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', '1.6.1' );
  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          * @private
  94          */
  95         # Persistent:
  96         var $mTagHooks, $mFunctionHooks;
  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          * @public
 120          */
 121         function Parser() {
 122                 $this->mTagHooks = array();
 123                 $this->mFunctionHooks = array();
 124                 $this->clearState();
 125         }
 126
 127         /**
 128          * Clear Parser state
 129          *
 130          * @private
 131          */
 132         function clearState() {
 133                 $this->mOutput = new ParserOutput;
 134                 $this->mAutonumber = 0;
 135                 $this->mLastSection = '';
 136                 $this->mDTopen = false;
 137                 $this->mVariables = false;
 138                 $this->mIncludeCount = array();
 139                 $this->mStripState = array();
 140                 $this->mArgStack = array();
 141                 $this->mInPre = false;
 142                 $this->mInterwikiLinkHolders = array(
 143                         'texts' => array(),
 144                         'titles' => array()
 145                 );
 146                 $this->mLinkHolders = array(
 147                         'namespaces' => array(),
 148                         'dbkeys' => array(),
 149                         'queries' => array(),
 150                         'texts' => array(),
 151                         'titles' => array()
 152                 );
 153                 $this->mRevisionId = null;
 154
 155                 /**
 156                  * Prefix for temporary replacement strings for the multipass parser.
 157                  * \x07 should never appear in input as it's disallowed in XML.
 158                  * Using it at the front also gives us a little extra robustness
 159                  * since it shouldn't match when butted up against identifier-like
 160                  * string constructs.
 161                  */
 162                 $this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 163
 164                 # Clear these on every parse, bug 4549
 165                 $this->mTemplates = array();
 166                 $this->mTemplatePath = array();
 167
 168                 $this->mShowToc = true;
 169                 $this->mForceTocPosition = false;
 170
 171                 wfRunHooks( 'ParserClearState', array( &$this ) );
 172         }
 173
 174         /**
 175          * Accessor for mUniqPrefix.
 176          *
 177          * @public
 178          */
 179         function UniqPrefix() {
 180                 return $this->mUniqPrefix;
 181         }
 182
 183         /**
 184          * Convert wikitext to HTML
 185          * Do not call this function recursively.
 186          *
 187          * @private
 188          * @param string $text Text we want to parse
 189          * @param Title &$title A title object
 190          * @param array $options
 191          * @param boolean $linestart
 192          * @param boolean $clearState
 193          * @param int $revid number to pass in {{REVISIONID}}
 194          * @return ParserOutput a ParserOutput
 195          */
 196         function parse( $text, &$title, $options, $linestart = true, $clearState = true, $revid = null ) {
 197                 /**
 198                  * First pass--just handle <nowiki> sections, pass the rest off
 199                  * to internalParse() which does all the real work.
 200                  */
 201
 202                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
 203                 $fname = 'Parser::parse';
 204                 wfProfileIn( $fname );
 205
 206                 if ( $clearState ) {
 207                         $this->clearState();
 208                 }
 209
 210                 $this->mOptions = $options;
 211                 $this->mTitle =& $title;
 212                 $this->mRevisionId = $revid;
 213                 $this->mOutputType = OT_HTML;
 214
 215                 //$text = $this->strip( $text, $this->mStripState );
 216                 // VOODOO MAGIC FIX! Sometimes the above segfaults in PHP5.
 217                 $x =& $this->mStripState;
 218
 219                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$x ) );
 220                 $text = $this->strip( $text, $x );
 221                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$x ) );
 222
 223                 # Hook to suspend the parser in this state
 224                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$x ) ) ) {
 225                         wfProfileOut( $fname );
 226                         return $text ;
 227                 }
 228
 229                 $text = $this->internalParse( $text );
 230
 231                 $text = $this->unstrip( $text, $this->mStripState );
 232
 233                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 234                 $fixtags = array(
 235                         # french spaces, last one Guillemet-left
 236                         # only if there is something before the space
 237                         '/(.) (?=\\?|:|;|!|\\302\\273)/' => '\\1&nbsp;\\2',
 238                         # french spaces, Guillemet-right
 239                         '/(\\302\\253) /' => '\\1&nbsp;',
 240                         '/<center *>(.*)<\\/center *>/i' => '<div class="center">\\1</div>',
 241                 );
 242                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 243
 244                 # only once and last
 245                 $text = $this->doBlockLevels( $text, $linestart );
 246
 247                 $this->replaceLinkHolders( $text );
 248
 249                 # the position of the parserConvert() call should not be changed. it
 250                 # assumes that the links are all replaced and the only thing left
 251                 # is the <nowiki> mark.
 252                 # Side-effects: this calls $this->mOutput->setTitleText()
 253                 $text = $wgContLang->parserConvert( $text, $this );
 254
 255                 $text = $this->unstripNoWiki( $text, $this->mStripState );
 256
 257                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 258
 259                 $text = Sanitizer::normalizeCharReferences( $text );
 260
 261                 if (($wgUseTidy and $this->mOptions->mTidy) or $wgAlwaysUseTidy) {
 262                         $text = Parser::tidy($text);
 263                 } else {
 264                         # attempt to sanitize at least some nesting problems
 265                         # (bug #2702 and quite a few others)
 266                         $tidyregs = array(
 267                                 # ''Something [http://www.cool.com cool''] -->
 268                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 269                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 270                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 271                                 # fix up an anchor inside another anchor, only
 272                                 # at least for a single single nested link (bug 3695)
 273                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 274                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 275                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 276                                 # contains a div, so fix it up here; replace
 277                                 # div with escaped text
 278                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 279                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 280                                 # remove empty italic or bold tag pairs, some
 281                                 # introduced by rules above
 282                                 '/<([bi])><\/\\1>/' => ''
 283                         );
 284
 285                         $text = preg_replace(
 286                                 array_keys( $tidyregs ),
 287                                 array_values( $tidyregs ),
 288                                 $text );
 289                 }
 290
 291                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 292
 293                 $this->mOutput->setText( $text );
 294                 wfProfileOut( $fname );
 295
 296                 return $this->mOutput;
 297         }
 298
 299         /**
 300          * Get a random string
 301          *
 302          * @private
 303          * @static
 304          */
 305         function getRandomString() {
 306                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 307         }
 308
 309         function &getTitle() { return $this->mTitle; }
 310         function getOptions() { return $this->mOptions; }
 311
 312         /**
 313          * Replaces all occurrences of <$tag>content</$tag> in the text
 314          * with a random marker and returns the new text. the output parameter
 315          * $content will be an associative array filled with data on the form
 316          * $unique_marker => content.
 317          *
 318          * If $content is already set, the additional entries will be appended
 319          * If $tag is set to STRIP_COMMENTS, the function will extract
 320          * <!-- HTML comments -->
 321          *
 322          * @private
 323          * @static
 324          */
 325         function extractTagsAndParams($tag, $text, &$content, &$tags, &$params, $uniq_prefix = ''){
 326                 $rnd = $uniq_prefix . '-' . $tag . Parser::getRandomString();
 327                 if ( !$content ) {
 328                         $content = array( );
 329                 }
 330                 $n = 1;
 331                 $stripped = '';
 332
 333                 if ( !$tags ) {
 334                         $tags = array( );
 335                 }
 336
 337                 if ( !$params ) {
 338                         $params = array( );
 339                 }
 340
 341                 if( $tag == STRIP_COMMENTS ) {
 342                         $start = '/<!--()/';
 343                         $end   = '/-->/';
 344                 } else {
 345                         $start = "/<$tag(\\s+[^>]*|\\s*\/?)>/i";
 346                         $end   = "/<\\/$tag\\s*>/i";
 347                 }
 348
 349                 while ( '' != $text ) {
 350                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 351                         $stripped .= $p[0];
 352                         if( count( $p ) < 3 ) {
 353                                 break;
 354                         }
 355                         $attributes = $p[1];
 356                         $inside     = $p[2];
 357
 358                         // If $attributes ends with '/', we have an empty element tag, <tag />
 359                         if( $tag != STRIP_COMMENTS && substr( $attributes, -1 ) == '/' ) {
 360                                 $attributes = substr( $attributes, 0, -1);
 361                                 $empty = '/';
 362                         } else {
 363                                 $empty = '';
 364                         }
 365
 366                         $marker = $rnd . sprintf('%08X', $n++);
 367                         $stripped .= $marker;
 368
 369                         $tags[$marker] = "<$tag$attributes$empty>";
 370                         $params[$marker] = Sanitizer::decodeTagAttributes( $attributes );
 371
 372                         if ( $empty === '/' ) {
 373                                 // Empty element tag, <tag />
 374                                 $content[$marker] = null;
 375                                 $text = $inside;
 376                         } else {
 377                                 $q = preg_split( $end, $inside, 2 );
 378                                 $content[$marker] = $q[0];
 379                                 if( count( $q ) < 2 ) {
 380                                         # No end tag -- let it run out to the end of the text.
 381                                         break;
 382                                 } else {
 383                                         $text = $q[1];
 384                                 }
 385                         }
 386                 }
 387                 return $stripped;
 388         }
 389
 390         /**
 391          * Wrapper function for extractTagsAndParams
 392          * for cases where $tags and $params isn't needed
 393          * i.e. where tags will never have params, like <nowiki>
 394          *
 395          * @private
 396          * @static
 397          */
 398         function extractTags( $tag, $text, &$content, $uniq_prefix = '' ) {
 399                 $dummy_tags = array();
 400                 $dummy_params = array();
 401
 402                 return Parser::extractTagsAndParams( $tag, $text, $content,
 403                         $dummy_tags, $dummy_params, $uniq_prefix );
 404         }
 405
 406         /**
 407          * Strips and renders nowiki, pre, math, hiero
 408          * If $render is set, performs necessary rendering operations on plugins
 409          * Returns the text, and fills an array with data needed in unstrip()
 410          * If the $state is already a valid strip state, it adds to the state
 411          *
 412          * @param bool $stripcomments when set, HTML comments <!-- like this -->
 413          *  will be stripped in addition to other tags. This is important
 414          *  for section editing, where these comments cause confusion when
 415          *  counting the sections in the wikisource
 416          *
 417          * @private
 418          */
 419         function strip( $text, &$state, $stripcomments = false ) {
 420                 $render = ($this->mOutputType == OT_HTML);
 421                 $html_content = array();
 422                 $nowiki_content = array();
 423                 $math_content = array();
 424                 $pre_content = array();
 425                 $comment_content = array();
 426                 $ext_content = array();
 427                 $ext_tags = array();
 428                 $ext_params = array();
 429                 $gallery_content = array();
 430
 431                 # Replace any instances of the placeholders
 432                 $uniq_prefix = $this->mUniqPrefix;
 433                 #$text = str_replace( $uniq_prefix, wfHtmlEscapeFirst( $uniq_prefix ), $text );
 434
 435                 # html
 436                 global $wgRawHtml;
 437                 if( $wgRawHtml ) {
 438                         $text = Parser::extractTags('html', $text, $html_content, $uniq_prefix);
 439                         foreach( $html_content as $marker => $content ) {
 440                                 if ($render ) {
 441                                         # Raw and unchecked for validity.
 442                                         $html_content[$marker] = $content;
 443                                 } else {
 444                                         $html_content[$marker] = '<html>'.$content.'</html>';
 445                                 }
 446                         }
 447                 }
 448
 449                 # nowiki
 450                 $text = Parser::extractTags('nowiki', $text, $nowiki_content, $uniq_prefix);
 451                 foreach( $nowiki_content as $marker => $content ) {
 452                         if( $render ){
 453                                 $nowiki_content[$marker] = wfEscapeHTMLTagsOnly( $content );
 454                         } else {
 455                                 $nowiki_content[$marker] = '<nowiki>'.$content.'</nowiki>';
 456                         }
 457                 }
 458
 459                 # math
 460                 if( $this->mOptions->getUseTeX() ) {
 461                         $text = Parser::extractTags('math', $text, $math_content, $uniq_prefix);
 462                         foreach( $math_content as $marker => $content ){
 463                                 if( $render ) {
 464                                         $math_content[$marker] = renderMath( $content );
 465                                 } else {
 466                                         $math_content[$marker] = '<math>'.$content.'</math>';
 467                                 }
 468                         }
 469                 }
 470
 471                 # pre
 472                 $text = Parser::extractTags('pre', $text, $pre_content, $uniq_prefix);
 473                 foreach( $pre_content as $marker => $content ){
 474                         if( $render ){
 475                                 $pre_content[$marker] = '<pre>' . wfEscapeHTMLTagsOnly( $content ) . '</pre>';
 476                         } else {
 477                                 $pre_content[$marker] = '<pre>'.$content.'</pre>';
 478                         }
 479                 }
 480
 481                 # gallery
 482                 $text = Parser::extractTags('gallery', $text, $gallery_content, $uniq_prefix);
 483                 foreach( $gallery_content as $marker => $content ) {
 484                         require_once( 'ImageGallery.php' );
 485                         if ( $render ) {
 486                                 $gallery_content[$marker] = $this->renderImageGallery( $content );
 487                         } else {
 488                                 $gallery_content[$marker] = '<gallery>'.$content.'</gallery>';
 489                         }
 490                 }
 491
 492                 # Comments
 493                 $text = Parser::extractTags(STRIP_COMMENTS, $text, $comment_content, $uniq_prefix);
 494                 foreach( $comment_content as $marker => $content ){
 495                         $comment_content[$marker] = '<!--'.$content.'-->';
 496                 }
 497
 498                 # Extensions
 499                 foreach ( $this->mTagHooks as $tag => $callback ) {
 500                         $ext_content[$tag] = array();
 501                         $text = Parser::extractTagsAndParams( $tag, $text, $ext_content[$tag],
 502                                 $ext_tags[$tag], $ext_params[$tag], $uniq_prefix );
 503                         foreach( $ext_content[$tag] as $marker => $content ) {
 504                                 $full_tag = $ext_tags[$tag][$marker];
 505                                 $params = $ext_params[$tag][$marker];
 506                                 if ( $render )
 507                                         $ext_content[$tag][$marker] = call_user_func_array( $callback, array( $content, $params, &$this ) );
 508                                 else {
 509                                         if ( is_null( $content ) ) {
 510                                                 // Empty element tag
 511                                                 $ext_content[$tag][$marker] = $full_tag;
 512                                         } else {
 513                                                 $ext_content[$tag][$marker] = "$full_tag$content</$tag>";
 514                                         }
 515                                 }
 516                         }
 517                 }
 518
 519                 # Unstrip comments unless explicitly told otherwise.
 520                 # (The comments are always stripped prior to this point, so as to
 521                 # not invoke any extension tags / parser hooks contained within
 522                 # a comment.)
 523                 if ( !$stripcomments ) {
 524                         $tempstate = array( 'comment' => $comment_content );
 525                         $text = $this->unstrip( $text, $tempstate );
 526                         $comment_content = array();
 527                 }
 528
 529                 # Merge state with the pre-existing state, if there is one
 530                 if ( $state ) {
 531                         $state['html'] = $state['html'] + $html_content;
 532                         $state['nowiki'] = $state['nowiki'] + $nowiki_content;
 533                         $state['math'] = $state['math'] + $math_content;
 534                         $state['pre'] = $state['pre'] + $pre_content;
 535                         $state['gallery'] = $state['gallery'] + $gallery_content;
 536                         $state['comment'] = $state['comment'] + $comment_content;
 537
 538                         foreach( $ext_content as $tag => $array ) {
 539                                 if ( array_key_exists( $tag, $state ) ) {
 540                                         $state[$tag] = $state[$tag] + $array;
 541                                 }
 542                         }
 543                 } else {
 544                         $state = array(
 545                           'html' => $html_content,
 546                           'nowiki' => $nowiki_content,
 547                           'math' => $math_content,
 548                           'pre' => $pre_content,
 549                           'gallery' => $gallery_content,
 550                           'comment' => $comment_content,
 551                         ) + $ext_content;
 552                 }
 553                 return $text;
 554         }
 555
 556         /**
 557          * restores pre, math, and hiero removed by strip()
 558          *
 559          * always call unstripNoWiki() after this one
 560          * @private
 561          */
 562         function unstrip( $text, &$state ) {
 563                 if ( !is_array( $state ) ) {
 564                         return $text;
 565                 }
 566
 567                 # Must expand in reverse order, otherwise nested tags will be corrupted
 568                 foreach( array_reverse( $state, true ) as $tag => $contentDict ) {
 569                         if( $tag != 'nowiki' && $tag != 'html' ) {
 570                                 foreach( array_reverse( $contentDict, true ) as $uniq => $content ) {
 571                                         $text = str_replace( $uniq, $content, $text );
 572                                 }
 573                         }
 574                 }
 575
 576                 return $text;
 577         }
 578
 579         /**
 580          * always call this after unstrip() to preserve the order
 581          *
 582          * @private
 583          */
 584         function unstripNoWiki( $text, &$state ) {
 585                 if ( !is_array( $state ) ) {
 586                         return $text;
 587                 }
 588
 589                 # Must expand in reverse order, otherwise nested tags will be corrupted
 590                 for ( $content = end($state['nowiki']); $content !== false; $content = prev( $state['nowiki'] ) ) {
 591                         $text = str_replace( key( $state['nowiki'] ), $content, $text );
 592                 }
 593
 594                 global $wgRawHtml;
 595                 if ($wgRawHtml) {
 596                         for ( $content = end($state['html']); $content !== false; $content = prev( $state['html'] ) ) {
 597                                 $text = str_replace( key( $state['html'] ), $content, $text );
 598                         }
 599                 }
 600
 601                 return $text;
 602         }
 603
 604         /**
 605          * Add an item to the strip state
 606          * Returns the unique tag which must be inserted into the stripped text
 607          * The tag will be replaced with the original text in unstrip()
 608          *
 609          * @private
 610          */
 611         function insertStripItem( $text, &$state ) {
 612                 $rnd = $this->mUniqPrefix . '-item' . Parser::getRandomString();
 613                 if ( !$state ) {
 614                         $state = array(
 615                           'html' => array(),
 616                           'nowiki' => array(),
 617                           'math' => array(),
 618                           'pre' => array(),
 619                           'comment' => array(),
 620                           'gallery' => array(),
 621                         );
 622                 }
 623                 $state['item'][$rnd] = $text;
 624                 return $rnd;
 625         }
 626
 627         /**
 628          * Interface with html tidy, used if $wgUseTidy = true.
 629          * If tidy isn't able to correct the markup, the original will be
 630          * returned in all its glory with a warning comment appended.
 631          *
 632          * Either the external tidy program or the in-process tidy extension
 633          * will be used depending on availability. Override the default
 634          * $wgTidyInternal setting to disable the internal if it's not working.
 635          *
 636          * @param string $text Hideous HTML input
 637          * @return string Corrected HTML output
 638          * @public
 639          * @static
 640          */
 641         function tidy( $text ) {
 642                 global $wgTidyInternal;
 643                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 644 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 645 '<head><title>test</title></head><body>'.$text.'</body></html>';
 646                 if( $wgTidyInternal ) {
 647                         $correctedtext = Parser::internalTidy( $wrappedtext );
 648                 } else {
 649                         $correctedtext = Parser::externalTidy( $wrappedtext );
 650                 }
 651                 if( is_null( $correctedtext ) ) {
 652                         wfDebug( "Tidy error detected!\n" );
 653                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 654                 }
 655                 return $correctedtext;
 656         }
 657
 658         /**
 659          * Spawn an external HTML tidy process and get corrected markup back from it.
 660          *
 661          * @private
 662          * @static
 663          */
 664         function externalTidy( $text ) {
 665                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 666                 $fname = 'Parser::externalTidy';
 667                 wfProfileIn( $fname );
 668
 669                 $cleansource = '';
 670                 $opts = ' -utf8';
 671
 672                 $descriptorspec = array(
 673                         0 => array('pipe', 'r'),
 674                         1 => array('pipe', 'w'),
 675                         2 => array('file', '/dev/null', 'a')
 676                 );
 677                 $pipes = array();
 678                 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 679                 if (is_resource($process)) {
 680                         // Theoretically, this style of communication could cause a deadlock
 681                         // here. If the stdout buffer fills up, then writes to stdin could
 682                         // block. This doesn't appear to happen with tidy, because tidy only
 683                         // writes to stdout after it's finished reading from stdin. Search
 684                         // for tidyParseStdin and tidySaveStdout in console/tidy.c
 685                         fwrite($pipes[0], $text);
 686                         fclose($pipes[0]);
 687                         while (!feof($pipes[1])) {
 688                                 $cleansource .= fgets($pipes[1], 1024);
 689                         }
 690                         fclose($pipes[1]);
 691                         proc_close($process);
 692                 }
 693
 694                 wfProfileOut( $fname );
 695
 696                 if( $cleansource == '' && $text != '') {
 697                         // Some kind of error happened, so we couldn't get the corrected text.
 698                         // Just give up; we'll use the source text and append a warning.
