; }
4006         function getAllowSpecialInclusion()         { return $this->mAllowSpecialInclusion; }
4007         function getTidy()                          { return $this->mTidy; }
4008
4009         function setUseTeX( $x )                    { return wfSetVar( $this->mUseTeX, $x ); }
4010         function setUseDynamicDates( $x )           { return wfSetVar( $this->mUseDynamicDates, $x ); }
4011         function setInterwikiMagic( $x )            { return wfSetVar( $this->mInterwikiMagic, $x ); }
4012         function setAllowExternalImages( $x )       { return wfSetVar( $this->mAllowExternalImages, $x ); }
4013         function setAllowExternalImagesFrom( $x )   { return wfSetVar( $this->mAllowExternalImagesFrom, $x ); }
4014         function setDateFormat( $x )                { return wfSetVar( $this->mDateFormat, $x ); }
4015         function setEditSection( $x )               { return wfSetVar( $this->mEditSection, $x ); }
4016         function setNumberHeadings( $x )            { return wfSetVar( $this->mNumberHeadings, $x ); }
4017         function setAllowSpecialInclusion( $x )     { return wfSetVar( $this->mAllowSpecialInclusion, $x ); }
4018         function setTidy( $x )                      { return wfSetVar( $this->mTidy, $x); }
4019         function setSkin( &$x ) { $this->mSkin =& $x; }
4020
4021         function ParserOptions() {
4022                 global $wgUser;
4023                 $this->initialiseFromUser( $wgUser );
4024         }
4025
4026         /**
4027          * Get parser options
4028          * @static
4029          */
4030         function newFromUser( &$user ) {
4031                 $popts = new ParserOptions;
4032                 $popts->initialiseFromUser( $user );
4033                 return $popts;
4034         }
4035
4036         /** Get user options */
4037         function initialiseFromUser( &$userInput ) {
4038                 global $wgUseTeX, $wgUseDynamicDates, $wgInterwikiMagic, $wgAllowExternalImages;
4039                 global $wgAllowExternalImagesFrom, $wgAllowSpecialInclusion;
4040                 $fname = 'ParserOptions::initialiseFromUser';
4041                 wfProfileIn( $fname );
4042                 if ( !$userInput ) {
4043                         $user = new User;
4044                         $user->setLoaded( true );
4045                 } else {
4046                         $user =& $userInput;
4047                 }
4048
4049                 $this->mUseTeX = $wgUseTeX;
4050                 $this->mUseDynamicDates = $wgUseDynamicDates;
4051                 $this->mInterwikiMagic = $wgInterwikiMagic;
4052                 $this->mAllowExternalImages = $wgAllowExternalImages;
4053                 $this->mAllowExternalImagesFrom = $wgAllowExternalImagesFrom;
4054                 wfProfileIn( $fname.'-skin' );
4055                 $this->mSkin =& $user->getSkin();
4056                 wfProfileOut( $fname.'-skin' );
4057                 $this->mDateFormat = $user->getOption( 'date' );
4058                 $this->mEditSection = true;
4059                 $this->mNumberHeadings = $user->getOption( 'numberheadings' );
4060                 $this->mAllowSpecialInclusion = $wgAllowSpecialInclusion;
4061                 $this->mTidy = false;
4062                 wfProfileOut( $fname );
4063         }
4064 }
4065
4066 /**
4067  * Callback function used by Parser::replaceLinkHolders()
4068  * to substitute link placeholders.
4069  */
4070 function &wfOutputReplaceMatches( $matches ) {
4071         global $wgOutputReplace;
4072         return $wgOutputReplace[$matches[1]];
4073 }
4074
4075 /**
4076  * Return the total number of articles
4077  */
4078 function wfNumberOfArticles() {
4079         global $wgNumberOfArticles;
4080
4081         wfLoadSiteStats();
4082         return $wgNumberOfArticles;
4083 }
4084
4085 /**
4086  * Return the number of files
4087  */
4088 function wfNumberOfFiles() {
4089         $fname = 'Parser::wfNumberOfFiles';
4090
4091         wfProfileIn( $fname );
4092         $dbr =& wfGetDB( DB_SLAVE );
4093         $res = $dbr->selectField('image', 'COUNT(*)', array(), $fname );
4094         wfProfileOut( $fname );
4095
4096         return $res;
4097 }
4098
4099 /**
4100  * Get various statistics from the database
4101  * @access private
4102  */
4103 function wfLoadSiteStats() {
4104         global $wgNumberOfArticles, $wgTotalViews, $wgTotalEdits;
4105         $fname = 'wfLoadSiteStats';
4106
4107         if ( -1 != $wgNumberOfArticles ) return;
4108         $dbr =& wfGetDB( DB_SLAVE );
4109         $s = $dbr->selectRow( 'site_stats',
4110                 array( 'ss_total_views', 'ss_total_edits', 'ss_good_articles' ),
4111                 array( 'ss_row_id' => 1 ), $fname
4112         );
4113
4114         if ( $s === false ) {
4115                 return;
4116         } else {
4117                 $wgTotalViews = $s->ss_total_views;
4118                 $wgTotalEdits = $s->ss_total_edits;
4119                 $wgNumberOfArticles = $s->ss_good_articles;
4120         }
4121 }
4122
4123 /**
4124  * Escape html tags
4125  * Basically replacing " > and < with HTML entities ( &quot;, &gt;, &lt;)
4126  *
4127  * @param string $in Text that might contain HTML tags
4128  * @return string Escaped string
4129  */
4130 function wfEscapeHTMLTagsOnly( $in ) {
4131         return str_replace(
4132                 array( '"', '>', '<' ),
4133                 array( '&quot;', '&gt;', '&lt;' ),
4134                 $in );
4135 }
4136
4137 ?>