includes/Parser.php

   1 <?php
   2 /**
   3  * File for Parser and related classes
   4  *
   5  * @package MediaWiki
   6  * @subpackage Parser
   7  */
   8
   9 /** */
  10 require_once( 'Sanitizer.php' );
  11 require_once( 'HttpFunctions.php' );
  12
  13 /**
  14  * Update this version number when the ParserOutput format
  15  * changes in an incompatible way, so the parser cache
  16  * can automatically discard old data.
  17  */
  18 define( 'MW_PARSER_VERSION', '1.6.0' );
  19
  20 /**
  21  * Variable substitution O(N^2) attack
  22  *
  23  * Without countermeasures, it would be possible to attack the parser by saving
  24  * a page filled with a large number of inclusions of large pages. The size of
  25  * the generated page would be proportional to the square of the input size.
  26  * Hence, we limit the number of inclusions of any given page, thus bringing any
  27  * attack back to O(N).
  28  */
  29
  30 define( 'MAX_INCLUDE_REPEAT', 100 );
  31 define( 'MAX_INCLUDE_SIZE', 1000000 ); // 1 Million
  32
  33 define( 'RLH_FOR_UPDATE', 1 );
  34
  35 # Allowed values for $mOutputType
  36 define( 'OT_HTML', 1 );
  37 define( 'OT_WIKI', 2 );
  38 define( 'OT_MSG' , 3 );
  39
  40 # string parameter for extractTags which will cause it
  41 # to strip HTML comments in addition to regular
  42 # <XML>-style tags. This should not be anything we
  43 # may want to use in wikisyntax
  44 define( 'STRIP_COMMENTS', 'HTMLCommentStrip' );
  45
  46 # Constants needed for external link processing
  47 define( 'HTTP_PROTOCOLS', 'http:\/\/|https:\/\/' );
  48 # Everything except bracket, space, or control characters
  49 define( 'EXT_LINK_URL_CLASS', '[^][<>"\\x00-\\x20\\x7F]' );
  50 # Including space
  51 define( 'EXT_LINK_TEXT_CLASS', '[^\]\\x00-\\x1F\\x7F]' );
  52 define( 'EXT_IMAGE_FNAME_CLASS', '[A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]' );
  53 define( 'EXT_IMAGE_EXTENSIONS', 'gif|png|jpg|jpeg' );
  54 define( 'EXT_LINK_BRACKETED',  '/\[(\b(' . wfUrlProtocols() . ')'.EXT_LINK_URL_CLASS.'+) *('.EXT_LINK_TEXT_CLASS.'*?)\]/S' );
  55 define( 'EXT_IMAGE_REGEX',
  56         '/^('.HTTP_PROTOCOLS.')'.  # Protocol
  57         '('.EXT_LINK_URL_CLASS.'+)\\/'.  # Hostname and path
  58         '('.EXT_IMAGE_FNAME_CLASS.'+)\\.((?i)'.EXT_IMAGE_EXTENSIONS.')$/S' # Filename
  59 );
  60
  61 /**
  62  * PHP Parser
  63  *
  64  * Processes wiki markup
  65  *
  66  * <pre>
  67  * There are three main entry points into the Parser class:
  68  * parse()
  69  *   produces HTML output
  70  * preSaveTransform().
  71  *   produces altered wiki markup.
  72  * transformMsg()
  73  *   performs brace substitution on MediaWiki messages
  74  *
  75  * Globals used:
  76  *    objects:   $wgLang, $wgContLang
  77  *
  78  * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
  79  *
  80  * settings:
  81  *  $wgUseTex*, $wgUseDynamicDates*, $wgInterwikiMagic*,
  82  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  83  *  $wgLocaltimezone, $wgAllowSpecialInclusion*
  84  *
  85  *  * only within ParserOptions
  86  * </pre>
  87  *
  88  * @package MediaWiki
  89  */
  90 class Parser
  91 {
  92         /**#@+
  93          * @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                 $this->mTemplatePath = array();
 159
 160                 wfRunHooks( 'ParserClearState', array( &$this ) );
 161         }
 162
 163         /**
 164          * Accessor for mUniqPrefix.
 165          *
 166          * @public
 167          */
 168         function UniqPrefix() {
 169                 return $this->mUniqPrefix;
 170         }
 171
 172         /**
 173          * Convert wikitext to HTML
 174          * Do not call this function recursively.