 699                         return null;
 700                 } else {
 701                         return $cleansource;
 702                 }
 703         }
 704
 705         /**
 706          * Use the HTML tidy PECL extension to use the tidy library in-process,
 707          * saving the overhead of spawning a new process. Currently written to
 708          * the PHP 4.3.x version of the extension, may not work on PHP 5.
 709          *
 710          * 'pear install tidy' should be able to compile the extension module.
 711          *
 712          * @private
 713          * @static
 714          */
 715         function internalTidy( $text ) {
 716                 global $wgTidyConf;
 717                 $fname = 'Parser::internalTidy';
 718                 wfProfileIn( $fname );
 719
 720                 tidy_load_config( $wgTidyConf );
 721                 tidy_set_encoding( 'utf8' );
 722                 tidy_parse_string( $text );
 723                 tidy_clean_repair();
 724                 if( tidy_get_status() == 2 ) {
 725                         // 2 is magic number for fatal error
 726                         // http://www.php.net/manual/en/function.tidy-get-status.php
 727                         $cleansource = null;
 728                 } else {
 729                         $cleansource = tidy_get_output();
 730                 }
 731                 wfProfileOut( $fname );
 732                 return $cleansource;
 733         }
 734
 735         /**
 736          * parse the wiki syntax used to render tables
 737          *
 738          * @private
 739          */
 740         function doTableStuff ( $t ) {
 741                 $fname = 'Parser::doTableStuff';
 742                 wfProfileIn( $fname );
 743
 744                 $t = explode ( "\n" , $t ) ;
 745                 $td = array () ; # Is currently a td tag open?
 746                 $ltd = array () ; # Was it TD or TH?
 747                 $tr = array () ; # Is currently a tr tag open?
 748                 $ltr = array () ; # tr attributes
 749                 $has_opened_tr = array(); # Did this table open a <tr> element?
 750                 $indent_level = 0; # indent level of the table
 751                 foreach ( $t AS $k => $x )
 752                 {
 753                         $x = trim ( $x ) ;
 754                         $fc = substr ( $x , 0 , 1 ) ;
 755                         if ( preg_match( '/^(:*)\{\|(.*)$/', $x, $matches ) ) {
 756                                 $indent_level = strlen( $matches[1] );
 757
 758                                 $attributes = $this->unstripForHTML( $matches[2] );
 759
 760                                 $t[$k] = str_repeat( '<dl><dd>', $indent_level ) .
 761                                         '<table' . Sanitizer::fixTagAttributes ( $attributes, 'table' ) . '>' ;
 762                                 array_push ( $td , false ) ;
 763                                 array_push ( $ltd , '' ) ;
 764                                 array_push ( $tr , false ) ;
 765                                 array_push ( $ltr , '' ) ;
 766                                 array_push ( $has_opened_tr, false );
 767                         }
 768                         else if ( count ( $td ) == 0 ) { } # Don't do any of the following
 769                         else if ( '|}' == substr ( $x , 0 , 2 ) ) {
 770                                 $z = "</table>" . substr ( $x , 2);
 771                                 $l = array_pop ( $ltd ) ;
 772                                 if ( !array_pop ( $has_opened_tr ) ) $z = "<tr><td></td></tr>" . $z ;
 773                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 774                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 775                                 array_pop ( $ltr ) ;
 776                                 $t[$k] = $z . str_repeat( '</dd></dl>', $indent_level );
 777                         }
 778                         else if ( '|-' == substr ( $x , 0 , 2 ) ) { # Allows for |---------------
 779                                 $x = substr ( $x , 1 ) ;
 780                                 while ( $x != '' && substr ( $x , 0 , 1 ) == '-' ) $x = substr ( $x , 1 ) ;
 781                                 $z = '' ;
 782                                 $l = array_pop ( $ltd ) ;
 783                                 array_pop ( $has_opened_tr );
 784                                 array_push ( $has_opened_tr , true ) ;
 785                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 786                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 787                                 array_pop ( $ltr ) ;
 788                                 $t[$k] = $z ;
 789                                 array_push ( $tr , false ) ;
 790                                 array_push ( $td , false ) ;
 791                                 array_push ( $ltd , '' ) ;
 792                                 $attributes = $this->unstripForHTML( $x );
 793                                 array_push ( $ltr , Sanitizer::fixTagAttributes ( $attributes, 'tr' ) ) ;
 794                         }
 795                         else if ( '|' == $fc || '!' == $fc || '|+' == substr ( $x , 0 , 2 ) ) { # Caption
 796                                 # $x is a table row
 797                                 if ( '|+' == substr ( $x , 0 , 2 ) ) {
 798                                         $fc = '+' ;
 799                                         $x = substr ( $x , 1 ) ;
 800                                 }
 801                                 $after = substr ( $x , 1 ) ;
 802                                 if ( $fc == '!' ) $after = str_replace ( '!!' , '||' , $after ) ;
 803
 804                                 // Split up multiple cells on the same line.
 805                                 // FIXME: This can result in improper nesting of tags processed
 806                                 // by earlier parser steps, but should avoid splitting up eg
 807                                 // attribute values containing literal "||".
 808                                 $after = wfExplodeMarkup( '||', $after );
 809
 810                                 $t[$k] = '' ;
 811
 812                                 # Loop through each table cell
 813                                 foreach ( $after AS $theline )
 814                                 {
 815                                         $z = '' ;
 816                                         if ( $fc != '+' )
 817                                         {
 818                                                 $tra = array_pop ( $ltr ) ;
 819                                                 if ( !array_pop ( $tr ) ) $z = '<tr'.$tra.">\n" ;
 820                                                 array_push ( $tr , true ) ;
 821                                                 array_push ( $ltr , '' ) ;
 822                                                 array_pop ( $has_opened_tr );
 823                                                 array_push ( $has_opened_tr , true ) ;
 824                                         }
 825
 826                                         $l = array_pop ( $ltd ) ;
 827                                         if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 828                                         if ( $fc == '|' ) $l = 'td' ;
 829                                         else if ( $fc == '!' ) $l = 'th' ;
 830                                         else if ( $fc == '+' ) $l = 'caption' ;
 831                                         else $l = '' ;
 832                                         array_push ( $ltd , $l ) ;
 833
 834                                         # Cell parameters
 835                                         $y = explode ( '|' , $theline , 2 ) ;
 836                                         # Note that a '|' inside an invalid link should not
 837                                         # be mistaken as delimiting cell parameters
 838                                         if ( strpos( $y[0], '[[' ) !== false ) {
 839                                                 $y = array ($theline);
 840                                         }
 841                                         if ( count ( $y ) == 1 )
 842                                                 $y = "{$z}<{$l}>{$y[0]}" ;
 843                                         else {
 844                                                 $attributes = $this->unstripForHTML( $y[0] );
 845                                                 $y = "{$z}<{$l}".Sanitizer::fixTagAttributes($attributes, $l).">{$y[1]}" ;
 846                                         }
 847                                         $t[$k] .= $y ;
 848                                         array_push ( $td , true ) ;
 849                                 }
 850                         }
 851                 }
 852
 853                 # Closing open td, tr && table
 854                 while ( count ( $td ) > 0 )
 855                 {
 856                         $l = array_pop ( $ltd ) ;
 857                         if ( array_pop ( $td ) ) $t[] = '</td>' ;
 858                         if ( array_pop ( $tr ) ) $t[] = '</tr>' ;
 859                         if ( !array_pop ( $has_opened_tr ) ) $t[] = "<tr><td></td></tr>" ;
 860                         $t[] = '</table>' ;
 861                 }
 862
 863                 $t = implode ( "\n" , $t ) ;
 864                 # special case: don't return empty table
 865                 if($t == "<table>\n<tr><td></td></tr>\n</table>")
 866                         $t = '';
 867                 wfProfileOut( $fname );
 868                 return $t ;
 869         }
 870
 871         /**
 872          * Helper function for parse() that transforms wiki markup into
 873          * HTML. Only called for $mOutputType == OT_HTML.
 874          *
 875          * @private
 876          */
 877         function internalParse( $text ) {
 878                 $args = array();
 879                 $isMain = true;
 880                 $fname = 'Parser::internalParse';
 881                 wfProfileIn( $fname );
 882
 883                 # Remove <noinclude> tags and <includeonly> sections
 884                 $text = strtr( $text, array( '<onlyinclude>' => '' , '</onlyinclude>' => '' ) );
 885                 $text = strtr( $text, array( '<noinclude>' => '', '</noinclude>' => '') );
 886                 $text = preg_replace( '/<includeonly>.*?<\/includeonly>/s', '', $text );
 887
 888                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ) );
 889
 890                 $text = $this->replaceVariables( $text, $args );
 891
 892                 // Tables need to come after variable replacement for things to work
 893                 // properly; putting them before other transformations should keep
 894                 // exciting things like link expansions from showing up in surprising
 895                 // places.
 896                 $text = $this->doTableStuff( $text );
 897
 898                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 899
 900                 $text = $this->stripToc( $text );
 901                 $text = $this->doHeadings( $text );
 902                 if($this->mOptions->getUseDynamicDates()) {
 903                         $df =& DateFormatter::getInstance();
 904                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 905                 }
 906                 $text = $this->doAllQuotes( $text );
 907                 $text = $this->replaceInternalLinks( $text );
 908                 $text = $this->replaceExternalLinks( $text );
 909
 910                 # replaceInternalLinks may sometimes leave behind
 911                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 912                 $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
 913
 914                 $text = $this->doMagicLinks( $text );
 915                 $text = $this->formatHeadings( $text, $isMain );
 916
 917                 wfProfileOut( $fname );
 918                 return $text;
 919         }
 920
 921         /**
 922          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 923          * magic external links.
 924          *
 925          * @private
 926          */
 927         function &doMagicLinks( &$text ) {
 928                 $text = $this->magicISBN( $text );
 929                 $text = $this->magicRFC( $text, 'RFC ', 'rfcurl' );
 930                 $text = $this->magicRFC( $text, 'PMID ', 'pubmedurl' );
 931                 return $text;
 932         }
 933
 934         /**
 935          * Parse headers and return html
 936          *
 937          * @private
 938          */
 939         function doHeadings( $text ) {
 940                 $fname = 'Parser::doHeadings';
 941                 wfProfileIn( $fname );
 942                 for ( $i = 6; $i >= 1; --$i ) {
 943                         $h = str_repeat( '=', $i );
 944                         $text = preg_replace( "/^{$h}(.+){$h}(\\s|$)/m",
 945                           "<h{$i}>\\1</h{$i}>\\2", $text );
 946                 }
 947                 wfProfileOut( $fname );
 948                 return $text;
 949         }
 950
 951         /**
 952          * Replace single quotes with HTML markup
 953          * @private
 954          * @return string the altered text
 955          */
 956         function doAllQuotes( $text ) {
 957                 $fname = 'Parser::doAllQuotes';
 958                 wfProfileIn( $fname );
 959                 $outtext = '';
 960                 $lines = explode( "\n", $text );
 961                 foreach ( $lines as $line ) {
 962                         $outtext .= $this->doQuotes ( $line ) . "\n";
 963                 }
 964                 $outtext = substr($outtext, 0,-1);
 965                 wfProfileOut( $fname );
 966                 return $outtext;
 967         }
 968
 969         /**
 970          * Helper function for doAllQuotes()
 971          * @private
 972          */
 973         function doQuotes( $text ) {
 974                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
 975                 if ( count( $arr ) == 1 )
 976                         return $text;
 977                 else
 978                 {
 979                         # First, do some preliminary work. This may shift some apostrophes from
 980                         # being mark-up to being text. It also counts the number of occurrences
 981                         # of bold and italics mark-ups.
 982                         $i = 0;
 983                         $numbold = 0;
 984                         $numitalics = 0;
 985                         foreach ( $arr as $r )
 986                         {
 987                                 if ( ( $i % 2 ) == 1 )
 988                                 {
 989                                         # If there are ever four apostrophes, assume the first is supposed to
 990                                         # be text, and the remaining three constitute mark-up for bold text.
 991                                         if ( strlen( $arr[$i] ) == 4 )
 992                                         {
 993                                                 $arr[$i-1] .= "'";
 994                                                 $arr[$i] = "'''";
 995                                         }
 996                                         # If there are more than 5 apostrophes in a row, assume they're all
 997                                         # text except for the last 5.
 998                                         else if ( strlen( $arr[$i] ) > 5 )
 999                                         {
1000                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1001                                                 $arr[$i] = "'''''";
1002                                         }
1003                                         # Count the number of occurrences of bold and italics mark-ups.
1004                                         # We are not counting sequences of five apostrophes.
1005                                         if ( strlen( $arr[$i] ) == 2 ) $numitalics++;  else
1006                                         if ( strlen( $arr[$i] ) == 3 ) $numbold++;     else
1007                                         if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
1008                                 }
1009                                 $i++;
1010                         }
1011
1012                         # If there is an odd number of both bold and italics, it is likely
1013                         # that one of the bold ones was meant to be an apostrophe followed
1014                         # by italics. Which one we cannot know for certain, but it is more
1015                         # likely to be one that has a single-letter word before it.
1016                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1017                         {
1018                                 $i = 0;
1019                                 $firstsingleletterword = -1;
1020                                 $firstmultiletterword = -1;
1021                                 $firstspace = -1;
1022                                 foreach ( $arr as $r )
1023                                 {
1024                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1025                                         {
1026                                                 $x1 = substr ($arr[$i-1], -1);
1027                                                 $x2 = substr ($arr[$i-1], -2, 1);
1028                                                 if ($x1 == ' ') {
1029                                                         if ($firstspace == -1) $firstspace = $i;
1030                                                 } else if ($x2 == ' ') {
1031                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1032                                                 } else {
1033                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1034                                                 }
1035                                         }
1036                                         $i++;
1037                                 }
1038
1039                                 # If there is a single-letter word, use it!
1040                                 if ($firstsingleletterword > -1)
1041                                 {
1042                                         $arr [ $firstsingleletterword ] = "''";
1043                                         $arr [ $firstsingleletterword-1 ] .= "'";
1044                                 }
1045                                 # If not, but there's a multi-letter word, use that one.
1046                                 else if ($firstmultiletterword > -1)
1047                                 {
1048                                         $arr [ $firstmultiletterword ] = "''";
1049                                         $arr [ $firstmultiletterword-1 ] .= "'";
1050                                 }
1051                                 # ... otherwise use the first one that has neither.
1052                                 # (notice that it is possible for all three to be -1 if, for example,
1053                                 # there is only one pentuple-apostrophe in the line)
1054                                 else if ($firstspace > -1)
1055                                 {
1056                                         $arr [ $firstspace ] = "''";
1057                                         $arr [ $firstspace-1 ] .= "'";
1058                                 }
1059                         }
1060
1061                         # Now let's actually convert our apostrophic mush to HTML!
1062                         $output = '';
1063                         $buffer = '';
1064                         $state = '';
1065                         $i = 0;
1066                         foreach ($arr as $r)
1067                         {
1068                                 if (($i % 2) == 0)
1069                                 {
1070                                         if ($state == 'both')
1071                                                 $buffer .= $r;
1072                                         else
1073                                                 $output .= $r;
1074                                 }
1075                                 else
1076                                 {
1077                                         if (strlen ($r) == 2)
1078                                         {
1079                                                 if ($state == 'i')
1080                                                 { $output .= '</i>'; $state = ''; }
1081                                                 else if ($state == 'bi')
1082                                                 { $output .= '</i>'; $state = 'b'; }
1083                                                 else if ($state == 'ib')
1084                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1085                                                 else if ($state == 'both')
1086                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1087                                                 else # $state can be 'b' or ''
1088                                                 { $output .= '<i>'; $state .= 'i'; }
1089                                         }
1090                                         else if (strlen ($r) == 3)
1091                                         {
1092                                                 if ($state == 'b')
1093                                                 { $output .= '</b>'; $state = ''; }
1094                                                 else if ($state == 'bi')
1095                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1096                                                 else if ($state == 'ib')
1097                                                 { $output .= '</b>'; $state = 'i'; }
1098                                                 else if ($state == 'both')
1099                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1100                                                 else # $state can be 'i' or ''
1101                                                 { $output .= '<b>'; $state .= 'b'; }
1102                                         }
1103                                         else if (strlen ($r) == 5)
1104                                         {
1105                                                 if ($state == 'b')
1106                                                 { $output .= '</b><i>'; $state = 'i'; }
1107                                                 else if ($state == 'i')
1108                                                 { $output .= '</i><b>'; $state = 'b'; }
1109                                                 else if ($state == 'bi')
1110                                                 { $output .= '</i></b>'; $state = ''; }
1111                                                 else if ($state == 'ib')
1112                                                 { $output .= '</b></i>'; $state = ''; }
1113                                                 else if ($state == 'both')
1114                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1115                                                 else # ($state == '')
1116                                                 { $buffer = ''; $state = 'both'; }
1117                                         }
1118                                 }
1119                                 $i++;
1120                         }
1121                         # Now close all remaining tags.  Notice that the order is important.
1122                         if ($state == 'b' || $state == 'ib')
1123                                 $output .= '</b>';
1124                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
1125                                 $output .= '</i>';
1126                         if ($state == 'bi')
1127                                 $output .= '</b>';
1128                         if ($state == 'both')
1129                                 $output .= '<b><i>'.$buffer.'</i></b>';
1130                         return $output;
1131                 }
1132         }
1133
1134         /**
1135          * Replace external links
1136          *
1137          * Note: this is all very hackish and the order of execution matters a lot.
1138          * Make sure to run maintenance/parserTests.php if you change this code.
1139          *
1140          * @private
1141          */
1142         function replaceExternalLinks( $text ) {
1143                 global $wgContLang;
1144                 $fname = 'Parser::replaceExternalLinks';
1145                 wfProfileIn( $fname );
1146
1147                 $sk =& $this->mOptions->getSkin();
1148
1149                 $bits = preg_split( EXT_LINK_BRACKETED, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1150
1151                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
1152
1153                 $i = 0;
1154                 while ( $i<count( $bits ) ) {
1155                         $url = $bits[$i++];
1156                         $protocol = $bits[$i++];
1157                         $text = $bits[$i++];
1158                         $trail = $bits[$i++];
1159
1160                         # The characters '<' and '>' (which were escaped by
1161                         # removeHTMLtags()) should not be included in
1162                         # URLs, per RFC 2396.
1163                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1164                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1165                                 $url = substr($url, 0, $m2[0][1]);
1166                         }
1167
1168                         # If the link text is an image URL, replace it with an <img> tag
1169                         # This happened by accident in the original parser, but some people used it extensively
1170                         $img = $this->maybeMakeExternalImage( $text );
1171                         if ( $img !== false ) {
1172                                 $text = $img;
1173                         }
1174
1175                         $dtrail = '';
1176
1177                         # Set linktype for CSS - if URL==text, link is essentially free
1178                         $linktype = ($text == $url) ? 'free' : 'text';
1179
1180                         # No link text, e.g. [http://domain.tld/some.link]
1181                         if ( $text == '' ) {
1182                                 # Autonumber if allowed. See bug #5918
1183                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1184                                         $text = '[' . ++$this->mAutonumber . ']';
1185                                         $linktype = 'autonumber';
1186                                 } else {
1187                                         # Otherwise just use the URL
1188                                         $text = htmlspecialchars( $url );
1189                                         $linktype = 'free';
1190                                 }
1191                         } else {
1192                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1193                                 # Check for trail
1194                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1195                         }
1196
1197                         $text = $wgContLang->markNoConversion($text);
1198
1199                         # Normalize any HTML entities in input. They will be
1200                         # re-escaped by makeExternalLink().
1201                         $url = Sanitizer::decodeCharReferences( $url );
1202
1203                         # Process the trail (i.e. everything after this link up until start of the next link),
1204                         # replacing any non-bracketed links
1205                         $trail = $this->replaceFreeExternalLinks( $trail );
1206
1207                         # Use the encoded URL
1208                         # This means that users can paste URLs directly into the text
1209                         # Funny characters like &ouml; aren't valid in URLs anyway
1210                         # This was changed in August 2004
1211                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype, $this->mTitle->getNamespace() ) . $dtrail . $trail;
1212
1213                         # Register link in the output object.
