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