 175          *
 176          * @private
 177          * @param string $text Text we want to parse
 178          * @param Title &$title A title object
 179          * @param array $options
 180          * @param boolean $linestart
 181          * @param boolean $clearState
 182          * @param int $revid number to pass in {{REVISIONID}}
 183          * @return ParserOutput a ParserOutput
 184          */
 185         function parse( $text, &$title, $options, $linestart = true, $clearState = true, $revid = null ) {
 186                 /**
 187                  * First pass--just handle <nowiki> sections, pass the rest off
 188                  * to internalParse() which does all the real work.
 189                  */
 190
 191                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
 192                 $fname = 'Parser::parse';
 193                 wfProfileIn( $fname );
 194
 195                 if ( $clearState ) {
 196                         $this->clearState();
 197                 }
 198
 199                 $this->mOptions = $options;
 200                 $this->mTitle =& $title;
 201                 $this->mRevisionId = $revid;
 202                 $this->mOutputType = OT_HTML;
 203
 204                 //$text = $this->strip( $text, $this->mStripState );
 205                 // VOODOO MAGIC FIX! Sometimes the above segfaults in PHP5.
 206                 $x =& $this->mStripState;
 207
 208                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$x ) );
 209                 $text = $this->strip( $text, $x );
 210                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$x ) );
 211
 212                 # Hook to suspend the parser in this state
 213                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$x ) ) ) {
 214                         wfProfileOut( $fname );
 215                         return $text ;
 216                 }
 217
 218                 $text = $this->internalParse( $text );
 219
 220                 $text = $this->unstrip( $text, $this->mStripState );
 221
 222                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 223                 $fixtags = array(
 224                         # french spaces, last one Guillemet-left
 225                         # only if there is something before the space
 226                         '/(.) (?=\\?|:|;|!|\\302\\273)/' => '\\1&nbsp;\\2',
 227                         # french spaces, Guillemet-right
 228                         '/(\\302\\253) /' => '\\1&nbsp;',
 229                         '/<center *>(.*)<\\/center *>/i' => '<div class="center">\\1</div>',
 230                 );
 231                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 232
 233                 # only once and last
 234                 $text = $this->doBlockLevels( $text, $linestart );
 235
 236                 $this->replaceLinkHolders( $text );
 237
 238                 # the position of the parserConvert() call should not be changed. it
 239                 # assumes that the links are all replaced and the only thing left
 240                 # is the <nowiki> mark.
 241                 # Side-effects: this calls $this->mOutput->setTitleText()
 242                 $text = $wgContLang->parserConvert( $text, $this );
 243
 244                 $text = $this->unstripNoWiki( $text, $this->mStripState );
 245
 246                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 247
 248                 $text = Sanitizer::normalizeCharReferences( $text );
 249
 250                 if (($wgUseTidy and $this->mOptions->mTidy) or $wgAlwaysUseTidy) {
 251                         $text = Parser::tidy($text);
 252                 } else {
 253                         # attempt to sanitize at least some nesting problems
 254                         # (bug #2702 and quite a few others)
 255                         $tidyregs = array(
 256                                 # ''Something [http://www.cool.com cool''] -->
 257                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 258                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 259                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 260                                 # fix up an anchor inside another anchor, only
 261                                 # at least for a single single nested link (bug 3695)
 262                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 263                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 264                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 265                                 # contains a div, so fix it up here; replace
 266                                 # div with escaped text
 267                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 268                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 269                                 # remove empty italic or bold tag pairs, some
 270                                 # introduced by rules above
 271                                 '/<([bi])><\/\\1>/' => ''
 272                         );
 273
 274                         $text = preg_replace(
 275                                 array_keys( $tidyregs ),
 276                                 array_values( $tidyregs ),
 277                                 $text );
 278                 }
 279
 280                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 281
 282                 $this->mOutput->setText( $text );
 283                 wfProfileOut( $fname );
 284
 285                 return $this->mOutput;
 286         }
 287
 288         /**
 289          * Get a random string
 290          *
 291          * @private
 292          * @static
 293          */
 294         function getRandomString() {
 295                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 296         }
 297
 298         function &getTitle() { return $this->mTitle; }
 299         function getOptions() { return $this->mOptions; }
 300
 301         /**
 302          * Replaces all occurrences of <$tag>content</$tag> in the text
 303          * with a random marker and returns the new text. the output parameter
 304          * $content will be an associative array filled with data on the form
 305          * $unique_marker => content.