1214                         # Replace unnecessary URL escape codes with the referenced character
1215                         # This prevents spammers from hiding links from the filters
1216                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1217                         $this->mOutput->addExternalLink( $pasteurized );
1218                 }
1219
1220                 wfProfileOut( $fname );
1221                 return $s;
1222         }
1223
1224         /**
1225          * Replace anything that looks like a URL with a link
1226          * @private
1227          */
1228         function replaceFreeExternalLinks( $text ) {
1229                 global $wgContLang;
1230                 $fname = 'Parser::replaceFreeExternalLinks';
1231                 wfProfileIn( $fname );
1232
1233                 $bits = preg_split( '/(\b(?:' . wfUrlProtocols() . '))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1234                 $s = array_shift( $bits );
1235                 $i = 0;
1236
1237                 $sk =& $this->mOptions->getSkin();
1238
1239                 while ( $i < count( $bits ) ){
1240                         $protocol = $bits[$i++];
1241                         $remainder = $bits[$i++];
1242
1243                         if ( preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1244                                 # Found some characters after the protocol that look promising
1245                                 $url = $protocol . $m[1];
1246                                 $trail = $m[2];
1247
1248                                 # special case: handle urls as url args:
1249                                 # http://www.example.com/foo?=http://www.example.com/bar
1250                                 if(strlen($trail) == 0 &&
1251                                         isset($bits[$i]) &&
1252                                         preg_match('/^'. wfUrlProtocols() . '$/S', $bits[$i]) &&
1253                                         preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $bits[$i + 1], $m ))
1254                                 {
1255                                         # add protocol, arg
1256                                         $url .= $bits[$i] . $m[1]; # protocol, url as arg to previous link
1257                                         $i += 2;
1258                                         $trail = $m[2];
1259                                 }
1260
1261                                 # The characters '<' and '>' (which were escaped by
1262                                 # removeHTMLtags()) should not be included in
1263                                 # URLs, per RFC 2396.
1264                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1265                                         $trail = substr($url, $m2[0][1]) . $trail;
1266                                         $url = substr($url, 0, $m2[0][1]);
1267                                 }
1268
1269                                 # Move trailing punctuation to $trail
1270                                 $sep = ',;\.:!?';
1271                                 # If there is no left bracket, then consider right brackets fair game too
1272                                 if ( strpos( $url, '(' ) === false ) {
1273                                         $sep .= ')';
1274                                 }
1275
1276                                 $numSepChars = strspn( strrev( $url ), $sep );
1277                                 if ( $numSepChars ) {
1278                                         $trail = substr( $url, -$numSepChars ) . $trail;
1279                                         $url = substr( $url, 0, -$numSepChars );
1280                                 }
1281
1282                                 # Normalize any HTML entities in input. They will be
1283                                 # re-escaped by makeExternalLink() or maybeMakeExternalImage()
1284                                 $url = Sanitizer::decodeCharReferences( $url );
1285
1286                                 # Is this an external image?
1287                                 $text = $this->maybeMakeExternalImage( $url );
1288                                 if ( $text === false ) {
1289                                         # Not an image, make a link
1290                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free', $this->mTitle->getNamespace() );
1291                                         # Register it in the output object...
1292                                         # Replace unnecessary URL escape codes with their equivalent characters
1293                                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1294                                         $this->mOutput->addExternalLink( $pasteurized );
1295                                 }
1296                                 $s .= $text . $trail;
1297                         } else {
1298                                 $s .= $protocol . $remainder;
1299                         }
1300                 }
1301                 wfProfileOut( $fname );
1302                 return $s;
1303         }
1304
1305         /**
1306          * Replace unusual URL escape codes with their equivalent characters
1307          * @param string
1308          * @return string
1309          * @static
1310          * @fixme This can merge genuinely required bits in the path or query string,
1311          *        breaking legit URLs. A proper fix would treat the various parts of
1312          *        the URL differently; as a workaround, just use the output for
1313          *        statistical records, not for actual linking/output.
1314          */
1315         function replaceUnusualEscapes( $url ) {
1316                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1317                         array( 'Parser', 'replaceUnusualEscapesCallback' ), $url );
1318         }
1319
1320         /**
1321          * Callback function used in replaceUnusualEscapes().
1322          * Replaces unusual URL escape codes with their equivalent character
1323          * @static
1324          * @private
1325          */
1326         function replaceUnusualEscapesCallback( $matches ) {
1327                 $char = urldecode( $matches[0] );
1328                 $ord = ord( $char );
1329                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1330                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1331                         // No, shouldn't be escaped
1332                         return $char;
1333                 } else {
1334                         // Yes, leave it escaped
1335                         return $matches[0];
1336                 }
1337         }
1338
1339         /**
1340          * make an image if it's allowed, either through the global
1341          * option or through the exception
1342          * @private
1343          */
1344         function maybeMakeExternalImage( $url ) {
1345                 $sk =& $this->mOptions->getSkin();
1346                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1347                 $imagesexception = !empty($imagesfrom);
1348                 $text = false;
1349                 if ( $this->mOptions->getAllowExternalImages()
1350                      || ( $imagesexception && strpos( $url, $imagesfrom ) === 0 ) ) {
1351                         if ( preg_match( EXT_IMAGE_REGEX, $url ) ) {
1352                                 # Image found
1353                                 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
1354                         }
1355                 }
1356                 return $text;
1357         }
1358
1359         /**
1360          * Process [[ ]] wikilinks
1361          *
1362          * @private
1363          */
1364         function replaceInternalLinks( $s ) {
1365                 global $wgContLang;
1366                 static $fname = 'Parser::replaceInternalLinks' ;
1367
1368                 wfProfileIn( $fname );
1369
1370                 wfProfileIn( $fname.'-setup' );
1371                 static $tc = FALSE;
1372                 # the % is needed to support urlencoded titles as well
1373                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1374
1375                 $sk =& $this->mOptions->getSkin();
1376
1377                 #split the entire text string on occurences of [[
1378                 $a = explode( '[[', ' ' . $s );
1379                 #get the first element (all text up to first [[), and remove the space we added
1380                 $s = array_shift( $a );
1381                 $s = substr( $s, 1 );
1382
1383                 # Match a link having the form [[namespace:link|alternate]]trail
1384                 static $e1 = FALSE;
1385                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1386                 # Match cases where there is no "]]", which might still be images
1387                 static $e1_img = FALSE;
1388                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1389                 # Match the end of a line for a word that's not followed by whitespace,
1390                 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1391                 $e2 = wfMsgForContent( 'linkprefix' );
1392
1393                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1394
1395                 if( is_null( $this->mTitle ) ) {
1396                         wfDebugDieBacktrace( 'nooo' );
1397                 }
1398                 $nottalk = !$this->mTitle->isTalkPage();
1399
1400                 if ( $useLinkPrefixExtension ) {
1401                         if ( preg_match( $e2, $s, $m ) ) {
1402                                 $first_prefix = $m[2];
1403                         } else {
1404                                 $first_prefix = false;
1405                         }
1406                 } else {
1407                         $prefix = '';
1408                 }
1409
1410                 $selflink = $this->mTitle->getPrefixedText();
1411                 wfProfileOut( $fname.'-setup' );
1412
1413                 $checkVariantLink = sizeof($wgContLang->getVariants())>1;
1414                 $useSubpages = $this->areSubpagesAllowed();
1415
1416                 # Loop for each link
1417                 for ($k = 0; isset( $a[$k] ); $k++) {
1418                         $line = $a[$k];
1419                         if ( $useLinkPrefixExtension ) {
1420                                 wfProfileIn( $fname.'-prefixhandling' );
1421                                 if ( preg_match( $e2, $s, $m ) ) {
1422                                         $prefix = $m[2];
1423                                         $s = $m[1];
1424                                 } else {
1425                                         $prefix='';
1426                                 }
1427                                 # first link
1428                                 if($first_prefix) {
1429                                         $prefix = $first_prefix;
1430                                         $first_prefix = false;
1431                                 }
1432                                 wfProfileOut( $fname.'-prefixhandling' );
1433                         }
1434
1435                         $might_be_img = false;
1436
1437                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1438                                 $text = $m[2];
1439                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1440                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1441                                 # the real problem is with the $e1 regex
1442                                 # See bug 1300.
1443                                 #
1444                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1445                                 # and no image is in sight. See bug 2095.
1446                                 #
1447                                 if( $text !== '' &&
1448                                         preg_match( "/^\](.*)/s", $m[3], $n ) &&
1449                                         strpos($text, '[') !== false
1450                                 )
1451                                 {
1452                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1453                                         $m[3] = $n[1];
1454                                 }
1455                                 # fix up urlencoded title texts
1456                                 if(preg_match('/%/', $m[1] ))
1457                                         # Should anchors '#' also be rejected?
1458                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1459                                 $trail = $m[3];
1460                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1461                                 $might_be_img = true;
1462                                 $text = $m[2];
1463                                 if(preg_match('/%/', $m[1] )) $m[1] = urldecode($m[1]);
1464                                 $trail = "";
1465                         } else { # Invalid form; output directly
1466                                 $s .= $prefix . '[[' . $line ;
1467                                 continue;
1468                         }
1469
1470                         # Don't allow internal links to pages containing
1471                         # PROTO: where PROTO is a valid URL protocol; these
1472                         # should be external links.
1473                         if (preg_match('/^(\b(?:' . wfUrlProtocols() . '))/', $m[1])) {
1474                                 $s .= $prefix . '[[' . $line ;
1475                                 continue;
1476                         }
1477
1478                         # Make subpage if necessary
1479                         if( $useSubpages ) {
1480                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1481                         } else {
1482                                 $link = $m[1];
1483                         }
1484
1485                         $noforce = (substr($m[1], 0, 1) != ':');
1486                         if (!$noforce) {
1487                                 # Strip off leading ':'
1488                                 $link = substr($link, 1);
1489                         }
1490
1491                         $nt = Title::newFromText( $this->unstripNoWiki($link, $this->mStripState) );
1492                         if( !$nt ) {
1493                                 $s .= $prefix . '[[' . $line;
1494                                 continue;
1495                         }
1496
1497                         #check other language variants of the link
1498                         #if the article does not exist
1499                         if( $checkVariantLink
1500                             && $nt->getArticleID() == 0 ) {
1501                                 $wgContLang->findVariantLink($link, $nt);
1502                         }
1503
1504                         $ns = $nt->getNamespace();
1505                         $iw = $nt->getInterWiki();
1506
1507                         if ($might_be_img) { # if this is actually an invalid link
1508                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1509                                         $found = false;
1510                                         while (isset ($a[$k+1]) ) {
1511                                                 #look at the next 'line' to see if we can close it there
1512                                                 $spliced = array_splice( $a, $k + 1, 1 );
1513                                                 $next_line = array_shift( $spliced );
1514                                                 if( preg_match("/^(.*?]].*?)]](.*)$/sD", $next_line, $m) ) {
1515                                                 # the first ]] closes the inner link, the second the image
1516                                                         $found = true;
1517                                                         $text .= '[[' . $m[1];
1518                                                         $trail = $m[2];
1519                                                         break;
1520                                                 } elseif( preg_match("/^.*?]].*$/sD", $next_line, $m) ) {
1521                                                         #if there's exactly one ]] that's fine, we'll keep looking
1522                                                         $text .= '[[' . $m[0];
1523                                                 } else {
1524                                                         #if $next_line is invalid too, we need look no further
1525                                                         $text .= '[[' . $next_line;
1526                                                         break;
1527                                                 }
1528                                         }
1529                                         if ( !$found ) {
1530                                                 # we couldn't find the end of this imageLink, so output it raw
1531                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1532                                                 $text = $this->replaceInternalLinks($text);
1533                                                 $s .= $prefix . '[[' . $link . '|' . $text;
1534                                                 # note: no $trail, because without an end, there *is* no trail
1535                                                 continue;
1536                                         }
1537                                 } else { #it's not an image, so output it raw
1538                                         $s .= $prefix . '[[' . $link . '|' . $text;
1539                                         # note: no $trail, because without an end, there *is* no trail
1540                                         continue;
1541                                 }
1542                         }
1543
1544                         $wasblank = ( '' == $text );
1545                         if( $wasblank ) $text = $link;
1546
1547
1548                         # Link not escaped by : , create the various objects
1549                         if( $noforce ) {
1550
1551                                 # Interwikis
1552                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1553                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1554                                         $s = rtrim($s . "\n");
1555                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1556                                         continue;
1557                                 }
1558
1559                                 if ( $ns == NS_IMAGE ) {
1560                                         wfProfileIn( "$fname-image" );
1561                                         if ( !wfIsBadImage( $nt->getDBkey() ) ) {
1562                                                 # recursively parse links inside the image caption
1563                                                 # actually, this will parse them in any other parameters, too,
1564                                                 # but it might be hard to fix that, and it doesn't matter ATM
1565                                                 $text = $this->replaceExternalLinks($text);
1566                                                 $text = $this->replaceInternalLinks($text);
1567
1568                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1569                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text ) ) . $trail;
1570                                                 $this->mOutput->addImage( $nt->getDBkey() );
1571
1572                                                 wfProfileOut( "$fname-image" );
1573                                                 continue;
1574                                         }
1575                                         wfProfileOut( "$fname-image" );
1576
1577                                 }
1578
1579                                 if ( $ns == NS_CATEGORY ) {
1580                                         wfProfileIn( "$fname-category" );
1581                                         $s = rtrim($s . "\n"); # bug 87
1582
1583                                         if ( $wasblank ) {
1584                                                 if ( $this->mTitle->getNamespace() == NS_CATEGORY ) {
1585                                                         $sortkey = $this->mTitle->getText();
1586                                                 } else {
1587                                                         $sortkey = $this->mTitle->getPrefixedText();
1588                                                 }
1589                                         } else {
1590                                                 $sortkey = $text;
1591                                         }
1592                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1593                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1594                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1595
1596                                         /**
1597                                          * Strip the whitespace Category links produce, see bug 87
1598                                          * @todo We might want to use trim($tmp, "\n") here.
1599                                          */
1600                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1601
1602                                         wfProfileOut( "$fname-category" );
1603                                         continue;
1604                                 }
1605                         }
1606
1607                         if( ( $nt->getPrefixedText() === $selflink ) &&
1608                             ( $nt->getFragment() === '' ) ) {
1609                                 # Self-links are handled specially; generally de-link and change to bold.
1610                                 $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1611                                 continue;
1612                         }
1613
1614                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1615                         if( $ns == NS_MEDIA ) {
1616                                 $link = $sk->makeMediaLinkObj( $nt, $text );
1617                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1618                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1619                                 $this->mOutput->addImage( $nt->getDBkey() );
1620                                 continue;
1621                         } elseif( $ns == NS_SPECIAL ) {
1622                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1623                                 continue;
1624                         } elseif( $ns == NS_IMAGE ) {
1625                                 $img = Image::newFromTitle( $nt );
1626                                 if( $img->exists() ) {
1627                                         // Force a blue link if the file exists; may be a remote
1628                                         // upload on the shared repository, and we want to see its
1629                                         // auto-generated page.
1630                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1631                                         continue;
1632                                 }
1633                         }
1634                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1635                 }
1636                 wfProfileOut( $fname );
1637                 return $s;
1638         }
1639
1640         /**
1641          * Make a link placeholder. The text returned can be later resolved to a real link with
1642          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1643          * parsing of interwiki links, and secondly to allow all extistence checks and
1644          * article length checks (for stub links) to be bundled into a single query.
1645          *
1646          */
1647         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1648                 if ( ! is_object($nt) ) {
1649                         # Fail gracefully
1650                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1651                 } else {
1652                         # Separate the link trail from the rest of the link
1653                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1654
1655                         if ( $nt->isExternal() ) {
1656                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1657                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1658                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1659                         } else {
1660                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1661                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1662                                 $this->mLinkHolders['queries'][] = $query;
1663                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1664                                 $this->mLinkHolders['titles'][] = $nt;
1665
1666                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1667                         }
1668                 }
1669                 return $retVal;
1670         }
1671
1672         /**
1673          * Render a forced-blue link inline; protect against double expansion of
1674          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1675          * Since this little disaster has to split off the trail text to avoid
1676          * breaking URLs in the following text without breaking trails on the
1677          * wiki links, it's been made into a horrible function.
1678          *
1679          * @param Title $nt
1680          * @param string $text
1681          * @param string $query
1682          * @param string $trail
1683          * @param string $prefix
1684          * @return string HTML-wikitext mix oh yuck
1685          */
1686         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1687                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1688                 $sk =& $this->mOptions->getSkin();
1689                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1690                 return $this->armorLinks( $link ) . $trail;
1691         }
1692
1693         /**
1694          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1695          * going to go through further parsing steps before inline URL expansion.
1696          *
1697          * In particular this is important when using action=render, which causes
1698          * full URLs to be included.
1699          *
1700          * Oh man I hate our multi-layer parser!
1701          *
1702          * @param string more-or-less HTML
1703          * @return string less-or-more HTML with NOPARSE bits
1704          */
1705         function armorLinks( $text ) {
1706                 return preg_replace( "/\b(" . wfUrlProtocols() . ')/',
1707                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1708         }
1709
1710         /**
1711          * Return true if subpage links should be expanded on this page.
1712          * @return bool
1713          */
1714         function areSubpagesAllowed() {
1715                 # Some namespaces don't allow subpages
1716                 global $wgNamespacesWithSubpages;
1717                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1718         }
1719
1720         /**
1721          * Handle link to subpage if necessary
1722          * @param string $target the source of the link
1723          * @param string &$text the link text, modified as necessary
1724          * @return string the full name of the link
1725          * @private
1726          */
1727         function maybeDoSubpageLink($target, &$text) {
1728                 # Valid link forms:
1729                 # Foobar -- normal
1730                 # :Foobar -- override special treatment of prefix (images, language links)
1731                 # /Foobar -- convert to CurrentPage/Foobar
1732                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1733                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1734                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1735
1736                 $fname = 'Parser::maybeDoSubpageLink';
1737                 wfProfileIn( $fname );
1738                 $ret = $target; # default return value is no change
1739
1740                 # Some namespaces don't allow subpages,
1741                 # so only perform processing if subpages are allowed
1742                 if( $this->areSubpagesAllowed() ) {
1743                         # Look at the first character
1744                         if( $target != '' && $target{0} == '/' ) {
1745                                 # / at end means we don't want the slash to be shown
1746                                 if( substr( $target, -1, 1 ) == '/' ) {
1747                                         $target = substr( $target, 1, -1 );
1748                                         $noslash = $target;
1749                                 } else {
1750                                         $noslash = substr( $target, 1 );
1751                                 }
1752
1753                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash);
1754                                 if( '' === $text ) {
1755                                         $text = $target;
1756                                 } # this might be changed for ugliness reasons
1757                         } else {
1758                                 # check for .. subpage backlinks
1759                                 $dotdotcount = 0;
1760                                 $nodotdot = $target;
1761                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1762                                         ++$dotdotcount;
1763                                         $nodotdot = substr( $nodotdot, 3 );
1764                                 }
1765                                 if($dotdotcount > 0) {
1766                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1767                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1768                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1769                                                 # / at the end means don't show full path
1770                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1771                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1772                                                         if( '' === $text ) {
1773                                                                 $text = $nodotdot;
1774                                                         }
1775                                                 }
1776                                                 $nodotdot = trim( $nodotdot );
1777                                                 if( $nodotdot != '' ) {
1778                                                         $ret .= '/' . $nodotdot;
1779                                                 }
1780                                         }
1781                                 }
1782                         }
1783                 }
1784
1785                 wfProfileOut( $fname );
1786                 return $ret;
1787         }
1788
1789         /**#@+
1790          * Used by doBlockLevels()
1791          * @private
1792          */
1793         /* private */ function closeParagraph() {
1794                 $result = '';
1795                 if ( '' != $this->mLastSection ) {
1796                         $result = '</' . $this->mLastSection  . ">\n";
1797                 }
1798                 $this->mInPre = false;
1799                 $this->mLastSection = '';
1800                 return $result;
1801         }
1802         # getCommon() returns the length of the longest common substring
1803         # of both arguments, starting at the beginning of both.
1804         #
1805         /* private */ function getCommon( $st1, $st2 ) {
1806                 $fl = strlen( $st1 );
1807                 $shorter = strlen( $st2 );
1808                 if ( $fl < $shorter ) { $shorter = $fl; }
1809
1810                 for ( $i = 0; $i < $shorter; ++$i ) {
1811                         if ( $st1{$i} != $st2{$i} ) { break; }
1812                 }
1813                 return $i;
1814         }
1815         # These next three functions open, continue, and close the list
1816         # element appropriate to the prefix character passed into them.
1817         #
1818         /* private */ function openList( $char ) {
1819                 $result = $this->closeParagraph();
1820
1821                 if ( '*' == $char ) { $result .= '<ul><li>'; }
1822                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
1823                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
1824                 else if ( ';' == $char ) {
1825                         $result .= '<dl><dt>';
1826                         $this->mDTopen = true;
1827                 }
1828                 else { $result = '<!-- ERR 1 -->'; }
1829
1830                 return $result;
1831         }
1832
1833         /* private */ function nextItem( $char ) {
1834                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
1835                 else if ( ':' == $char || ';' == $char ) {
1836                         $close = '</dd>';
1837                         if ( $this->mDTopen ) { $close = '</dt>'; }
1838                         if ( ';' == $char ) {
1839                                 $this->mDTopen = true;
1840                                 return $close . '<dt>';
1841                         } else {
1842                                 $this->mDTopen = false;
1843                                 return $close . '<dd>';
1844                         }
1845                 }
1846                 return '<!-- ERR 2 -->';
1847         }
1848
1849         /* private */ function closeList( $char ) {
1850                 if ( '*' == $char ) { $text = '</li></ul>'; }
1851                 else if ( '#' == $char ) { $text = '</li></ol>'; }
1852                 else if ( ':' == $char ) {
1853                         if ( $this->mDTopen ) {
1854                                 $this->mDTopen = false;
1855                                 $text = '</dt></dl>';
1856                         } else {
1857                                 $text = '</dd></dl>';
1858                         }
1859                 }
1860                 else {  return '<!-- ERR 3 -->'; }
1861                 return $text."\n";
1862         }
1863         /**#@-*/
1864
1865         /**
1866          * Make lists from lines starting with ':', '*', '#', etc.
1867          *
1868          * @private
1869          * @return string the lists rendered as HTML
1870          */
1871         function doBlockLevels( $text, $linestart ) {
1872                 $fname = 'Parser::doBlockLevels';
1873                 wfProfileIn( $fname );
1874
1875                 # Parsing through the text line by line.  The main thing
1876                 # happening here is handling of block-level elements p, pre,
1877                 # and making lists from lines starting with * # : etc.
1878                 #
1879                 $textLines = explode( "\n", $text );
1880
1881                 $lastPrefix = $output = '';
1882                 $this->mDTopen = $inBlockElem = false;
1883                 $prefixLength = 0;
1884                 $paragraphStack = false;
1885
1886                 if ( !$linestart ) {
1887                         $output .= array_shift( $textLines );
1888                 }
1889                 foreach ( $textLines as $oLine ) {
1890                         $lastPrefixLength = strlen( $lastPrefix );
1891                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
1892                         $preOpenMatch = preg_match('/<pre/i', $oLine );
1893                         if ( !$this->mInPre ) {
1894                                 # Multiple prefixes may abut each other for nested lists.
1895                                 $prefixLength = strspn( $oLine, '*#:;' );
1896                                 $pref = substr( $oLine, 0, $prefixLength );
1897
1898                                 # eh?
1899                                 $pref2 = str_replace( ';', ':', $pref );
1900                                 $t = substr( $oLine, $prefixLength );
1901                                 $this->mInPre = !empty($preOpenMatch);
1902                         } else {
1903                                 # Don't interpret any other prefixes in preformatted text
1904                                 $prefixLength = 0;
1905                                 $pref = $pref2 = '';
1906                                 $t = $oLine;
1907                         }
1908
1909                         # List generation
1910                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
1911                                 # Same as the last item, so no need to deal with nesting or opening stuff
1912                                 $output .= $this->nextItem( substr( $pref, -1 ) );
1913                                 $paragraphStack = false;
1914
1915                                 if ( substr( $pref, -1 ) == ';') {
1916                                         # The one nasty exception: definition lists work like this:
1917                                         # ; title : definition text
1918                                         # So we check for : in the remainder text to split up the
1919                                         # title and definition, without b0rking links.
1920                                         $term = $t2 = '';
1921                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
1922                                                 $t = $t2;
1923                                                 $output .= $term . $this->nextItem( ':' );
1924                                         }
1925                                 }
1926                         } elseif( $prefixLength || $lastPrefixLength ) {
1927                                 # Either open or close a level...
1928                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
1929                                 $paragraphStack = false;
1930
1931                                 while( $commonPrefixLength < $lastPrefixLength ) {
1932                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
1933                                         --$lastPrefixLength;
1934                                 }
1935                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
1936                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
1937                                 }
1938                                 while ( $prefixLength > $commonPrefixLength ) {
1939                                         $char = substr( $pref, $commonPrefixLength, 1 );
1940                                         $output .= $this->openList( $char );
1941
1942                                         if ( ';' == $char ) {
1943                                                 # FIXME: This is dupe of code above
1944                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
1945                                                         $t = $t2;
1946                                                         $output .= $term . $this->nextItem( ':' );
1947                                                 }
1948                                         }
1949                                         ++$commonPrefixLength;
1950                                 }
1951                                 $lastPrefix = $pref2;
1952                         }
1953                         if( 0 == $prefixLength ) {
1954                                 wfProfileIn( "$fname-paragraph" );
1955                                 # No prefix (not in list)--go to paragraph mode
1956                                 // XXX: use a stack for nestable elements like span, table and div
1957                                 $openmatch = preg_match('/(<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
1958                                 $closematch = preg_match(
1959                                         '/(<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
1960                                         '<td|<th|<div|<\\/div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul)/iS', $t );
1961                                 if ( $openmatch or $closematch ) {
1962                                         $paragraphStack = false;
1963                                         # TODO bug 5718: paragraph closed
1964                                         $output .= $this->closeParagraph();
1965                                         if ( $preOpenMatch and !$preCloseMatch ) {
1966                                                 $this->mInPre = true;
1967                                         }
1968                                         if ( $closematch ) {
1969                                                 $inBlockElem = false;
1970                                         } else {
1971                                                 $inBlockElem = true;
1972                                         }
1973                                 } else if ( !$inBlockElem && !$this->mInPre ) {
1974                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
1975                                                 // pre
1976                                                 if ($this->mLastSection != 'pre') {
1977                                                         $paragraphStack = false;
1978                                                         $output .= $this->closeParagraph().'<pre>';
1979                                                         $this->mLastSection = 'pre';
1980                                                 }
1981                                                 $t = substr( $t, 1 );
1982                                         } else {
1983                                                 // paragraph
1984                                                 if ( '' == trim($t) ) {
1985                                                         if ( $paragraphStack ) {
1986                                                                 $output .= $paragraphStack.'<br />';
1987                                                                 $paragraphStack = false;
1988                                                                 $this->mLastSection = 'p';
1989                                                         } else {
1990                                                                 if ($this->mLastSection != 'p' ) {
1991                                                                         $output .= $this->closeParagraph();
1992                                                                         $this->mLastSection = '';
1993                                                                         $paragraphStack = '<p>';
1994                                                                 } else {
1995                                                                         $paragraphStack = '</p><p>';
1996                                                                 }
1997                                                         }
1998                                                 } else {
1999                                                         if ( $paragraphStack ) {
2000                                                                 $output .= $paragraphStack;
2001                                                                 $paragraphStack = false;
2002                                                                 $this->mLastSection = 'p';
2003                                                         } else if ($this->mLastSection != 'p') {
2004                                                                 $output .= $this->closeParagraph().'<p>';
2005                                                                 $this->mLastSection = 'p';
2006                                                         }
2007                                                 }
2008                                         }
2009                                 }
2010                                 wfProfileOut( "$fname-paragraph" );
2011                         }
2012                         // somewhere above we forget to get out of pre block (bug 785)
2013                         if($preCloseMatch && $this->mInPre) {
2014                                 $this->mInPre = false;
2015                         }
2016                         if ($paragraphStack === false) {
2017                                 $output .= $t."\n";
2018                         }
2019                 }
2020                 while ( $prefixLength ) {
2021                         $output .= $this->closeList( $pref2{$prefixLength-1} );
2022                         --$prefixLength;
2023                 }
2024                 if ( '' != $this->mLastSection ) {
2025                         $output .= '</' . $this->mLastSection . '>';
2026                         $this->mLastSection = '';
2027                 }
2028
2029                 wfProfileOut( $fname );
2030                 return $output;
2031         }
2032
2033         /**
2034          * Split up a string on ':', ignoring any occurences inside
2035          * <a>..</a> or <span>...</span>
2036          * @param string $str the string to split
2037          * @param string &$before set to everything before the ':'
2038          * @param string &$after set to everything after the ':'
2039          * return string the position of the ':', or false if none found
2040          */
2041         function findColonNoLinks($str, &$before, &$after) {
2042                 # I wonder if we should make this count all tags, not just <a>
2043                 # and <span>. That would prevent us from matching a ':' that
2044                 # comes in the middle of italics other such formatting....
2045                 # -- Wil
2046                 $fname = 'Parser::findColonNoLinks';
2047                 wfProfileIn( $fname );
2048                 $pos = 0;
2049                 do {
2050                         $colon = strpos($str, ':', $pos);
2051
2052                         if ($colon !== false) {
2053                                 $before = substr($str, 0, $colon);
2054                                 $after = substr($str, $colon + 1);
2055
2056                                 # Skip any ':' within <a> or <span> pairs
2057                                 $a = substr_count($before, '<a');
2058                                 $s = substr_count($before, '<span');
2059                                 $ca = substr_count($before, '</a>');
2060                                 $cs = substr_count($before, '</span>');
2061
2062                                 if ($a <= $ca and $s <= $cs) {
2063                                         # Tags are balanced before ':'; ok
2064                                         break;
2065                                 }
2066                                 $pos = $colon + 1;
2067                         }
2068                 } while ($colon !== false);
2069                 wfProfileOut( $fname );
2070                 return $colon;
2071         }
2072
2073         /**
2074          * Return value of a magic variable (like PAGENAME)
2075          *
2076          * @private
2077          */
2078         function getVariableValue( $index ) {
2079                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2080
2081                 /**
2082                  * Some of these require message or data lookups and can be
2083                  * expensive to check many times.
2084                  */
2085                 static $varCache = array();
2086                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$varCache ) ) )
2087                         if ( isset( $varCache[$index] ) )
2088                                 return $varCache[$index];
2089
2090                 $ts = time();
2091                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2092
2093                 switch ( $index ) {
2094                         case MAG_CURRENTMONTH:
2095                                 return $varCache[$index] = $wgContLang->formatNum( date( 'm', $ts ) );
2096                         case MAG_CURRENTMONTHNAME:
2097                                 return $varCache[$index] = $wgContLang->getMonthName( date( 'n', $ts ) );
2098                         case MAG_CURRENTMONTHNAMEGEN:
2099                                 return $varCache[$index] = $wgContLang->getMonthNameGen( date( 'n', $ts ) );
2100                         case MAG_CURRENTMONTHABBREV:
2101                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( date( 'n', $ts ) );
2102                         case MAG_CURRENTDAY:
2103                                 return $varCache[$index] = $wgContLang->formatNum( date( 'j', $ts ) );
2104                         case MAG_CURRENTDAY2:
2105                                 return $varCache[$index] = $wgContLang->formatNum( date( 'd', $ts ) );
2106                         case MAG_PAGENAME:
2107                                 return $this->mTitle->getText();
2108                         case MAG_PAGENAMEE:
2109                                 return $this->mTitle->getPartialURL();
2110                         case MAG_FULLPAGENAME:
2111                                 return $this->mTitle->getPrefixedText();
2112                         case MAG_FULLPAGENAMEE:
2113                                 return $this->mTitle->getPrefixedURL();
2114                         case MAG_SUBPAGENAME:
2115                                 return $this->mTitle->getSubpageText();
2116                         case MAG_SUBPAGENAMEE:
2117                                 return $this->mTitle->getSubpageUrlForm();
2118                         case MAG_BASEPAGENAME:
2119                                 return $this->mTitle->getBaseText();
2120                         case MAG_BASEPAGENAMEE:
2121                                 return wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2122                         case MAG_TALKPAGENAME:
2123                                 if( $this->mTitle->canTalk() ) {
2124                                         $talkPage = $this->mTitle->getTalkPage();
2125                                         return $talkPage->getPrefixedText();
2126                                 } else {
2127                                         return '';
2128                                 }
2129                         case MAG_TALKPAGENAMEE:
2130                                 if( $this->mTitle->canTalk() ) {
2131                                         $talkPage = $this->mTitle->getTalkPage();
2132                                         return $talkPage->getPrefixedUrl();
2133                                 } else {
2134                                         return '';
2135                                 }
2136                         case MAG_SUBJECTPAGENAME:
2137                                 $subjPage = $this->mTitle->getSubjectPage();
2138                                 return $subjPage->getPrefixedText();
2139                         case MAG_SUBJECTPAGENAMEE:
2140                                 $subjPage = $this->mTitle->getSubjectPage();
2141                                 return $subjPage->getPrefixedUrl();
2142                         case MAG_REVISIONID:
2143                                 return $this->mRevisionId;
2144                         case MAG_NAMESPACE:
2145                                 return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2146                         case MAG_NAMESPACEE:
2147                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2148                         case MAG_TALKSPACE:
2149                                 return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2150                         case MAG_TALKSPACEE:
2151                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2152                         case MAG_SUBJECTSPACE:
2153                                 return $this->mTitle->getSubjectNsText();
2154                         case MAG_SUBJECTSPACEE:
2155                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2156                         case MAG_CURRENTDAYNAME:
2157                                 return $varCache[$index] = $wgContLang->getWeekdayName( date( 'w', $ts ) + 1 );
2158                         case MAG_CURRENTYEAR:
2159                                 return $varCache[$index] = $wgContLang->formatNum( date( 'Y', $ts ), true );
2160                         case MAG_CURRENTTIME:
2161                                 return $varCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2162                         case MAG_CURRENTWEEK:
2163                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2164                                 // int to remove the padding
2165                                 return $varCache[$index] = $wgContLang->formatNum( (int)date( 'W', $ts ) );
2166                         case MAG_CURRENTDOW:
2167                                 return $varCache[$index] = $wgContLang->formatNum( date( 'w', $ts ) );
2168                         case MAG_NUMBEROFARTICLES:
2169                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfArticles() );
2170                         case MAG_NUMBEROFFILES:
2171                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfFiles() );
2172                         case MAG_NUMBEROFUSERS:
2173                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfUsers() );
2174                         case MAG_NUMBEROFPAGES:
2175                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfPages() );
2176                         case MAG_CURRENTTIMESTAMP:
2177                                 return $varCache[$index] = wfTimestampNow();
2178                         case MAG_CURRENTVERSION:
2179                                 global $wgVersion;
2180                                 return $wgVersion;
2181                         case MAG_SITENAME:
2182                                 return $wgSitename;
2183                         case MAG_SERVER:
2184                                 return $wgServer;
2185                         case MAG_SERVERNAME:
2186                                 return $wgServerName;
2187                         case MAG_SCRIPTPATH:
2188                                 return $wgScriptPath;
2189                         default:
2190                                 $ret = null;
2191                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$varCache, &$index, &$ret ) ) )
2192                                         return $ret;
2193                                 else
2194                                         return null;
2195                 }
2196         }
2197
2198         /**
2199          * initialise the magic variables (like CURRENTMONTHNAME)
2200          *
2201          * @private
2202          */
2203         function initialiseVariables() {
2204                 $fname = 'Parser::initialiseVariables';
2205                 wfProfileIn( $fname );
2206                 global $wgVariableIDs;
2207                 $this->mVariables = array();
2208                 foreach ( $wgVariableIDs as $id ) {
2209                         $mw =& MagicWord::get( $id );
2210                         $mw->addToArray( $this->mVariables, $id );
2211                 }
2212                 wfProfileOut( $fname );
2213         }
2214
2215         /**
2216          * parse any parentheses in format ((title|part|part))
2217          * and call callbacks to get a replacement text for any found piece
2218          *
2219          * @param string $text The text to parse
2220          * @param array $callbacks rules in form:
2221          *     '{' => array(                            # opening parentheses
2222          *                                      'end' => '}',   # closing parentheses
2223          *                                      'cb' => array(2 => callback,    # replacement callback to call if {{..}} is found
2224          *                                                                4 => callback         # replacement callback to call if {{{{..}}}} is found
2225          *                                                                )
2226          *                                      )
2227          * @private
2228          */
2229         function replace_callback ($text, $callbacks) {
2230                 $openingBraceStack = array();   # this array will hold a stack of parentheses which are not closed yet
2231                 $lastOpeningBrace = -1;                 # last not closed parentheses
2232
2233                 for ($i = 0; $i < strlen($text); $i++) {
2234                         # check for any opening brace
2235                         $rule = null;
2236                         $nextPos = -1;
2237                         foreach ($callbacks as $key => $value) {
2238                                 $pos = strpos ($text, $key, $i);
2239                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)) {
2240                                         $rule = $value;
2241                                         $nextPos = $pos;
2242                                 }
2243                         }
2244
2245                         if ($lastOpeningBrace >= 0) {
2246                                 $pos = strpos ($text, $openingBraceStack[$lastOpeningBrace]['braceEnd'], $i);
2247
2248                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)){
2249                                         $rule = null;
2250                                         $nextPos = $pos;
2251                                 }
2252
2253                                 $pos = strpos ($text, '|', $i);
2254
2255                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)){
2256                                         $rule = null;
2257                                         $nextPos = $pos;
2258                                 }
2259                         }
2260
2261                         if ($nextPos == -1)
2262                                 break;
2263
2264                         $i = $nextPos;
2265
2266                         # found openning brace, lets add it to parentheses stack
2267                         if (null != $rule) {
2268                                 $piece = array('brace' => $text[$i],
2269                                                            'braceEnd' => $rule['end'],
2270                                                            'count' => 1,
2271                                                            'title' => '',
2272                                                            'parts' => null);
2273
2274                                 # count openning brace characters
2275                                 while ($i+1 < strlen($text) && $text[$i+1] == $piece['brace']) {
2276                                         $piece['count']++;
2277                                         $i++;
2278                                 }
2279
2280                                 $piece['startAt'] = $i+1;
2281                                 $piece['partStart'] = $i+1;
2282
2283                                 # we need to add to stack only if openning brace count is enough for any given rule
2284                                 foreach ($rule['cb'] as $cnt => $fn) {
2285                                         if ($piece['count'] >= $cnt) {
2286                                                 $lastOpeningBrace ++;
2287                                                 $openingBraceStack[$lastOpeningBrace] = $piece;
2288                                                 break;
2289                                         }
2290                                 }
2291
2292                                 continue;
2293                         }
2294                         else if ($lastOpeningBrace >= 0) {
2295                                 # first check if it is a closing brace
2296                                 if ($openingBraceStack[$lastOpeningBrace]['braceEnd'] == $text[$i]) {
2297                                         # lets check if it is enough characters for closing brace
2298                                         $count = 1;
2299                                         while ($i+$count < strlen($text) && $text[$i+$count] == $text[$i])
2300                                                 $count++;
2301
2302                                         # if there are more closing parentheses than opening ones, we parse less
2303                                         if ($openingBraceStack[$lastOpeningBrace]['count'] < $count)
2304                                                 $count = $openingBraceStack[$lastOpeningBrace]['count'];
2305
2306                                         # check for maximum matching characters (if there are 5 closing characters, we will probably need only 3 - depending on the rules)
2307                                         $matchingCount = 0;
2308                                         $matchingCallback = null;
2309                                         foreach ($callbacks[$openingBraceStack[$lastOpeningBrace]['brace']]['cb'] as $cnt => $fn) {
2310                                                 if ($count >= $cnt && $matchingCount < $cnt) {
2311                                                         $matchingCount = $cnt;
2312                                                         $matchingCallback = $fn;
2313                                                 }
2314                                         }
2315
2316                                         if ($matchingCount == 0) {
2317                                                 $i += $count - 1;
2318                                                 continue;
2319                                         }
2320
2321                                         # lets set a title or last part (if '|' was found)
2322                                         if (null === $openingBraceStack[$lastOpeningBrace]['parts'])
2323                                                 $openingBraceStack[$lastOpeningBrace]['title'] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2324                                         else
2325                                                 $openingBraceStack[$lastOpeningBrace]['parts'][] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2326
2327                                         $pieceStart = $openingBraceStack[$lastOpeningBrace]['startAt'] - $matchingCount;
2328                                         $pieceEnd = $i + $matchingCount;
2329
2330                                         if( is_callable( $matchingCallback ) ) {
2331                                                 $cbArgs = array (
2332                                                                                  'text' => substr($text, $pieceStart, $pieceEnd - $pieceStart),
2333                                                                                  'title' => trim($openingBraceStack[$lastOpeningBrace]['title']),
2334                                                                                  'parts' => $openingBraceStack[$lastOpeningBrace]['parts'],
2335                                                                                  'lineStart' => (($pieceStart > 0) && ($text[$pieceStart-1] == '\n')),
2336                                                                                  );
2337                                                 # finally we can call a user callback and replace piece of text
2338                                                 $replaceWith = call_user_func( $matchingCallback, $cbArgs );
2339                                                 $text = substr($text, 0, $pieceStart) . $replaceWith . substr($text, $pieceEnd);
2340                                                 $i = $pieceStart + strlen($replaceWith) - 1;
2341                                         }
2342                                         else {
2343                                                 # null value for callback means that parentheses should be parsed, but not replaced
2344                                                 $i += $matchingCount - 1;
2345                                         }
2346
2347                                         # reset last openning parentheses, but keep it in case there are unused characters
2348                                         $piece = array('brace' => $openingBraceStack[$lastOpeningBrace]['brace'],
2349                                                                    'braceEnd' => $openingBraceStack[$lastOpeningBrace]['braceEnd'],
2350                                                                    'count' => $openingBraceStack[$lastOpeningBrace]['count'],
2351                                                                    'title' => '',
2352                                                                    'parts' => null,
2353                                                                    'startAt' => $openingBraceStack[$lastOpeningBrace]['startAt']);
2354                                         $openingBraceStack[$lastOpeningBrace--] = null;
2355
2356                                         if ($matchingCount < $piece['count']) {
2357                                                 $piece['count'] -= $matchingCount;
2358                                                 $piece['startAt'] -= $matchingCount;
2359                                                 $piece['partStart'] = $piece['startAt'];
2360                                                 # do we still qualify for any callback with remaining count?