 306          *
 307          * If $content is already set, the additional entries will be appended
 308          * If $tag is set to STRIP_COMMENTS, the function will extract
 309          * <!-- HTML comments -->
 310          *
 311          * @private
 312          * @static
 313          */
 314         function extractTagsAndParams($tag, $text, &$content, &$tags, &$params, $uniq_prefix = ''){
 315                 $rnd = $uniq_prefix . '-' . $tag . Parser::getRandomString();
 316                 if ( !$content ) {
 317                         $content = array( );
 318                 }
 319                 $n = 1;
 320                 $stripped = '';
 321
 322                 if ( !$tags ) {
 323                         $tags = array( );
 324                 }
 325
 326                 if ( !$params ) {
 327                         $params = array( );
 328                 }
 329
 330                 if( $tag == STRIP_COMMENTS ) {
 331                         $start = '/<!--()/';
 332                         $end   = '/-->/';
 333                 } else {
 334                         $start = "/<$tag(\\s+[^>]*|\\s*\/?)>/i";
 335                         $end   = "/<\\/$tag\\s*>/i";
 336                 }
 337
 338                 while ( '' != $text ) {
 339                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 340                         $stripped .= $p[0];
 341                         if( count( $p ) < 3 ) {
 342                                 break;
 343                         }
 344                         $attributes = $p[1];
 345                         $inside     = $p[2];
 346
 347                         // If $attributes ends with '/', we have an empty element tag, <tag />
 348                         if( $tag != STRIP_COMMENTS && substr( $attributes, -1 ) == '/' ) {
 349                                 $attributes = substr( $attributes, 0, -1);
 350                                 $empty = '/';
 351                         } else {
 352                                 $empty = '';
 353                         }
 354
 355                         $marker = $rnd . sprintf('%08X', $n++);
 356                         $stripped .= $marker;
 357
 358                         $tags[$marker] = "<$tag$attributes$empty>";
 359                         $params[$marker] = Sanitizer::decodeTagAttributes( $attributes );
 360
 361                         if ( $empty === '/' ) {
 362                                 // Empty element tag, <tag />
 363                                 $content[$marker] = null;
 364                                 $text = $inside;
 365                         } else {
 366                                 $q = preg_split( $end, $inside, 2 );
 367                                 $content[$marker] = $q[0];
 368                                 if( count( $q ) < 2 ) {
 369                                         # No end tag -- let it run out to the end of the text.
 370                                         break;
 371                                 } else {
 372                                         $text = $q[1];
 373                                 }
 374                         }
 375                 }
 376                 return $stripped;
 377         }
 378
 379         /**
 380          * Wrapper function for extractTagsAndParams
 381          * for cases where $tags and $params isn't needed
 382          * i.e. where tags will never have params, like <nowiki>
 383          *
 384          * @private
 385          * @static
 386          */
 387         function extractTags( $tag, $text, &$content, $uniq_prefix = '' ) {
 388                 $dummy_tags = array();
 389                 $dummy_params = array();
 390
 391                 return Parser::extractTagsAndParams( $tag, $text, $content,
 392                         $dummy_tags, $dummy_params, $uniq_prefix );
 393         }
 394
 395         /**
 396          * Strips and renders nowiki, pre, math, hiero
 397          * If $render is set, performs necessary rendering operations on plugins
 398          * Returns the text, and fills an array with data needed in unstrip()
 399          * If the $state is already a valid strip state, it adds to the state
 400          *
 401          * @param bool $stripcomments when set, HTML comments <!-- like this -->
 402          *  will be stripped in addition to other tags. This is important
 403          *  for section editing, where these comments cause confusion when
 404          *  counting the sections in the wikisource
 405          *
 406          * @private
 407          */
 408         function strip( $text, &$state, $stripcomments = false ) {
 409                 $render = ($this->mOutputType == OT_HTML);
 410                 $html_content = array();
 411                 $nowiki_content = array();
 412                 $math_content = array();
 413                 $pre_content = array();
 414                 $comment_content = array();
 415                 $ext_content = array();
 416                 $ext_tags = array();
 417                 $ext_params = array();
 418                 $gallery_content = array();
 419
 420                 # Replace any instances of the placeholders
 421                 $uniq_prefix = $this->mUniqPrefix;
 422                 #$text = str_replace( $uniq_prefix, wfHtmlEscapeFirst( $uniq_prefix ), $text );
 423
 424                 # html
 425                 global $wgRawHtml;
 426                 if( $wgRawHtml ) {
 427                         $text = Parser::extractTags('html', $text, $html_content, $uniq_prefix);
 428                         foreach( $html_content as $marker => $content ) {
 429                                 if ($render ) {
 430                                         # Raw and unchecked for validity.