2361                                                 foreach ($callbacks[$piece['brace']]['cb'] as $cnt => $fn) {
2362                                                         if ($piece['count'] >= $cnt) {
2363                                                                 $lastOpeningBrace ++;
2364                                                                 $openingBraceStack[$lastOpeningBrace] = $piece;
2365                                                                 break;
2366                                                         }
2367                                                 }
2368                                         }
2369                                         continue;
2370                                 }
2371
2372                                 # lets set a title if it is a first separator, or next part otherwise
2373                                 if ($text[$i] == '|') {
2374                                         if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2375                                                 $openingBraceStack[$lastOpeningBrace]['title'] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2376                                                 $openingBraceStack[$lastOpeningBrace]['parts'] = array();
2377                                         }
2378                                         else
2379                                                 $openingBraceStack[$lastOpeningBrace]['parts'][] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2380
2381                                         $openingBraceStack[$lastOpeningBrace]['partStart'] = $i + 1;
2382                                 }
2383                         }
2384                 }
2385
2386                 return $text;
2387         }
2388
2389         /**
2390          * Replace magic variables, templates, and template arguments
2391          * with the appropriate text. Templates are substituted recursively,
2392          * taking care to avoid infinite loops.
2393          *
2394          * Note that the substitution depends on value of $mOutputType:
2395          *  OT_WIKI: only {{subst:}} templates
2396          *  OT_MSG: only magic variables
2397          *  OT_HTML: all templates and magic variables
2398          *
2399          * @param string $tex The text to transform
2400          * @param array $args Key-value pairs representing template parameters to substitute
2401          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2402          * @private
2403          */
2404         function replaceVariables( $text, $args = array(), $argsOnly = false ) {
2405                 # Prevent too big inclusions
2406                 if( strlen( $text ) > MAX_INCLUDE_SIZE ) {
2407                         return $text;
2408                 }
2409
2410                 $fname = 'Parser::replaceVariables';
2411                 wfProfileIn( $fname );
2412
2413                 # This function is called recursively. To keep track of arguments we need a stack:
2414                 array_push( $this->mArgStack, $args );
2415
2416                 $braceCallbacks = array();
2417                 if ( !$argsOnly ) {
2418                         $braceCallbacks[2] = array( &$this, 'braceSubstitution' );
2419                 }
2420                 if ( $this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI ) {
2421                         $braceCallbacks[3] = array( &$this, 'argSubstitution' );
2422                 }
2423                 $callbacks = array();
2424                 $callbacks['{'] = array('end' => '}', 'cb' => $braceCallbacks);
2425                 $callbacks['['] = array('end' => ']', 'cb' => array(2=>null));
2426                 $text = $this->replace_callback ($text, $callbacks);
2427
2428                 array_pop( $this->mArgStack );
2429
2430                 wfProfileOut( $fname );
2431                 return $text;
2432         }
2433
2434         /**
2435          * Replace magic variables
2436          * @private
2437          */
2438         function variableSubstitution( $matches ) {
2439                 $fname = 'Parser::variableSubstitution';
2440                 $varname = $matches[1];
2441                 wfProfileIn( $fname );
2442                 if ( !$this->mVariables ) {
2443                         $this->initialiseVariables();
2444                 }
2445                 $skip = false;
2446                 if ( $this->mOutputType == OT_WIKI ) {
2447                         # Do only magic variables prefixed by SUBST
2448                         $mwSubst =& MagicWord::get( MAG_SUBST );
2449                         if (!$mwSubst->matchStartAndRemove( $varname ))
2450                                 $skip = true;
2451                         # Note that if we don't substitute the variable below,
2452                         # we don't remove the {{subst:}} magic word, in case
2453                         # it is a template rather than a magic variable.
2454                 }
2455                 if ( !$skip && array_key_exists( $varname, $this->mVariables ) ) {
2456                         $id = $this->mVariables[$varname];
2457                         $text = $this->getVariableValue( $id );
2458                         $this->mOutput->mContainsOldMagic = true;
2459                 } else {
2460                         $text = $matches[0];
2461                 }
2462                 wfProfileOut( $fname );
2463                 return $text;
2464         }
2465
2466         # Split template arguments
2467         function getTemplateArgs( $argsString ) {
2468                 if ( $argsString === '' ) {
2469                         return array();
2470                 }
2471
2472                 $args = explode( '|', substr( $argsString, 1 ) );
2473
2474                 # If any of the arguments contains a '[[' but no ']]', it needs to be
2475                 # merged with the next arg because the '|' character between belongs
2476                 # to the link syntax and not the template parameter syntax.
2477                 $argc = count($args);
2478
2479                 for ( $i = 0; $i < $argc-1; $i++ ) {
2480                         if ( substr_count ( $args[$i], '[[' ) != substr_count ( $args[$i], ']]' ) ) {
2481                                 $args[$i] .= '|'.$args[$i+1];
2482                                 array_splice($args, $i+1, 1);
2483                                 $i--;
2484                                 $argc--;
2485                         }
2486                 }
2487
2488                 return $args;
2489         }
2490
2491         /**
2492          * Return the text of a template, after recursively
2493          * replacing any variables or templates within the template.
2494          *
2495          * @param array $piece The parts of the template
2496          *  $piece['text']: matched text
2497          *  $piece['title']: the title, i.e. the part before the |
2498          *  $piece['parts']: the parameter array
2499          * @return string the text of the template
2500          * @private
2501          */
2502         function braceSubstitution( $piece ) {
2503                 global $wgContLang, $wgLang, $wgAllowDisplayTitle, $action;
2504                 $fname = 'Parser::braceSubstitution';
2505                 wfProfileIn( $fname );
2506
2507                 # Flags
2508                 $found = false;             # $text has been filled
2509                 $nowiki = false;            # wiki markup in $text should be escaped
2510                 $noparse = false;           # Unsafe HTML tags should not be stripped, etc.
2511                 $noargs = false;            # Don't replace triple-brace arguments in $text
2512                 $replaceHeadings = false;   # Make the edit section links go to the template not the article
2513                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2514                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2515
2516                 # Title object, where $text came from
2517                 $title = NULL;
2518
2519                 $linestart = '';
2520
2521                 # $part1 is the bit before the first |, and must contain only title characters
2522                 # $args is a list of arguments, starting from index 0, not including $part1
2523
2524                 $part1 = $piece['title'];
2525                 # If the third subpattern matched anything, it will start with |
2526
2527                 if (null == $piece['parts']) {
2528                         $replaceWith = $this->variableSubstitution (array ($piece['text'], $piece['title']));
2529                         if ($replaceWith != $piece['text']) {
2530                                 $text = $replaceWith;
2531                                 $found = true;
2532                                 $noparse = true;
2533                                 $noargs = true;
2534                         }
2535                 }
2536
2537                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2538                 $argc = count( $args );
2539
2540                 # SUBST
2541                 if ( !$found ) {
2542                         $mwSubst =& MagicWord::get( MAG_SUBST );
2543                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor ($this->mOutputType == OT_WIKI) ) {
2544                                 # One of two possibilities is true:
2545                                 # 1) Found SUBST but not in the PST phase
2546                                 # 2) Didn't find SUBST and in the PST phase
2547                                 # In either case, return without further processing
2548                                 $text = $piece['text'];
2549                                 $found = true;
2550                                 $noparse = true;
2551                                 $noargs = true;
2552                         }
2553                 }
2554
2555                 # MSG, MSGNW, INT and RAW
2556                 if ( !$found ) {
2557                         # Check for MSGNW:
2558                         $mwMsgnw =& MagicWord::get( MAG_MSGNW );
2559                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2560                                 $nowiki = true;
2561                         } else {
2562                                 # Remove obsolete MSG:
2563                                 $mwMsg =& MagicWord::get( MAG_MSG );
2564                                 $mwMsg->matchStartAndRemove( $part1 );
2565                         }
2566
2567                         # Check for RAW:
2568                         $mwRaw =& MagicWord::get( MAG_RAW );
2569                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2570                                 $forceRawInterwiki = true;
2571                         }
2572
2573                         # Check if it is an internal message
2574                         $mwInt =& MagicWord::get( MAG_INT );
2575                         if ( $mwInt->matchStartAndRemove( $part1 ) ) {
2576                                 if ( $this->incrementIncludeCount( 'int:'.$part1 ) ) {
2577                                         $text = $linestart . wfMsgReal( $part1, $args, true,
2578                                                 /* ui language */false, /* transform */false );
2579                                         $found = true;
2580                                 }
2581                         }
2582                 }
2583
2584                 # NS
2585                 if ( !$found ) {
2586                         # Check for NS: (namespace expansion)
2587                         $mwNs = MagicWord::get( MAG_NS );
2588                         if ( $mwNs->matchStartAndRemove( $part1 ) ) {
2589                                 if ( intval( $part1 ) || $part1 == "0" ) {
2590                                         $text = $linestart . $wgContLang->getNsText( intval( $part1 ) );
2591                                         $found = true;
2592                                 } else {
2593                                         $index = Namespace::getCanonicalIndex( strtolower( $part1 ) );
2594                                         if ( !is_null( $index ) ) {
2595                                                 $text = $linestart . $wgContLang->getNsText( $index );
2596                                                 $found = true;
2597                                         }
2598                                 }
2599                         }
2600                 }
2601
2602                 # URLENCODE
2603                 if( !$found ) {
2604                         $urlencode =& MagicWord::get( MAG_URLENCODE );
2605                         if( $urlencode->matchStartAndRemove( $part1 ) ) {
2606                                 $text = $linestart . urlencode( $part1 );
2607                                 $found = true;
2608                         }
2609                 }
2610
2611                 # LCFIRST, UCFIRST, LC and UC
2612                 if ( !$found ) {
2613                         $lcfirst =& MagicWord::get( MAG_LCFIRST );
2614                         $ucfirst =& MagicWord::get( MAG_UCFIRST );
2615                         $lc =& MagicWord::get( MAG_LC );
2616                         $uc =& MagicWord::get( MAG_UC );
2617                         if ( $lcfirst->matchStartAndRemove( $part1 ) ) {
2618                                 $text = $linestart . $wgContLang->lcfirst( $part1 );
2619                                 $found = true;
2620                         } else if ( $ucfirst->matchStartAndRemove( $part1 ) ) {
2621                                 $text = $linestart . $wgContLang->ucfirst( $part1 );
2622                                 $found = true;
2623                         } else if ( $lc->matchStartAndRemove( $part1 ) ) {
2624                                 $text = $linestart . $wgContLang->lc( $part1 );
2625                                 $found = true;
2626                         } else if ( $uc->matchStartAndRemove( $part1 ) ) {
2627                                  $text = $linestart . $wgContLang->uc( $part1 );
2628                                  $found = true;
2629                         }
2630                 }
2631
2632                 # LOCALURL and FULLURL
2633                 if ( !$found ) {
2634                         $mwLocal =& MagicWord::get( MAG_LOCALURL );
2635                         $mwLocalE =& MagicWord::get( MAG_LOCALURLE );
2636                         $mwFull =& MagicWord::get( MAG_FULLURL );
2637                         $mwFullE =& MagicWord::get( MAG_FULLURLE );
2638
2639
2640                         if ( $mwLocal->matchStartAndRemove( $part1 ) ) {
2641                                 $func = 'getLocalURL';
2642                         } elseif ( $mwLocalE->matchStartAndRemove( $part1 ) ) {
2643                                 $func = 'escapeLocalURL';
2644                         } elseif ( $mwFull->matchStartAndRemove( $part1 ) ) {
2645                                 $func = 'getFullURL';
2646                         } elseif ( $mwFullE->matchStartAndRemove( $part1 ) ) {
2647                                 $func = 'escapeFullURL';
2648                         } else {
2649                                 $func = false;
2650                         }
2651
2652                         if ( $func !== false ) {
2653                                 $title = Title::newFromText( $part1 );
2654                                 # Due to order of execution of a lot of bits, the values might be encoded
2655                                 # before arriving here; if that's true, then the title can't be created
2656                                 # and the variable will fail. If we can't get a decent title from the first
2657                                 # attempt, url-decode and try for a second.
2658                                 if( is_null( $title ) )
2659                                         $title = Title::newFromUrl( urldecode( $part1 ) );
2660                                 if ( !is_null( $title ) ) {
2661                                         if ( $argc > 0 ) {
2662                                                 $text = $linestart . $title->$func( $args[0] );
2663                                         } else {
2664                                                 $text = $linestart . $title->$func();
2665                                         }
2666                                         $found = true;
2667                                 }
2668                         }
2669                 }
2670
2671                 $lang = $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
2672                 # GRAMMAR
2673                 if ( !$found && $argc == 1 ) {
2674                         $mwGrammar =& MagicWord::get( MAG_GRAMMAR );
2675                         if ( $mwGrammar->matchStartAndRemove( $part1 ) ) {
2676                                 $text = $linestart . $lang->convertGrammar( $args[0], $part1 );
2677                                 $found = true;
2678                         }
2679                 }
2680
2681                 # PLURAL
2682                 if ( !$found && $argc >= 2 ) {
2683                         $mwPluralForm =& MagicWord::get( MAG_PLURAL );
2684                         if ( $mwPluralForm->matchStartAndRemove( $part1 ) ) {
2685                                 if ($argc==2) {$args[2]=$args[1];}
2686                                 $text = $linestart . $lang->convertPlural( $part1, $args[0], $args[1], $args[2]);
2687                                 $found = true;
2688                         }
2689                 }
2690
2691                 # DISPLAYTITLE
2692                 if ( !$found && $argc == 1 && $wgAllowDisplayTitle ) {
2693                         $mwDT =& MagicWord::get( MAG_DISPLAYTITLE );
2694                         if ( $mwDT->matchStartAndRemove( $part1 ) ) {
2695
2696                                 # Set title in parser output object
2697                                 $param = $args[0];
2698                                 $parserOptions = new ParserOptions;
2699                                 $local_parser = new Parser ();
2700                                 $t2 = $local_parser->parse ( $param, $this->mTitle, $parserOptions, false );
2701                                 $this->mOutput->mHTMLtitle = $t2->GetText();
2702
2703                                 # Add subtitle
2704                                 $t = $this->mTitle->getPrefixedText();
2705                                 $this->mOutput->mSubtitle .= wfMsg('displaytitle', $t);
2706                                 $text = "" ;
2707                                 $found = true ;
2708                         }
2709                 }
2710
2711                 # NUMBEROFPAGES, NUMBEROFUSERS, NUMBEROFARTICLES, and NUMBEROFFILES
2712                 if( !$found ) {
2713                         $mwWordsToCheck = array( MAG_NUMBEROFPAGES => 'wfNumberOfPages',
2714                                                                          MAG_NUMBEROFUSERS => 'wfNumberOfUsers',
2715                                                                          MAG_NUMBEROFARTICLES => 'wfNumberOfArticles',
2716                                                                          MAG_NUMBEROFFILES => 'wfNumberOfFiles' );
2717                         foreach( $mwWordsToCheck as $word => $func ) {
2718                                 $mwCurrentWord =& MagicWord::get( $word );
2719                                 if( $mwCurrentWord->matchStartAndRemove( $part1 ) ) {
2720                                         $mwRawSuffix =& MagicWord::get( MAG_RAWSUFFIX );
2721                                         if( $mwRawSuffix->match( $args[0] ) ) {
2722                                                 # Raw and unformatted
2723                                                 $text = $linestart . call_user_func( $func );
2724                                         } else {
2725                                                 # Formatted according to the content default
2726                                                 $text = $linestart . $wgContLang->formatNum( call_user_func( $func ) );
2727                                         }
2728                                         $found = true;
2729                                 }
2730                         }
2731                 }
2732
2733                 # Extensions
2734                 if ( !$found && substr( $part1, 0, 1 ) == '#' ) {
2735                         $colonPos = strpos( $part1, ':' );
2736                         if ( $colonPos !== false ) {
2737                                 $function = strtolower( substr( $part1, 1, $colonPos - 1 ) );
2738                                 if ( isset( $this->mFunctionHooks[$function] ) ) {
2739                                         $funcArgs = array_map( 'trim', $args );
2740                                         $funcArgs = array_merge( array( &$this, trim( substr( $part1, $colonPos + 1 ) ) ), $funcArgs );
2741                                         $result = call_user_func_array( $this->mFunctionHooks[$function], $funcArgs );
2742                                         $found = true;
2743
2744                                         // The text is usually already parsed, doesn't need triple-brace tags expanded, etc.
2745                                         //$noargs = true;
2746                                         //$noparse = true;
2747
2748                                         if ( is_array( $result ) ) {
2749                                                 $text = $linestart . $result[0];
2750                                                 unset( $result[0] );
2751
2752                                                 // Extract flags into the local scope
2753                                                 // This allows callers to set flags such as nowiki, noparse, found, etc.
2754                                                 extract( $result );
2755                                         } else {
2756                                                 $text = $linestart . $result;
2757                                         }
2758                                 }
2759                         }
2760                 }
2761
2762                 # Template table test
2763
2764                 # Did we encounter this template already? If yes, it is in the cache
2765                 # and we need to check for loops.
2766                 if ( !$found && isset( $this->mTemplates[$piece['title']] ) ) {
2767                         $found = true;
2768
2769                         # Infinite loop test
2770                         if ( isset( $this->mTemplatePath[$part1] ) ) {
2771                                 $noparse = true;
2772                                 $noargs = true;
2773                                 $found = true;
2774                                 $text = $linestart .
2775                                         '{{' . $part1 . '}}' .
2776                                         '<!-- WARNING: template loop detected -->';
2777                                 wfDebug( "$fname: template loop broken at '$part1'\n" );
2778                         } else {
2779                                 # set $text to cached message.
2780                                 $text = $linestart . $this->mTemplates[$piece['title']];
2781                         }
2782                 }
2783
2784                 # Load from database
2785                 $lastPathLevel = $this->mTemplatePath;
2786                 if ( !$found ) {
2787                         $ns = NS_TEMPLATE;
2788                         # declaring $subpage directly in the function call
2789                         # does not work correctly with references and breaks
2790                         # {{/subpage}}-style inclusions
2791                         $subpage = '';
2792                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
2793                         if ($subpage !== '') {
2794                                 $ns = $this->mTitle->getNamespace();
2795                         }
2796                         $title = Title::newFromText( $part1, $ns );
2797
2798                         if ( !is_null( $title ) ) {
2799                                 if ( !$title->isExternal() ) {
2800                                         # Check for excessive inclusion
2801                                         $dbk = $title->getPrefixedDBkey();
2802                                         if ( $this->incrementIncludeCount( $dbk ) ) {
2803                                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->mOutputType != OT_WIKI ) {
2804                                                         $text = SpecialPage::capturePath( $title );
2805                                                         if ( is_string( $text ) ) {
2806                                                                 $found = true;
2807                                                                 $noparse = true;
2808                                                                 $noargs = true;
2809                                                                 $isHTML = true;
2810                                                                 $this->disableCache();
2811                                                         }
2812                                                 } else {
2813                                                         $articleContent = $this->fetchTemplate( $title );
2814                                                         if ( $articleContent !== false ) {
2815                                                                 $found = true;
2816                                                                 $text = $articleContent;
2817                                                                 $replaceHeadings = true;
2818                                                         }
2819                                                 }
2820                                         }
2821
2822                                         # If the title is valid but undisplayable, make a link to it
2823                                         if ( $this->mOutputType == OT_HTML && !$found ) {
2824                                                 $text = '[['.$title->getPrefixedText().']]';
2825                                                 $found = true;
2826                                         }
2827                                 } elseif ( $title->isTrans() ) {
2828                                         // Interwiki transclusion
2829                                         if ( $this->mOutputType == OT_HTML && !$forceRawInterwiki ) {
2830                                                 $text = $this->interwikiTransclude( $title, 'render' );
2831                                                 $isHTML = true;
2832                                                 $noparse = true;
2833                                         } else {
2834                                                 $text = $this->interwikiTransclude( $title, 'raw' );
2835                                                 $replaceHeadings = true;
2836                                         }
2837                                         $found = true;
2838                                 }
2839
2840                                 # Template cache array insertion
2841                                 # Use the original $piece['title'] not the mangled $part1, so that
2842                                 # modifiers such as RAW: produce separate cache entries
2843                                 if( $found ) {
2844                                         $this->mTemplates[$piece['title']] = $text;
2845                                         $text = $linestart . $text;
2846                                 }
2847                         }
2848                 }
2849
2850                 # Recursive parsing, escaping and link table handling
2851                 # Only for HTML output
2852                 if ( $nowiki && $found && $this->mOutputType == OT_HTML ) {
2853                         $text = wfEscapeWikiText( $text );
2854                 } elseif ( ($this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI) && $found ) {
2855                         if ( $noargs ) {
2856                                 $assocArgs = array();
2857                         } else {
2858                                 # Clean up argument array
2859                                 $assocArgs = array();
2860                                 $index = 1;
2861                                 foreach( $args as $arg ) {
2862                                         $eqpos = strpos( $arg, '=' );
2863                                         if ( $eqpos === false ) {
2864                                                 $assocArgs[$index++] = $arg;
2865                                         } else {
2866                                                 $name = trim( substr( $arg, 0, $eqpos ) );
2867                                                 $value = trim( substr( $arg, $eqpos+1 ) );
2868                                                 if ( $value === false ) {
2869                                                         $value = '';
2870                                                 }
2871                                                 if ( $name !== false ) {
2872                                                         $assocArgs[$name] = $value;
2873                                                 }
2874                                         }
2875                                 }
2876
2877                                 # Add a new element to the templace recursion path
2878                                 $this->mTemplatePath[$part1] = 1;
2879                         }
2880
2881                         if ( !$noparse ) {
2882                                 # If there are any <onlyinclude> tags, only include them
2883                                 if ( in_string( '<onlyinclude>', $text ) && in_string( '</onlyinclude>', $text ) ) {
2884                                         preg_match_all( '/<onlyinclude>(.*?)\n?<\/onlyinclude>/s', $text, $m );
2885                                         $text = '';
2886                                         foreach ($m[1] as $piece)
2887                                                 $text .= $piece;
2888                                 }
2889                                 # Remove <noinclude> sections and <includeonly> tags
2890                                 $text = preg_replace( '/<noinclude>.*?<\/noinclude>/s', '', $text );
2891                                 $text = strtr( $text, array( '<includeonly>' => '' , '</includeonly>' => '' ) );
2892
2893                                 if( $this->mOutputType == OT_HTML ) {
2894                                         # Strip <nowiki>, <pre>, etc.
2895                                         $text = $this->strip( $text, $this->mStripState );
2896                                         $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'replaceVariables' ), $assocArgs );
2897                                 }
2898                                 $text = $this->replaceVariables( $text, $assocArgs );
2899
2900                                 # If the template begins with a table or block-level
2901                                 # element, it should be treated as beginning a new line.
2902                                 if (!$piece['lineStart'] && preg_match('/^({\\||:|;|#|\*)/', $text)) {
2903                                         $text = "\n" . $text;
2904                                 }
2905                         } elseif ( !$noargs ) {
2906                                 # $noparse and !$noargs
2907                                 # Just replace the arguments, not any double-brace items
2908                                 # This is used for rendered interwiki transclusion
2909                                 $text = $this->replaceVariables( $text, $assocArgs, true );
2910                         }
2911                 }
2912                 # Prune lower levels off the recursion check path
2913                 $this->mTemplatePath = $lastPathLevel;
2914
2915                 if ( !$found ) {
2916                         wfProfileOut( $fname );
2917                         return $piece['text'];
2918                 } else {
2919                         if ( $isHTML ) {
2920                                 # Replace raw HTML by a placeholder
2921                                 # Add a blank line preceding, to prevent it from mucking up
2922                                 # immediately preceding headings
2923                                 $text = "\n\n" . $this->insertStripItem( $text, $this->mStripState );
2924                         } else {
2925                                 # replace ==section headers==
2926                                 # XXX this needs to go away once we have a better parser.
2927                                 if ( $this->mOutputType != OT_WIKI && $replaceHeadings ) {
2928                                         if( !is_null( $title ) )
2929                                                 $encodedname = base64_encode($title->getPrefixedDBkey());
2930                                         else
2931                                                 $encodedname = base64_encode("");
2932                                         $m = preg_split('/(^={1,6}.*?={1,6}\s*?$)/m', $text, -1,
2933                                                 PREG_SPLIT_DELIM_CAPTURE);
2934                                         $text = '';
2935                                         $nsec = 0;
2936                                         for( $i = 0; $i < count($m); $i += 2 ) {
2937                                                 $text .= $m[$i];
2938                                                 if (!isset($m[$i + 1]) || $m[$i + 1] == "") continue;
2939                                                 $hl = $m[$i + 1];
2940                                                 if( strstr($hl, "<!--MWTEMPLATESECTION") ) {
2941                                                         $text .= $hl;
2942                                                         continue;
2943                                                 }
2944                                                 preg_match('/^(={1,6})(.*?)(={1,6})\s*?$/m', $hl, $m2);
2945                                                 $text .= $m2[1] . $m2[2] . "<!--MWTEMPLATESECTION="
2946                                                         . $encodedname . "&" . base64_encode("$nsec") . "-->" . $m2[3];
2947
2948                                                 $nsec++;
2949                                         }
2950                                 }
2951                         }
2952                 }
2953
2954                 # Prune lower levels off the recursion check path
2955                 $this->mTemplatePath = $lastPathLevel;
2956
2957                 if ( !$found ) {
2958                         wfProfileOut( $fname );
2959                         return $piece['text'];
2960                 } else {
2961                         wfProfileOut( $fname );
2962                         return $text;
2963                 }
2964         }
2965
2966         /**
2967          * Fetch the unparsed text of a template and register a reference to it.
2968          */
2969         function fetchTemplate( $title ) {
2970                 $text = false;
2971                 // Loop to fetch the article, with up to 1 redirect
2972                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
2973                         $rev = Revision::newFromTitle( $title );
2974                         $this->mOutput->addTemplate( $title, $title->getArticleID() );
2975                         if ( !$rev ) {
2976                                 break;
2977                         }
2978                         $text = $rev->getText();
2979                         if ( $text === false ) {
2980                                 break;
2981                         }
2982                         // Redirect?
2983                         $title = Title::newFromRedirect( $text );
2984                 }
2985                 return $text;
2986         }
2987
2988         /**
2989          * Transclude an interwiki link.
2990          */
2991         function interwikiTransclude( $title, $action ) {
2992                 global $wgEnableScaryTranscluding, $wgCanonicalNamespaceNames;
2993
2994                 if (!$wgEnableScaryTranscluding)
2995                         return wfMsg('scarytranscludedisabled');
2996
2997                 // The namespace will actually only be 0 or 10, depending on whether there was a leading :
2998                 // But we'll handle it generally anyway
2999                 if ( $title->getNamespace() ) {
3000                         // Use the canonical namespace, which should work anywhere
3001                         $articleName = $wgCanonicalNamespaceNames[$title->getNamespace()] . ':' . $title->getDBkey();
3002                 } else {
3003                         $articleName = $title->getDBkey();
3004                 }
3005
3006                 $url = str_replace('$1', urlencode($articleName), Title::getInterwikiLink($title->getInterwiki()));
3007                 $url .= "?action=$action";
3008                 if (strlen($url) > 255)
3009                         return wfMsg('scarytranscludetoolong');
3010                 return $this->fetchScaryTemplateMaybeFromCache($url);
3011         }
3012
3013         function fetchScaryTemplateMaybeFromCache($url) {
3014                 global $wgTranscludeCacheExpiry;
3015                 $dbr =& wfGetDB(DB_SLAVE);
3016                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3017                                 array('tc_url' => $url));
3018                 if ($obj) {
3019                         $time = $obj->tc_time;
3020                         $text = $obj->tc_contents;
3021                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
3022                                 return $text;
3023                         }
3024                 }
3025
3026                 $text = wfGetHTTP($url);
3027                 if (!$text)
3028                         return wfMsg('scarytranscludefailed', $url);
3029
3030                 $dbw =& wfGetDB(DB_MASTER);
3031                 $dbw->replace('transcache', array('tc_url'), array(
3032                         'tc_url' => $url,
3033                         'tc_time' => time(),
3034                         'tc_contents' => $text));
3035                 return $text;
3036         }
3037
3038
3039         /**
3040          * Triple brace replacement -- used for template arguments
3041          * @private
3042          */
3043         function argSubstitution( $matches ) {
3044                 $arg = trim( $matches['title'] );
3045                 $text = $matches['text'];
3046                 $inputArgs = end( $this->mArgStack );
3047
3048                 if ( array_key_exists( $arg, $inputArgs ) ) {
3049                         $text = $inputArgs[$arg];
3050                 } else if ($this->mOutputType == OT_HTML && null != $matches['parts'] && count($matches['parts']) > 0) {
3051                         $text = $matches['parts'][0];
3052                 }
3053
3054                 return $text;
3055         }
3056
3057         /**
3058          * Returns true if the function is allowed to include this entity
3059          * @private
3060          */
3061         function incrementIncludeCount( $dbk ) {
3062                 if ( !array_key_exists( $dbk, $this->mIncludeCount ) ) {
3063                         $this->mIncludeCount[$dbk] = 0;
3064                 }
3065                 if ( ++$this->mIncludeCount[$dbk] <= MAX_INCLUDE_REPEAT ) {
3066                         return true;
3067                 } else {
3068                         return false;
3069                 }
3070         }
3071
3072         /**
3073          * Detect __TOC__ magic word and set a placeholder
3074          */
3075         function stripToc( $text ) {
3076                 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
3077                 # do not add TOC
3078                 $mw = MagicWord::get( MAG_NOTOC );
3079                 if( $mw->matchAndRemove( $text ) ) {
3080                         $this->mShowToc = false;
3081                 }
3082
3083                 $mw = MagicWord::get( MAG_TOC );
3084                 if( $mw->match( $text ) ) {
3085                         $this->mShowToc = true;
3086                         $this->mForceTocPosition = true;
3087
3088                         // Set a placeholder. At the end we'll fill it in with the TOC.
3089                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3090
3091                         // Only keep the first one.
3092                         $text = $mw->replace( '', $text );
3093                 }
3094                 return $text;
3095         }
3096
3097         /**
3098          * This function accomplishes several tasks:
3099          * 1) Auto-number headings if that option is enabled
3100          * 2) Add an [edit] link to sections for logged in users who have enabled the option
3101          * 3) Add a Table of contents on the top for users who have enabled the option
3102          * 4) Auto-anchor headings
3103          *
3104          * It loops through all headlines, collects the necessary data, then splits up the
3105          * string and re-inserts the newly formatted headlines.
3106          *
3107          * @param string $text
3108          * @param boolean $isMain
3109          * @private
3110          */
3111         function formatHeadings( $text, $isMain=true ) {
3112                 global $wgMaxTocLevel, $wgContLang;
3113
3114                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3115                 if( !$this->mTitle->userCanEdit() ) {
3116                         $showEditLink = 0;
3117                 } else {
3118                         $showEditLink = $this->mOptions->getEditSection();
3119                 }
3120
3121                 # Inhibit editsection links if requested in the page
3122                 $esw =& MagicWord::get( MAG_NOEDITSECTION );
3123                 if( $esw->matchAndRemove( $text ) ) {
3124                         $showEditLink = 0;
3125                 }
3126
3127                 # Get all headlines for numbering them and adding funky stuff like [edit]
3128                 # links - this is for later, but we need the number of headlines right now
3129                 $numMatches = preg_match_all( '/<H([1-6])(.*?'.'>)(.*?)<\/H[1-6] *>/i', $text, $matches );
3130
3131                 # if there are fewer than 4 headlines in the article, do not show TOC
3132                 # unless it's been explicitly enabled.
3133                 $enoughToc = $this->mShowToc &&
3134                         (($numMatches >= 4) || $this->mForceTocPosition);
3135
3136                 # Allow user to stipulate that a page should have a "new section"
3137                 # link added via __NEWSECTIONLINK__
3138                 $mw =& MagicWord::get( MAG_NEWSECTIONLINK );
3139                 if( $mw->matchAndRemove( $text ) )
3140                         $this->mOutput->setNewSection( true );
3141
3142                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3143                 # override above conditions and always show TOC above first header
3144                 $mw =& MagicWord::get( MAG_FORCETOC );
3145                 if ($mw->matchAndRemove( $text ) ) {
3146                         $this->mShowToc = true;
3147                         $enoughToc = true;
3148                 }
3149
3150                 # Never ever show TOC if no headers
3151                 if( $numMatches < 1 ) {
3152                         $enoughToc = false;
3153                 }
3154
3155                 # We need this to perform operations on the HTML
3156                 $sk =& $this->mOptions->getSkin();
3157
3158                 # headline counter
3159                 $headlineCount = 0;
3160                 $sectionCount = 0; # headlineCount excluding template sections
3161
3162                 # Ugh .. the TOC should have neat indentation levels which can be
3163                 # passed to the skin functions. These are determined here
3164                 $toc = '';
3165                 $full = '';
3166                 $head = array();
3167                 $sublevelCount = array();
3168                 $levelCount = array();
3169                 $toclevel = 0;
3170                 $level = 0;
3171                 $prevlevel = 0;
3172                 $toclevel = 0;
3173                 $prevtoclevel = 0;
3174
3175                 foreach( $matches[3] as $headline ) {
3176                         $istemplate = 0;
3177                         $templatetitle = '';
3178                         $templatesection = 0;
3179                         $numbering = '';
3180
3181                         if (preg_match("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", $headline, $mat)) {
3182                                 $istemplate = 1;
3183                                 $templatetitle = base64_decode($mat[1]);
3184                                 $templatesection = 1 + (int)base64_decode($mat[2]);
3185                                 $headline = preg_replace("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", "", $headline);
3186                         }
3187
3188                         if( $toclevel ) {
3189                                 $prevlevel = $level;
3190                                 $prevtoclevel = $toclevel;
3191                         }
3192                         $level = $matches[1][$headlineCount];
3193
3194                         if( $doNumberHeadings || $enoughToc ) {
3195
3196                                 if ( $level > $prevlevel ) {
3197                                         # Increase TOC level
3198                                         $toclevel++;
3199                                         $sublevelCount[$toclevel] = 0;
3200                                         if( $toclevel<$wgMaxTocLevel ) {
3201                                                 $toc .= $sk->tocIndent();
3202                                         }
3203                                 }
3204                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3205                                         # Decrease TOC level, find level to jump to
3206
3207                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
3208                                                 # Can only go down to level 1
3209                                                 $toclevel = 1;
3210                                         } else {
3211                                                 for ($i = $toclevel; $i > 0; $i--) {
3212                                                         if ( $levelCount[$i] == $level ) {
3213                                                                 # Found last matching level
3214                                                                 $toclevel = $i;
3215                                                                 break;
3216                                                         }
3217                                                         elseif ( $levelCount[$i] < $level ) {
3218                                                                 # Found first matching level below current level
3219                                                                 $toclevel = $i + 1;
3220                                                                 break;
3221                                                         }
3222                                                 }
3223                                         }
3224                                         if( $toclevel<$wgMaxTocLevel ) {
3225                                                 $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3226                                         }
3227                                 }
3228                                 else {
3229                                         # No change in level, end TOC line
3230                                         if( $toclevel<$wgMaxTocLevel ) {
3231                                                 $toc .= $sk->tocLineEnd();
3232                                         }
3233                                 }
3234
3235                                 $levelCount[$toclevel] = $level;
3236
3237                                 # count number of headlines for each level
3238                                 @$sublevelCount[$toclevel]++;
3239                                 $dot = 0;
3240                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3241                                         if( !empty( $sublevelCount[$i] ) ) {
3242                                                 if( $dot ) {
3243                                                         $numbering .= '.';
3244                                                 }
3245                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3246                                                 $dot = 1;
3247                                         }
3248                                 }
3249                         }
3250
3251                         # The canonized header is a version of the header text safe to use for links
3252                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3253                         $canonized_headline = $this->unstrip( $headline, $this->mStripState );
3254                         $canonized_headline = $this->unstripNoWiki( $canonized_headline, $this->mStripState );
3255
3256                         # Remove link placeholders by the link text.
3257                         #     <!--LINK number-->
3258                         # turns into
3259                         #     link text with suffix
3260                         $canonized_headline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
3261                                                             "\$this->mLinkHolders['texts'][\$1]",
3262                                                             $canonized_headline );
3263                         $canonized_headline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
3264                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
3265                                                             $canonized_headline );
3266
3267                         # strip out HTML
3268                         $canonized_headline = preg_replace( '/<.*?' . '>/','',$canonized_headline );
3269                         $tocline = trim( $canonized_headline );
3270                         # Save headline for section edit hint before it's escaped
3271                         $headline_hint = trim( $canonized_headline );
3272                         $canonized_headline = Sanitizer::escapeId( $tocline );
3273                         $refers[$headlineCount] = $canonized_headline;
3274
3275                         # count how many in assoc. array so we can track dupes in anchors
3276                         @$refers[$canonized_headline]++;
3277                         $refcount[$headlineCount]=$refers[$canonized_headline];
3278
3279                         # Don't number the heading if it is the only one (looks silly)
3280                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3281                                 # the two are different if the line contains a link
3282                                 $headline=$numbering . ' ' . $headline;
3283                         }
3284
3285                         # Create the anchor for linking from the TOC to the section
3286                         $anchor = $canonized_headline;
3287                         if($refcount[$headlineCount] > 1 ) {
3288                                 $anchor .= '_' . $refcount[$headlineCount];
3289                         }
3290                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3291                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3292                         }
3293                         if( $showEditLink && ( !$istemplate || $templatetitle !== "" ) ) {
3294                                 if ( empty( $head[$headlineCount] ) ) {
3295                                         $head[$headlineCount] = '';
3296                                 }
3297                                 if( $istemplate )
3298                                         $head[$headlineCount] .= $sk->editSectionLinkForOther($templatetitle, $templatesection);
3299                                 else
3300                                         $head[$headlineCount] .= $sk->editSectionLink($this->mTitle, $sectionCount+1, $headline_hint);
3301                         }
3302
3303                         # give headline the correct <h#> tag
3304                         @$head[$headlineCount] .= "<a name=\"$anchor\"></a><h".$level.$matches[2][$headlineCount] .$headline.'</h'.$level.'>';
3305
3306                         $headlineCount++;
3307                         if( !$istemplate )
3308                                 $sectionCount++;
3309                 }
3310
3311                 if( $enoughToc ) {
3312                         if( $toclevel<$wgMaxTocLevel ) {
3313                                 $toc .= $sk->tocUnindent( $toclevel - 1 );
3314                         }
3315                         $toc = $sk->tocList( $toc );
3316                 }
3317
3318                 # split up and insert constructed headlines
3319
3320                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3321                 $i = 0;
3322
3323                 foreach( $blocks as $block ) {
3324                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
3325                                 # This is the [edit] link that appears for the top block of text when
3326                                 # section editing is enabled
3327
3328                                 # Disabled because it broke block formatting
3329                                 # For example, a bullet point in the top line
3330                                 # $full .= $sk->editSectionLink(0);
3331                         }
3332                         $full .= $block;
3333                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3334                                 # Top anchor now in skin
3335                                 $full = $full.$toc;
3336                         }
3337
3338                         if( !empty( $head[$i] ) ) {
3339                                 $full .= $head[$i];
3340                         }
3341                         $i++;
3342                 }
3343                 if( $this->mForceTocPosition ) {
3344                         return str_replace( '<!--MWTOC-->', $toc, $full );
3345                 } else {
3346                         return $full;
3347                 }
3348         }
3349
3350         /**
3351          * Return an HTML link for the "ISBN 123456" text
3352          * @private
3353          */
3354         function magicISBN( $text ) {
3355                 $fname = 'Parser::magicISBN';
3356                 wfProfileIn( $fname );
3357
3358                 $a = split( 'ISBN ', ' '.$text );
3359                 if ( count ( $a ) < 2 ) {
3360                         wfProfileOut( $fname );
3361                         return $text;
3362                 }
3363                 $text = substr( array_shift( $a ), 1);
3364                 $valid = '0123456789-Xx';
3365
3366                 foreach ( $a as $x ) {
3367                         # hack: don't replace inside thumbnail title/alt
3368                         # attributes
3369                         if(preg_match('/<[^>]+(alt|title)="[^">]*$/', $text)) {
3370                                 $text .= "ISBN $x";
3371                                 continue;
3372                         }
3373
3374                         $isbn = $blank = '' ;
3375                         while ( ' ' == $x{0} ) {
3376                                 $blank .= ' ';
3377                                 $x = substr( $x, 1 );
3378                         }
3379                         if ( $x == '' ) { # blank isbn
3380                                 $text .= "ISBN $blank";
3381                                 continue;
3382                         }
3383                         while ( strstr( $valid, $x{0} ) != false ) {
3384                                 $isbn .= $x{0};
3385                                 $x = substr( $x, 1 );
3386                         }
3387                         $num = str_replace( '-', '', $isbn );
3388                         $num = str_replace( ' ', '', $num );
3389                         $num = str_replace( 'x', 'X', $num );
3390
3391                         if ( '' == $num ) {
3392                                 $text .= "ISBN $blank$x";
3393                         } else {
3394                                 $titleObj = Title::makeTitle( NS_SPECIAL, 'Booksources' );
3395                                 $text .= '<a href="' .