 431                                         $html_content[$marker] = $content;
 432                                 } else {
 433                                         $html_content[$marker] = '<html>'.$content.'</html>';
 434                                 }
 435                         }
 436                 }
 437
 438                 # nowiki
 439                 $text = Parser::extractTags('nowiki', $text, $nowiki_content, $uniq_prefix);
 440                 foreach( $nowiki_content as $marker => $content ) {
 441                         if( $render ){
 442                                 $nowiki_content[$marker] = wfEscapeHTMLTagsOnly( $content );
 443                         } else {
 444                                 $nowiki_content[$marker] = '<nowiki>'.$content.'</nowiki>';
 445                         }
 446                 }
 447
 448                 # math
 449                 if( $this->mOptions->getUseTeX() ) {
 450                         $text = Parser::extractTags('math', $text, $math_content, $uniq_prefix);
 451                         foreach( $math_content as $marker => $content ){
 452                                 if( $render ) {
 453                                         $math_content[$marker] = renderMath( $content );
 454                                 } else {
 455                                         $math_content[$marker] = '<math>'.$content.'</math>';
 456                                 }
 457                         }
 458                 }
 459
 460                 # pre
 461                 $text = Parser::extractTags('pre', $text, $pre_content, $uniq_prefix);
 462                 foreach( $pre_content as $marker => $content ){
 463                         if( $render ){
 464                                 $pre_content[$marker] = '<pre>' . wfEscapeHTMLTagsOnly( $content ) . '</pre>';
 465                         } else {
 466                                 $pre_content[$marker] = '<pre>'.$content.'</pre>';
 467                         }
 468                 }
 469
 470                 # gallery
 471                 $text = Parser::extractTags('gallery', $text, $gallery_content, $uniq_prefix);
 472                 foreach( $gallery_content as $marker => $content ) {
 473                         require_once( 'ImageGallery.php' );
 474                         if ( $render ) {
 475                                 $gallery_content[$marker] = $this->renderImageGallery( $content );
 476                         } else {
 477                                 $gallery_content[$marker] = '<gallery>'.$content.'</gallery>';
 478                         }
 479                 }
 480
 481                 # Comments
 482                 $text = Parser::extractTags(STRIP_COMMENTS, $text, $comment_content, $uniq_prefix);
 483                 foreach( $comment_content as $marker => $content ){
 484                         $comment_content[$marker] = '<!--'.$content.'-->';
 485                 }
 486
 487                 # Extensions
 488                 foreach ( $this->mTagHooks as $tag => $callback ) {
 489                         $ext_content[$tag] = array();
 490                         $text = Parser::extractTagsAndParams( $tag, $text, $ext_content[$tag],
 491                                 $ext_tags[$tag], $ext_params[$tag], $uniq_prefix );
 492                         foreach( $ext_content[$tag] as $marker => $content ) {
 493                                 $full_tag = $ext_tags[$tag][$marker];
 494                                 $params = $ext_params[$tag][$marker];
 495                                 if ( $render )
 496                                         $ext_content[$tag][$marker] = call_user_func_array( $callback, array( $content, $params, &$this ) );
 497                                 else {
 498                                         if ( is_null( $content ) ) {
 499                                                 // Empty element tag
 500                                                 $ext_content[$tag][$marker] = $full_tag;
 501                                         } else {
 502                                                 $ext_content[$tag][$marker] = "$full_tag$content</$tag>";
 503                                         }
 504                                 }
 505                         }
 506                 }
 507
 508                 # Unstrip comments unless explicitly told otherwise.
 509                 # (The comments are always stripped prior to this point, so as to
 510                 # not invoke any extension tags / parser hooks contained within
 511                 # a comment.)
 512                 if ( !$stripcomments ) {
 513                         $tempstate = array( 'comment' => $comment_content );
 514                         $text = $this->unstrip( $text, $tempstate );
 515                         $comment_content = array();
 516                 }
 517
 518                 # Merge state with the pre-existing state, if there is one
 519                 if ( $state ) {
 520                         $state['html'] = $state['html'] + $html_content;
 521                         $state['nowiki'] = $state['nowiki'] + $nowiki_content;
 522                         $state['math'] = $state['math'] + $math_content;
 523                         $state['pre'] = $state['pre'] + $pre_content;
 524                         $state['gallery'] = $state['gallery'] + $gallery_content;
 525                         $state['comment'] = $state['comment'] + $comment_content;
 526
 527                         foreach( $ext_content as $tag => $array ) {
 528                                 if ( array_key_exists( $tag, $state ) ) {
 529                                         $state[$tag] = $state[$tag] + $array;
 530                                 }
 531                         }
 532                 } else {
 533                         $state = array(
 534                           'html' => $html_content,
 535                           'nowiki' => $nowiki_content,
 536                           'math' => $math_content,
 537                           'pre' => $pre_content,
 538                           'gallery' => $gallery_content,
 539                           'comment' => $comment_content,
 540                         ) + $ext_content;
 541                 }
 542                 return $text;
 543         }
 544
 545         /**
 546          * restores pre, math, and hiero removed by strip()
 547          *
 548          * always call unstripNoWiki() after this one
 549          * @private
 550          */
 551         function unstrip( $text, &$state ) {
 552                 if ( !is_array( $state ) ) {
 553                         return $text;
 554                 }
 555
 556                 # Must expand in reverse order, otherwise nested tags will be corrupted
 557                 foreach( array_reverse( $state, true ) as $tag => $contentDict ) {
 558                         if( $tag != 'nowiki' && $tag != 'html' ) {
 559                                 foreach( array_reverse( $contentDict, true ) as $uniq => $content ) {
 560                                         $text = str_replace( $uniq, $content, $text );
 561                                 }
 562                         }
 563                 }
 564
 565                 return $text;
 566         }
 567
 568         /**
 569          * always call this after unstrip() to preserve the order
 570          *
 571          * @private
 572          */
 573         function unstripNoWiki( $text, &$state ) {
 574                 if ( !is_array( $state ) ) {
 575                         return $text;
 576                 }
 577
 578                 # Must expand in reverse order, otherwise nested tags will be corrupted
 579                 for ( $content = end($state['nowiki']); $content !== false; $content = prev( $state['nowiki'] ) ) {
 580                         $text = str_replace( key( $state['nowiki'] ), $content, $text );
 581                 }
 582
 583                 global $wgRawHtml;
 584                 if ($wgRawHtml) {
 585                         for ( $content = end($state['html']); $content !== false; $content = prev( $state['html'] ) ) {
 586                                 $text = str_replace( key( $state['html'] ), $content, $text );
 587                         }
 588                 }
 589
 590                 return $text;
 591         }
 592
 593         /**
 594          * Add an item to the strip state
 595          * Returns the unique tag which must be inserted into the stripped text
 596          * The tag will be replaced with the original text in unstrip()
 597          *
 598          * @private
 599          */
 600         function insertStripItem( $text, &$state ) {
 601                 $rnd = $this->mUniqPrefix . '-item' . Parser::getRandomString();
 602                 if ( !$state ) {
 603                         $state = array(
 604                           'html' => array(),
 605                           'nowiki' => array(),
 606                           'math' => array(),
 607                           'pre' => array(),
 608                           'comment' => array(),
 609                           'gallery' => array(),
 610                         );
 611                 }
 612                 $state['item'][$rnd] = $text;
 613                 return $rnd;
 614         }
 615
 616         /**
 617          * Interface with html tidy, used if $wgUseTidy = true.
 618          * If tidy isn't able to correct the markup, the original will be
 619          * returned in all its glory with a warning comment appended.
 620          *
 621          * Either the external tidy program or the in-process tidy extension
 622          * will be used depending on availability. Override the default
 623          * $wgTidyInternal setting to disable the internal if it's not working.