3396                                         $titleObj->escapeLocalUrl( 'isbn='.$num ) .
3397                                         "\" class=\"internal\">ISBN $isbn</a>";
3398                                 $text .= $x;
3399                         }
3400                 }
3401                 wfProfileOut( $fname );
3402                 return $text;
3403         }
3404
3405         /**
3406          * Return an HTML link for the "RFC 1234" text
3407          *
3408          * @private
3409          * @param string $text     Text to be processed
3410          * @param string $keyword  Magic keyword to use (default RFC)
3411          * @param string $urlmsg   Interface message to use (default rfcurl)
3412          * @return string
3413          */
3414         function magicRFC( $text, $keyword='RFC ', $urlmsg='rfcurl'  ) {
3415
3416                 $valid = '0123456789';
3417                 $internal = false;
3418
3419                 $a = split( $keyword, ' '.$text );
3420                 if ( count ( $a ) < 2 ) {
3421                         return $text;
3422                 }
3423                 $text = substr( array_shift( $a ), 1);
3424
3425                 /* Check if keyword is preceed by [[.
3426                  * This test is made here cause of the array_shift above
3427                  * that prevent the test to be done in the foreach.
3428                  */
3429                 if ( substr( $text, -2 ) == '[[' ) {
3430                         $internal = true;
3431                 }
3432
3433                 foreach ( $a as $x ) {
3434                         /* token might be empty if we have RFC RFC 1234 */
3435                         if ( $x=='' ) {
3436                                 $text.=$keyword;
3437                                 continue;
3438                                 }
3439
3440                         # hack: don't replace inside thumbnail title/alt
3441                         # attributes
3442                         if(preg_match('/<[^>]+(alt|title)="[^">]*$/', $text)) {
3443                                 $text .= $keyword . $x;
3444                                 continue;
3445                         }
3446
3447                         $id = $blank = '' ;
3448
3449                         /** remove and save whitespaces in $blank */
3450                         while ( $x{0} == ' ' ) {
3451                                 $blank .= ' ';
3452                                 $x = substr( $x, 1 );
3453                         }
3454
3455                         /** remove and save the rfc number in $id */
3456                         while ( strstr( $valid, $x{0} ) != false ) {
3457                                 $id .= $x{0};
3458                                 $x = substr( $x, 1 );
3459                         }
3460
3461                         if ( $id == '' ) {
3462                                 /* call back stripped spaces*/
3463                                 $text .= $keyword.$blank.$x;
3464                         } elseif( $internal ) {
3465                                 /* normal link */
3466                                 $text .= $keyword.$id.$x;
3467                         } else {
3468                                 /* build the external link*/
3469                                 $url = wfMsg( $urlmsg, $id);
3470                                 $sk =& $this->mOptions->getSkin();
3471                                 $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
3472                                 $text .= "<a href='{$url}'{$la}>{$keyword}{$id}</a>{$x}";
3473                         }
3474
3475                         /* Check if the next RFC keyword is preceed by [[ */
3476                         $internal = ( substr($x,-2) == '[[' );
3477                 }
3478                 return $text;
3479         }
3480
3481         /**
3482          * Transform wiki markup when saving a page by doing \r\n -> \n
3483          * conversion, substitting signatures, {{subst:}} templates, etc.
3484          *
3485          * @param string $text the text to transform
3486          * @param Title &$title the Title object for the current article
3487          * @param User &$user the User object describing the current user
3488          * @param ParserOptions $options parsing options
3489          * @param bool $clearState whether to clear the parser state first
3490          * @return string the altered wiki markup
3491          * @public
3492          */
3493         function preSaveTransform( $text, &$title, &$user, $options, $clearState = true ) {
3494                 $this->mOptions = $options;
3495                 $this->mTitle =& $title;
3496                 $this->mOutputType = OT_WIKI;
3497
3498                 if ( $clearState ) {
3499                         $this->clearState();
3500                 }
3501
3502                 $stripState = false;
3503                 $pairs = array(
3504                         "\r\n" => "\n",
3505                 );
3506                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3507                 $text = $this->strip( $text, $stripState, true );
3508                 $text = $this->pstPass2( $text, $stripState, $user );
3509                 $text = $this->unstrip( $text, $stripState );
3510                 $text = $this->unstripNoWiki( $text, $stripState );
3511                 return $text;
3512         }
3513
3514         /**
3515          * Pre-save transform helper function
3516          * @private
3517          */
3518         function pstPass2( $text, &$stripState, &$user ) {
3519                 global $wgContLang, $wgLocaltimezone;
3520
3521                 /* Note: This is the timestamp saved as hardcoded wikitext to
3522                  * the database, we use $wgContLang here in order to give
3523                  * everyone the same signature and use the default one rather
3524                  * than the one selected in each user's preferences.
3525                  */
3526                 if ( isset( $wgLocaltimezone ) ) {
3527                         $oldtz = getenv( 'TZ' );
3528                         putenv( 'TZ='.$wgLocaltimezone );
3529                 }
3530                 $d = $wgContLang->timeanddate( date( 'YmdHis' ), false, false) .
3531                   ' (' . date( 'T' ) . ')';
3532                 if ( isset( $wgLocaltimezone ) ) {
3533                         putenv( 'TZ='.$oldtz );
3534                 }
3535
3536                 # Variable replacement
3537                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3538                 $text = $this->replaceVariables( $text );
3539
3540                 # Strip out <nowiki> etc. added via replaceVariables
3541                 $text = $this->strip( $text, $stripState );
3542
3543                 # Signatures
3544                 $sigText = $this->getUserSig( $user );
3545                 $text = strtr( $text, array(
3546                         '~~~~~' => $d,
3547                         '~~~~' => "$sigText $d",
3548                         '~~~' => $sigText
3549                 ) );
3550
3551                 # Context links: [[|name]] and [[name (context)|]]
3552                 #
3553                 global $wgLegalTitleChars;
3554                 $tc = "[$wgLegalTitleChars]";
3555                 $np = str_replace( array( '(', ')' ), array( '', '' ), $tc ); # No parens
3556
3557                 $namespacechar = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
3558                 $conpat = "/^({$np}+) \\(({$tc}+)\\)$/";
3559
3560                 $p1 = "/\[\[({$np}+) \\(({$np}+)\\)\\|]]/";             # [[page (context)|]]
3561                 $p2 = "/\[\[\\|({$tc}+)]]/";                                    # [[|page]]
3562                 $p3 = "/\[\[(:*$namespacechar+):({$np}+)\\|]]/";                # [[namespace:page|]] and [[:namespace:page|]]
3563                 $p4 = "/\[\[(:*$namespacechar+):({$np}+) \\(({$np}+)\\)\\|]]/"; # [[ns:page (cont)|]] and [[:ns:page (cont)|]]
3564                 $context = '';
3565                 $t = $this->mTitle->getText();
3566                 if ( preg_match( $conpat, $t, $m ) ) {
3567                         $context = $m[2];
3568                 }
3569                 $text = preg_replace( $p4, '[[\\1:\\2 (\\3)|\\2]]', $text );
3570                 $text = preg_replace( $p1, '[[\\1 (\\2)|\\1]]', $text );
3571                 $text = preg_replace( $p3, '[[\\1:\\2|\\2]]', $text );
3572
3573                 if ( '' == $context ) {
3574                         $text = preg_replace( $p2, '[[\\1]]', $text );
3575                 } else {
3576                         $text = preg_replace( $p2, "[[\\1 ({$context})|\\1]]", $text );
3577                 }
3578
3579                 # Trim trailing whitespace
3580                 # MAG_END (__END__) tag allows for trailing
3581                 # whitespace to be deliberately included
3582                 $text = rtrim( $text );
3583                 $mw =& MagicWord::get( MAG_END );
3584                 $mw->matchAndRemove( $text );
3585
3586                 return $text;
3587         }
3588
3589         /**
3590          * Fetch the user's signature text, if any, and normalize to
3591          * validated, ready-to-insert wikitext.
3592          *
3593          * @param User $user
3594          * @return string
3595          * @private
3596          */
3597         function getUserSig( &$user ) {
3598                 $username = $user->getName();
3599                 $nickname = $user->getOption( 'nickname' );
3600                 $nickname = $nickname === '' ? $username : $nickname;
3601
3602                 if( $user->getBoolOption( 'fancysig' ) !== false ) {
3603                         # Sig. might contain markup; validate this
3604                         if( $this->validateSig( $nickname ) !== false ) {
3605                                 # Validated; clean up (if needed) and return it
3606                                 return $this->cleanSig( $nickname, true );
3607                         } else {
3608                                 # Failed to validate; fall back to the default
3609                                 $nickname = $username;
3610                                 wfDebug( "Parser::getUserSig: $username has bad XML tags in signature.\n" );
3611                         }
3612                 }
3613
3614                 # If we're still here, make it a link to the user page
3615                 $userpage = $user->getUserPage();
3616                 return( '[[' . $userpage->getPrefixedText() . '|' . wfEscapeWikiText( $nickname ) . ']]' );
3617         }
3618
3619         /**
3620          * Check that the user's signature contains no bad XML
3621          *
3622          * @param string $text
3623          * @return mixed An expanded string, or false if invalid.
3624          */
3625         function validateSig( $text ) {
3626                 return( wfIsWellFormedXmlFragment( $text ) ? $text : false );
3627         }
3628
3629         /**
3630          * Clean up signature text
3631          *
3632          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures
3633          * 2) Substitute all transclusions
3634          *
3635          * @param string $text
3636          * @param $parsing Whether we're cleaning (preferences save) or parsing
3637          * @return string Signature text
3638          */
3639         function cleanSig( $text, $parsing = false ) {
3640                 global $wgTitle;
3641                 $this->startExternalParse( $wgTitle, new ParserOptions(), $parsing ? OT_WIKI : OT_MSG );
3642
3643                 $substWord = MagicWord::get( MAG_SUBST );
3644                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3645                 $substText = '{{' . $substWord->getSynonym( 0 );
3646
3647                 $text = preg_replace( $substRegex, $substText, $text );
3648                 $text = preg_replace( '/~{3,5}/', '', $text );
3649                 $text = $this->replaceVariables( $text );
3650
3651                 $this->clearState();
3652                 return $text;
3653         }
3654
3655         /**
3656          * Set up some variables which are usually set up in parse()
3657          * so that an external function can call some class members with confidence
3658          * @public
3659          */
3660         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
3661                 $this->mTitle =& $title;
3662                 $this->mOptions = $options;
3663                 $this->mOutputType = $outputType;
3664                 if ( $clearState ) {
3665                         $this->clearState();
3666                 }
3667         }
3668
3669         /**
3670          * Transform a MediaWiki message by replacing magic variables.
3671          *
3672          * @param string $text the text to transform
3673          * @param ParserOptions $options  options
3674          * @return string the text with variables substituted
3675          * @public
3676          */
3677         function transformMsg( $text, $options ) {
3678                 global $wgTitle;
3679                 static $executing = false;
3680
3681                 $fname = "Parser::transformMsg";
3682
3683                 # Guard against infinite recursion
3684                 if ( $executing ) {
3685                         return $text;
3686                 }
3687                 $executing = true;
3688
3689                 wfProfileIn($fname);
3690
3691                 $this->mTitle = $wgTitle;
3692                 $this->mOptions = $options;
3693                 $this->mOutputType = OT_MSG;
3694                 $this->clearState();
3695                 $text = $this->replaceVariables( $text );
3696
3697                 $executing = false;
3698                 wfProfileOut($fname);
3699                 return $text;
3700         }
3701
3702         /**
3703          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
3704          * The callback should have the following form:
3705          *    function myParserHook( $text, $params, &$parser ) { ... }
3706          *
3707          * Transform and return $text. Use $parser for any required context, e.g. use
3708          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
3709          *
3710          * @public
3711          *
3712          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
3713          * @param mixed $callback The callback function (and object) to use for the tag
3714          *
3715          * @return The old value of the mTagHooks array associated with the hook
3716          */
3717         function setHook( $tag, $callback ) {
3718                 $oldVal = @$this->mTagHooks[$tag];
3719                 $this->mTagHooks[$tag] = $callback;
3720
3721                 return $oldVal;
3722         }
3723
3724         /**
3725          * Create a function, e.g. {{sum:1|2|3}}
3726          * The callback function should have the form:
3727          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
3728          *
3729          * The callback may either return the text result of the function, or an array with the text
3730          * in element 0, and a number of flags in the other elements. The names of the flags are
3731          * specified in the keys. Valid flags are:
3732          *   found                     The text returned is valid, stop processing the template. This
3733          *                             is on by default.
3734          *   nowiki                    Wiki markup in the return value should be escaped
3735          *   noparse                   Unsafe HTML tags should not be stripped, etc.
3736          *   noargs                    Don't replace triple-brace arguments in the return value
3737          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
3738          *
3739          * @public
3740          *
3741          * @param string $name The function name. Function names are case-insensitive.
3742          * @param mixed $callback The callback function (and object) to use
3743          *
3744          * @return The old callback function for this name, if any
3745          */
3746         function setFunctionHook( $name, $callback ) {
3747                 $name = strtolower( $name );
3748                 $oldVal = @$this->mFunctionHooks[$name];
3749                 $this->mFunctionHooks[$name] = $callback;
3750                 return $oldVal;
3751         }
3752
3753         /**
3754          * Replace <!--LINK--> link placeholders with actual links, in the buffer
3755          * Placeholders created in Skin::makeLinkObj()
3756          * Returns an array of links found, indexed by PDBK:
3757          *  0 - broken
3758          *  1 - normal link
3759          *  2 - stub
3760          * $options is a bit field, RLH_FOR_UPDATE to select for update
3761          */
3762         function replaceLinkHolders( &$text, $options = 0 ) {
3763                 global $wgUser;
3764                 global $wgOutputReplace;
3765
3766                 $fname = 'Parser::replaceLinkHolders';
3767                 wfProfileIn( $fname );
3768
3769                 $pdbks = array();
3770                 $colours = array();
3771                 $sk =& $this->mOptions->getSkin();
3772                 $linkCache =& LinkCache::singleton();
3773
3774                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
3775                         wfProfileIn( $fname.'-check' );
3776                         $dbr =& wfGetDB( DB_SLAVE );
3777                         $page = $dbr->tableName( 'page' );
3778                         $threshold = $wgUser->getOption('stubthreshold');
3779
3780                         # Sort by namespace
3781                         asort( $this->mLinkHolders['namespaces'] );
3782
3783                         # Generate query
3784                         $query = false;
3785                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3786                                 # Make title object
3787                                 $title = $this->mLinkHolders['titles'][$key];
3788
3789                                 # Skip invalid entries.
3790                                 # Result will be ugly, but prevents crash.
3791                                 if ( is_null( $title ) ) {
3792                                         continue;
3793                                 }
3794                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
3795
3796                                 # Check if it's a static known link, e.g. interwiki
3797                                 if ( $title->isAlwaysKnown() ) {
3798                                         $colours[$pdbk] = 1;
3799                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
3800                                         $colours[$pdbk] = 1;
3801                                         $this->mOutput->addLink( $title, $id );
3802                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
3803                                         $colours[$pdbk] = 0;
3804                                 } else {
3805                                         # Not in the link cache, add it to the query
3806                                         if ( !isset( $current ) ) {
3807                                                 $current = $ns;
3808                                                 $query =  "SELECT page_id, page_namespace, page_title";
3809                                                 if ( $threshold > 0 ) {
3810                                                         $query .= ', page_len, page_is_redirect';
3811                                                 }
3812                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
3813                                         } elseif ( $current != $ns ) {
3814                                                 $current = $ns;
3815                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
3816                                         } else {
3817                                                 $query .= ', ';
3818                                         }
3819
3820                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
3821                                 }
3822                         }
3823                         if ( $query ) {
3824                                 $query .= '))';
3825                                 if ( $options & RLH_FOR_UPDATE ) {
3826                                         $query .= ' FOR UPDATE';
3827                                 }
3828
3829                                 $res = $dbr->query( $query, $fname );
3830
3831                                 # Fetch data and form into an associative array
3832                                 # non-existent = broken
3833                                 # 1 = known
3834                                 # 2 = stub
3835                                 while ( $s = $dbr->fetchObject($res) ) {
3836                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
3837                                         $pdbk = $title->getPrefixedDBkey();
3838                                         $linkCache->addGoodLinkObj( $s->page_id, $title );
3839                                         $this->mOutput->addLink( $title, $s->page_id );
3840
3841                                         if ( $threshold >  0 ) {
3842                                                 $size = $s->page_len;
3843                                                 if ( $s->page_is_redirect || $s->page_namespace != 0 || $size >= $threshold ) {
3844                                                         $colours[$pdbk] = 1;
3845                                                 } else {
3846                                                         $colours[$pdbk] = 2;
3847                                                 }
3848                                         } else {
3849                                                 $colours[$pdbk] = 1;
3850                                         }
3851                                 }
3852                         }
3853                         wfProfileOut( $fname.'-check' );
3854
3855                         # Construct search and replace arrays
3856                         wfProfileIn( $fname.'-construct' );
3857                         $wgOutputReplace = array();
3858                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3859                                 $pdbk = $pdbks[$key];
3860                                 $searchkey = "<!--LINK $key-->";
3861                                 $title = $this->mLinkHolders['titles'][$key];
3862                                 if ( empty( $colours[$pdbk] ) ) {
3863                                         $linkCache->addBadLinkObj( $title );
3864                                         $colours[$pdbk] = 0;
3865                                         $this->mOutput->addLink( $title, 0 );
3866                                         $wgOutputReplace[$searchkey] = $sk->makeBrokenLinkObj( $title,
3867                                                                         $this->mLinkHolders['texts'][$key],
3868                                                                         $this->mLinkHolders['queries'][$key] );
3869                                 } elseif ( $colours[$pdbk] == 1 ) {
3870                                         $wgOutputReplace[$searchkey] = $sk->makeKnownLinkObj( $title,
3871                                                                         $this->mLinkHolders['texts'][$key],
3872                                                                         $this->mLinkHolders['queries'][$key] );
3873                                 } elseif ( $colours[$pdbk] == 2 ) {
3874                                         $wgOutputReplace[$searchkey] = $sk->makeStubLinkObj( $title,
3875                                                                         $this->mLinkHolders['texts'][$key],
3876                                                                         $this->mLinkHolders['queries'][$key] );
3877                                 }
3878                         }
3879                         wfProfileOut( $fname.'-construct' );
3880
3881                         # Do the thing
3882                         wfProfileIn( $fname.'-replace' );
3883
3884                         $text = preg_replace_callback(
3885                                 '/(<!--LINK .*?-->)/',
3886                                 "wfOutputReplaceMatches",
3887                                 $text);
3888
3889                         wfProfileOut( $fname.'-replace' );
3890                 }
3891
3892                 # Now process interwiki link holders
3893                 # This is quite a bit simpler than internal links
3894                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
3895                         wfProfileIn( $fname.'-interwiki' );
3896                         # Make interwiki link HTML
3897                         $wgOutputReplace = array();
3898                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
3899                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
3900                                 $wgOutputReplace[$key] = $sk->makeLinkObj( $title, $link );
3901                         }
3902
3903                         $text = preg_replace_callback(
3904                                 '/<!--IWLINK (.*?)-->/',
3905                                 "wfOutputReplaceMatches",
3906                                 $text );
3907                         wfProfileOut( $fname.'-interwiki' );
3908                 }
3909
3910                 wfProfileOut( $fname );
3911                 return $colours;
3912         }
3913
3914         /**
3915          * Replace <!--LINK--> link placeholders with plain text of links
3916          * (not HTML-formatted).
3917          * @param string $text
3918          * @return string
3919          */
3920         function replaceLinkHoldersText( $text ) {
3921                 $fname = 'Parser::replaceLinkHoldersText';
3922                 wfProfileIn( $fname );
3923
3924                 $text = preg_replace_callback(
3925                         '/<!--(LINK|IWLINK) (.*?)-->/',
3926                         array( &$this, 'replaceLinkHoldersTextCallback' ),
3927                         $text );
3928
3929                 wfProfileOut( $fname );
3930                 return $text;
3931         }
3932
3933         /**
3934          * @param array $matches
3935          * @return string
3936          * @private
3937          */
3938         function replaceLinkHoldersTextCallback( $matches ) {
3939                 $type = $matches[1];
3940                 $key  = $matches[2];
3941                 if( $type == 'LINK' ) {
3942                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
3943                                 return $this->mLinkHolders['texts'][$key];
3944                         }
3945                 } elseif( $type == 'IWLINK' ) {
3946                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
3947                                 return $this->mInterwikiLinkHolders['texts'][$key];
3948                         }
3949                 }
3950                 return $matches[0];
3951         }
3952
3953         /**
3954          * Renders an image gallery from a text with one line per image.
3955          * text labels may be given by using |-style alternative text. E.g.
3956          *   Image:one.jpg|The number "1"
3957          *   Image:tree.jpg|A tree
3958          * given as text will return the HTML of a gallery with two images,
3959          * labeled 'The number "1"' and
3960          * 'A tree'.
3961          */
3962         function renderImageGallery( $text ) {
3963                 # Setup the parser
3964                 $parserOptions = new ParserOptions;
3965                 $localParser = new Parser();
3966
3967                 $ig = new ImageGallery();
3968                 $ig->setShowBytes( false );
3969                 $ig->setShowFilename( false );
3970                 $ig->setParsing();
3971                 $lines = explode( "\n", $text );
3972
3973                 foreach ( $lines as $line ) {
3974                         # match lines like these:
3975                         # Image:someimage.jpg|This is some image
3976                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
3977                         # Skip empty lines
3978                         if ( count( $matches ) == 0 ) {
3979                                 continue;
3980                         }
3981                         $nt =& Title::newFromText( $matches[1] );
3982                         if( is_null( $nt ) ) {
3983                                 # Bogus title. Ignore these so we don't bomb out later.
3984                                 continue;
3985                         }
3986                         if ( isset( $matches[3] ) ) {
3987                                 $label = $matches[3];
3988                         } else {
3989                                 $label = '';
3990                         }
3991
3992                         $pout = $localParser->parse( $label , $this->mTitle, $parserOptions );
3993                         $html = $pout->getText();
3994
3995                         $ig->add( new Image( $nt ), $html );
3996
3997                         # Only add real images (bug #5586)
3998                         if ( $nt->getNamespace() == NS_IMAGE ) {
3999                                 $this->mOutput->addImage( $nt->getDBkey() );
4000                         }
4001
4002                         # Register links with the parent parser
4003                         foreach( $pout->getLinks() as $ns => $keys ) {
4004                                 foreach( $keys as $dbk => $id )
4005                                         $this->mOutput->addLink( Title::makeTitle( $ns, $dbk ), $id );
4006                         }
4007
4008                 }
4009                 return $ig->toHTML();
4010         }
4011
4012         /**
4013          * Parse image options text and use it to make an image
4014          */
4015         function makeImage( &$nt, $options ) {
4016                 global $wgUseImageResize;
4017
4018                 $align = '';
4019
4020                 # Check if the options text is of the form "options|alt text"
4021                 # Options are:
4022                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
4023                 #  * left               no resizing, just left align. label is used for alt= only
4024                 #  * right              same, but right aligned
4025                 #  * none               same, but not aligned
4026                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
4027                 #  * center             center the image
4028                 #  * framed             Keep original image size, no magnify-button.
4029
4030                 $part = explode( '|', $options);
4031
4032                 $mwThumb  =& MagicWord::get( MAG_IMG_THUMBNAIL );
4033                 $mwManualThumb =& MagicWord::get( MAG_IMG_MANUALTHUMB );
4034                 $mwLeft   =& MagicWord::get( MAG_IMG_LEFT );
4035                 $mwRight  =& MagicWord::get( MAG_IMG_RIGHT );
4036                 $mwNone   =& MagicWord::get( MAG_IMG_NONE );
4037                 $mwWidth  =& MagicWord::get( MAG_IMG_WIDTH );
4038                 $mwCenter =& MagicWord::get( MAG_IMG_CENTER );
4039                 $mwFramed =& MagicWord::get( MAG_IMG_FRAMED );
4040                 $caption = '';
4041
4042                 $width = $height = $framed = $thumb = false;
4043                 $manual_thumb = '' ;
4044
4045                 foreach( $part as $key => $val ) {
4046                         if ( $wgUseImageResize && ! is_null( $mwThumb->matchVariableStartToEnd($val) ) ) {
4047                                 $thumb=true;
4048                         } elseif ( ! is_null( $match = $mwManualThumb->matchVariableStartToEnd($val) ) ) {
4049                                 # use manually specified thumbnail
4050                                 $thumb=true;
4051                                 $manual_thumb = $match;
4052                         } elseif ( ! is_null( $mwRight->matchVariableStartToEnd($val) ) ) {
4053                                 # remember to set an alignment, don't render immediately
4054                                 $align = 'right';
4055                         } elseif ( ! is_null( $mwLeft->matchVariableStartToEnd($val) ) ) {
4056                                 # remember to set an alignment, don't render immediately
4057                                 $align = 'left';
4058                         } elseif ( ! is_null( $mwCenter->matchVariableStartToEnd($val) ) ) {
4059                                 # remember to set an alignment, don't render immediately
4060                                 $align = 'center';
4061                         } elseif ( ! is_null( $mwNone->matchVariableStartToEnd($val) ) ) {
4062                                 # remember to set an alignment, don't render immediately
4063                                 $align = 'none';
4064                         } elseif ( $wgUseImageResize && ! is_null( $match = $mwWidth->matchVariableStartToEnd($val) ) ) {
4065                                 wfDebug( "MAG_IMG_WIDTH match: $match\n" );
4066                                 # $match is the image width in pixels
4067                                 if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $match, $m ) ) {
4068                                         $width = intval( $m[1] );
4069                                         $height = intval( $m[2] );
4070                                 } else {
4071                                         $width = intval($match);
4072                                 }
4073                         } elseif ( ! is_null( $mwFramed->matchVariableStartToEnd($val) ) ) {
4074                                 $framed=true;
4075                         } else {
4076                                 $caption = $val;
4077                         }
4078                 }
4079                 # Strip bad stuff out of the alt text
4080                 $alt = $this->replaceLinkHoldersText( $caption );
4081
4082                 # make sure there are no placeholders in thumbnail attributes
4083                 # that are later expanded to html- so expand them now and
4084                 # remove the tags
4085                 $alt = $this->unstrip($alt, $this->mStripState);
4086                 $alt = Sanitizer::stripAllTags( $alt );
4087
4088                 # Linker does the rest
4089                 $sk =& $this->mOptions->getSkin();
4090                 return $sk->makeImageLinkObj( $nt, $caption, $alt, $align, $width, $height, $framed, $thumb, $manual_thumb );
4091         }
4092
4093         /**
4094          * Set a flag in the output object indicating that the content is dynamic and
4095          * shouldn't be cached.
4096          */
4097         function disableCache() {
4098                 wfDebug( "Parser output marked as uncacheable.\n" );
4099                 $this->mOutput->mCacheTime = -1;
4100         }
4101
4102         /**#@+
4103          * Callback from the Sanitizer for expanding items found in HTML attribute
4104          * values, so they can be safely tested and escaped.
4105          * @param string $text
4106          * @param array $args
4107          * @return string
4108          * @private
4109          */
4110         function attributeStripCallback( &$text, $args ) {
4111                 $text = $this->replaceVariables( $text, $args );
4112                 $text = $this->unstripForHTML( $text );
4113                 return $text;
4114         }
4115
4116         function unstripForHTML( $text ) {
4117                 $text = $this->unstrip( $text, $this->mStripState );
4118                 $text = $this->unstripNoWiki( $text, $this->mStripState );
4119                 return $text;
4120         }
4121         /**#@-*/
4122
4123         /**#@+
4124          * Accessor/mutator
4125          */
4126         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4127         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4128         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4129         /**#@-*/
4130
4131         /**#@+
4132          * Accessor
4133          */
4134         function getTags() { return array_keys( $this->mTagHooks ); }
4135         /**#@-*/
4136 }
4137
4138 /**
4139  * @todo document
4140  * @package MediaWiki
4141  */
4142 class ParserOutput
4143 {
4144         var $mText,             # The output text
4145                 $mLanguageLinks,    # List of the full text of language links, in the order they appear
4146                 $mCategories,       # Map of category names to sort keys
4147                 $mContainsOldMagic, # Boolean variable indicating if the input contained variables like {{CURRENTDAY}}
4148                 $mCacheTime,        # Time when this object was generated, or -1 for uncacheable. Used in ParserCache.
4149                 $mVersion,          # Compatibility check
4150                 $mTitleText,        # title text of the chosen language variant
4151                 $mLinks,            # 2-D map of NS/DBK to ID for the links in the document. ID=zero for broken.
4152                 $mTemplates,        # 2-D map of NS/DBK to ID for the template references. ID=zero for broken.
4153                 $mImages,           # DB keys of the images used, in the array key only
4154                 $mExternalLinks,    # External link URLs, in the key only
4155                 $mHTMLtitle,            # Display HTML title
4156                 $mSubtitle,                     # Additional subtitle
4157                 $mNewSection;           # Show a new section link?
4158
4159         function ParserOutput( $text = '', $languageLinks = array(), $categoryLinks = array(),
4160                 $containsOldMagic = false, $titletext = '' )
4161         {
4162                 $this->mText = $text;
4163                 $this->mLanguageLinks = $languageLinks;
4164                 $this->mCategories = $categoryLinks;
4165                 $this->mContainsOldMagic = $containsOldMagic;
4166                 $this->mCacheTime = '';
4167                 $this->mVersion = MW_PARSER_VERSION;
4168                 $this->mTitleText = $titletext;
4169                 $this->mLinks = array();
4170                 $this->mTemplates = array();
4171                 $this->mImages = array();
4172                 $this->mExternalLinks = array();
4173                 $this->mHTMLtitle = "" ;
4174                 $this->mSubtitle = "" ;
4175                 $this->mNewSection = false;
4176         }
4177
4178         function getText()                   { return $this->mText; }
4179         function &getLanguageLinks()          { return $this->mLanguageLinks; }
4180         function getCategoryLinks()          { return array_keys( $this->mCategories ); }
4181         function &getCategories()            { return $this->mCategories; }
4182         function getCacheTime()              { return $this->mCacheTime; }
4183         function getTitleText()              { return $this->mTitleText; }
4184         function &getLinks()                 { return $this->mLinks; }
4185         function &getTemplates()             { return $this->mTemplates; }
4186         function &getImages()                { return $this->mImages; }
4187         function &getExternalLinks()         { return $this->mExternalLinks; }
4188
4189         function containsOldMagic()          { return $this->mContainsOldMagic; }
4190         function setText( $text )            { return wfSetVar( $this->mText, $text ); }
4191         function setLanguageLinks( $ll )     { return wfSetVar( $this->mLanguageLinks, $ll ); }
4192         function setCategoryLinks( $cl )     { return wfSetVar( $this->mCategories, $cl ); }
4193         function setContainsOldMagic( $com ) { return wfSetVar( $this->mContainsOldMagic, $com ); }
4194         function setCacheTime( $t )          { return wfSetVar( $this->mCacheTime, $t ); }
4195         function setTitleText( $t )          { return wfSetVar ($this->mTitleText, $t); }
4196
4197         function addCategory( $c, $sort )    { $this->mCategories[$c] = $sort; }
4198         function addImage( $name )           { $this->mImages[$name] = 1; }
4199         function addLanguageLink( $t )       { $this->mLanguageLinks[] = $t; }
4200         function addExternalLink( $url )     { $this->mExternalLinks[$url] = 1; }
4201
4202         function setNewSection( $value ) {
4203                 $this->mNewSection = (bool)$value;
4204         }
4205         function getNewSection() {
4206                 return (bool)$this->mNewSection;
4207         }
4208
4209         function addLink( $title, $id ) {
4210                 $ns = $title->getNamespace();
4211                 $dbk = $title->getDBkey();
4212                 if ( !isset( $this->mLinks[$ns] ) ) {
4213                         $this->mLinks[$ns] = array();
4214                 }
4215                 $this->mLinks[$ns][$dbk] = $id;
4216         }
4217
4218         function addTemplate( $title, $id ) {
4219                 $ns = $title->getNamespace();
4220                 $dbk = $title->getDBkey();
4221                 if ( !isset( $this->mTemplates[$ns] ) ) {
4222                         $this->mTemplates[$ns] = array();
4223                 }
4224                 $this->mTemplates[$ns][$dbk] = $id;
4225         }
4226
4227         /**
4228          * Return true if this cached output object predates the global or
4229          * per-article cache invalidation timestamps, or if it comes from
4230          * an incompatible older version.
4231          *
4232          * @param string $touched the affected article's last touched timestamp
4233          * @return bool
4234          * @public
4235          */
4236         function expired( $touched ) {
4237                 global $wgCacheEpoch;
4238                 return $this->getCacheTime() == -1 || // parser says it's uncacheable
4239                        $this->getCacheTime() < $touched ||
4240                        $this->getCacheTime() <= $wgCacheEpoch ||
4241                        !isset( $this->mVersion ) ||
4242                        version_compare( $this->mVersion, MW_PARSER_VERSION, "lt" );
4243         }
4244 }
4245
4246 /**
4247  * Set options of the Parser
4248  * @todo document
4249  * @package MediaWiki
4250  */
4251 class ParserOptions
4252 {
4253         # All variables are private
4254         var $mUseTeX;                    # Use texvc to expand <math> tags
4255         var $mUseDynamicDates;           # Use DateFormatter to format dates
4256         var $mInterwikiMagic;            # Interlanguage links are removed and returned in an array
4257         var $mAllowExternalImages;       # Allow external images inline
4258         var $mAllowExternalImagesFrom;   # If not, any exception?
4259         var $mSkin;                      # Reference to the preferred skin
4260         var $mDateFormat;                # Date format index
4261         var $mEditSection;               # Create "edit section" links
4262         var $mNumberHeadings;            # Automatically number headings
4263         var $mAllowSpecialInclusion;     # Allow inclusion of special pages
4264         var $mTidy;                      # Ask for tidy cleanup
4265         var $mInterfaceMessage;          # Which lang to call for PLURAL and GRAMMAR
4266
4267         function getUseTeX()                        { return $this->mUseTeX; }
4268         function getUseDynamicDates()               { return $this->mUseDynamicDates; }
4269         function getInterwikiMagic()                { return $this->mInterwikiMagic; }
4270         function getAllowExternalImages()           { return $this->mAllowExternalImages; }
4271         function getAllowExternalImagesFrom()       { return $this->mAllowExternalImagesFrom; }
4272         function &getSkin()                         { return $this->mSkin; }
4273         function getDateFormat()                    { return $this->mDateFormat; }
4274         function getEditSection()                   { return $this->mEditSection; }
4275         function getNumberHeadings()                { return $this->mNumberHeadings; }
4276         function getAllowSpecialInclusion()         { return $this->mAllowSpecialInclusion; }
4277         function getTidy()                          { return $this->mTidy; }
4278         function getInterfaceMessage()              { return $this->mInterfaceMessage; }
4279
4280         function setUseTeX( $x )                    { return wfSetVar( $this->mUseTeX, $x ); }
4281         function setUseDynamicDates( $x )           { return wfSetVar( $this->mUseDynamicDates, $x ); }
4282         function setInterwikiMagic( $x )            { return wfSetVar( $this->mInterwikiMagic, $x ); }
4283         function setAllowExternalImages( $x )       { return wfSetVar( $this->mAllowExternalImages, $x ); }
4284         function setAllowExternalImagesFrom( $x )   { return wfSetVar( $this->mAllowExternalImagesFrom, $x ); }
4285         function setDateFormat( $x )                { return wfSetVar( $this->mDateFormat, $x ); }
4286         function setEditSection( $x )               { return wfSetVar( $this->mEditSection, $x ); }
4287         function setNumberHeadings( $x )            { return wfSetVar( $this->mNumberHeadings, $x ); }
4288         function setAllowSpecialInclusion( $x )     { return wfSetVar( $this->mAllowSpecialInclusion, $x ); }
4289         function setTidy( $x )                      { return wfSetVar( $this->mTidy, $x); }
4290         function setSkin( &$x ) { $this->mSkin =& $x; }
4291         function setInterfaceMessage( $x )          { return wfSetVar( $this->mInterfaceMessage, $x); }
4292
4293         function ParserOptions() {
4294                 global $wgUser;
4295                 $this->initialiseFromUser( $wgUser );
4296         }
4297
4298         /**
4299          * Get parser options
4300          * @static
4301          */
4302         function newFromUser( &$user ) {
4303                 $popts = new ParserOptions;
4304                 $popts->initialiseFromUser( $user );
4305                 return $popts;
4306         }
4307
4308         /** Get user options */
4309         function initialiseFromUser( &$userInput ) {
4310                 global $wgUseTeX, $wgUseDynamicDates, $wgInterwikiMagic, $wgAllowExternalImages;
4311                 global $wgAllowExternalImagesFrom, $wgAllowSpecialInclusion;
4312                 $fname = 'ParserOptions::initialiseFromUser';
4313                 wfProfileIn( $fname );
4314                 if ( !$userInput ) {
4315                         $user = new User;
4316                         $user->setLoaded( true );
4317                 } else {
4318                         $user =& $userInput;
4319                 }
4320
4321                 $this->mUseTeX = $wgUseTeX;
4322                 $this->mUseDynamicDates = $wgUseDynamicDates;
4323                 $this->mInterwikiMagic = $wgInterwikiMagic;
4324                 $this->mAllowExternalImages = $wgAllowExternalImages;
4325                 $this->mAllowExternalImagesFrom = $wgAllowExternalImagesFrom;
4326                 wfProfileIn( $fname.'-skin' );
4327                 $this->mSkin =& $user->getSkin();
4328                 wfProfileOut( $fname.'-skin' );
4329                 $this->mDateFormat = $user->getOption( 'date' );
4330                 $this->mEditSection = true;
4331                 $this->mNumberHeadings = $user->getOption( 'numberheadings' );
4332                 $this->mAllowSpecialInclusion = $wgAllowSpecialInclusion;
4333                 $this->mTidy = false;
4334                 $this->mInterfaceMessage = false;
4335                 wfProfileOut( $fname );
4336         }
4337 }
4338
4339 /**
4340  * Callback function used by Parser::replaceLinkHolders()
4341  * to substitute link placeholders.
4342  */
4343 function &wfOutputReplaceMatches( $matches ) {
4344         global $wgOutputReplace;
4345         return $wgOutputReplace[$matches[1]];
4346 }
4347
4348 /**
4349  * Return the total number of articles
4350  */
4351 function wfNumberOfArticles() {
4352         global $wgNumberOfArticles;
4353
4354         wfLoadSiteStats();
4355         return $wgNumberOfArticles;
4356 }
4357
4358 /**
4359  * Return the number of files
4360  */
4361 function wfNumberOfFiles() {
4362         $fname = 'wfNumberOfFiles';
4363
4364         wfProfileIn( $fname );
4365         $dbr =& wfGetDB( DB_SLAVE );
4366         $numImages = $dbr->selectField('site_stats', 'ss_images', array(), $fname );
4367         wfProfileOut( $fname );
4368
4369         return $numImages;
4370 }
4371
4372 /**
4373  * Return the number of user accounts
4374  * @return integer
4375  */
4376 function wfNumberOfUsers() {
4377         wfProfileIn( 'wfNumberOfUsers' );
4378         $dbr =& wfGetDB( DB_SLAVE );
4379         $count = $dbr->selectField( 'site_stats', 'ss_users', array(), 'wfNumberOfUsers' );
4380         wfProfileOut( 'wfNumberOfUsers' );
4381         return (int)$count;
4382 }
4383
4384 /**
4385  * Return the total number of pages
4386  * @return integer
4387  */
4388 function wfNumberOfPages() {
4389         wfProfileIn( 'wfNumberOfPages' );
4390         $dbr =& wfGetDB( DB_SLAVE );
4391         $count = $dbr->selectField( 'site_stats', 'ss_total_pages', array(), 'wfNumberOfPages' );
4392         wfProfileOut( 'wfNumberOfPages' );
4393         return (int)$count;
4394 }
4395
4396 /**
4397  * Get various statistics from the database
4398  * @private
4399  */
4400 function wfLoadSiteStats() {
4401         global $wgNumberOfArticles, $wgTotalViews, $wgTotalEdits;
4402         $fname = 'wfLoadSiteStats';
4403
4404         if ( -1 != $wgNumberOfArticles ) return;
4405         $dbr =& wfGetDB( DB_SLAVE );
4406         $s = $dbr->selectRow( 'site_stats',
4407                 array( 'ss_total_views', 'ss_total_edits', 'ss_good_articles' ),
4408                 array( 'ss_row_id' => 1 ), $fname
4409         );
4410
4411         if ( $s === false ) {
4412                 return;
4413         } else {
4414                 $wgTotalViews = $s->ss_total_views;
4415                 $wgTotalEdits = $s->ss_total_edits;
4416                 $wgNumberOfArticles = $s->ss_good_articles;
4417         }
4418 }
4419
4420 /**
4421  * Escape html tags
4422  * Basically replacing " > and < with HTML entities ( &quot;, &gt;, &lt;)
4423  *
4424  * @param $in String: text that might contain HTML tags.
4425  * @return string Escaped string
4426  */
4427 function wfEscapeHTMLTagsOnly( $in ) {
4428         return str_replace(
4429                 array( '"', '>', '<' ),
4430                 array( '&quot;', '&gt;', '&lt;' ),
4431                 $in );
4432 }
4433
4434 ?>