 624          *
 625          * @param string $text Hideous HTML input
 626          * @return string Corrected HTML output
 627          * @public
 628          * @static
 629          */
 630         function tidy( $text ) {
 631                 global $wgTidyInternal;
 632                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 633 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 634 '<head><title>test</title></head><body>'.$text.'</body></html>';
 635                 if( $wgTidyInternal ) {
 636                         $correctedtext = Parser::internalTidy( $wrappedtext );
 637                 } else {
 638                         $correctedtext = Parser::externalTidy( $wrappedtext );
 639                 }
 640                 if( is_null( $correctedtext ) ) {
 641                         wfDebug( "Tidy error detected!\n" );
 642                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 643                 }
 644                 return $correctedtext;
 645         }
 646
 647         /**
 648          * Spawn an external HTML tidy process and get corrected markup back from it.
 649          *
 650          * @private
 651          * @static
 652          */
 653         function externalTidy( $text ) {
 654                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 655                 $fname = 'Parser::externalTidy';
 656                 wfProfileIn( $fname );
 657
 658                 $cleansource = '';
 659                 $opts = ' -utf8';
 660
 661                 $descriptorspec = array(
 662                         0 => array('pipe', 'r'),
 663                         1 => array('pipe', 'w'),
 664                         2 => array('file', '/dev/null', 'a')
 665                 );
 666                 $pipes = array();
 667                 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 668                 if (is_resource($process)) {
 669                         // Theoretically, this style of communication could cause a deadlock
 670                         // here. If the stdout buffer fills up, then writes to stdin could
 671                         // block. This doesn't appear to happen with tidy, because tidy only
 672                         // writes to stdout after it's finished reading from stdin. Search
 673                         // for tidyParseStdin and tidySaveStdout in console/tidy.c
 674                         fwrite($pipes[0], $text);
 675                         fclose($pipes[0]);
 676                         while (!feof($pipes[1])) {
 677                                 $cleansource .= fgets($pipes[1], 1024);
 678                         }
 679                         fclose($pipes[1]);
 680                         proc_close($process);
 681                 }
 682
 683                 wfProfileOut( $fname );
 684
 685                 if( $cleansource == '' && $text != '') {
 686                         // Some kind of error happened, so we couldn't get the corrected text.
 687                         // Just give up; we'll use the source text and append a warning.
 688                         return null;
 689                 } else {
 690                         return $cleansource;
 691                 }
 692         }
 693
 694         /**
 695          * Use the HTML tidy PECL extension to use the tidy library in-process,
 696          * saving the overhead of spawning a new process. Currently written to
 697          * the PHP 4.3.x version of the extension, may not work on PHP 5.
 698          *
 699          * 'pear install tidy' should be able to compile the extension module.
 700          *
 701          * @private
 702          * @static
 703          */
 704         function internalTidy( $text ) {
 705                 global $wgTidyConf;
 706                 $fname = 'Parser::internalTidy';
 707                 wfProfileIn( $fname );
 708
 709                 tidy_load_config( $wgTidyConf );
 710                 tidy_set_encoding( 'utf8' );
 711                 tidy_parse_string( $text );
 712                 tidy_clean_repair();
 713                 if( tidy_get_status() == 2 ) {
 714                         // 2 is magic number for fatal error
 715                         // http://www.php.net/manual/en/function.tidy-get-status.php
 716                         $cleansource = null;
 717                 } else {
 718                         $cleansource = tidy_get_output();
 719                 }
 720                 wfProfileOut( $fname );
 721                 return $cleansource;
 722         }
 723
 724         /**
 725          * parse the wiki syntax used to render tables
 726          *
 727          * @private
 728          */
 729         function doTableStuff ( $t ) {
 730                 $fname = 'Parser::doTableStuff';
 731                 wfProfileIn( $fname );
 732
 733                 $t = explode ( "\n" , $t ) ;
 734                 $td = array () ; # Is currently a td tag open?
 735                 $ltd = array () ; # Was it TD or TH?
 736                 $tr = array () ; # Is currently a tr tag open?
 737                 $ltr = array () ; # tr attributes
 738                 $has_opened_tr = array(); # Did this table open a <tr> element?
 739                 $indent_level = 0; # indent level of the table
 740                 foreach ( $t AS $k => $x )
 741                 {
 742                         $x = trim ( $x ) ;
 743                         $fc = substr ( $x , 0 , 1 ) ;
 744                         if ( preg_match( '/^(:*)\{\|(.*)$/', $x, $matches ) ) {
 745                                 $indent_level = strlen( $matches[1] );
 746
 747                                 $attributes = $this->unstripForHTML( $matches[2] );
 748
 749                                 $t[$k] = str_repeat( '<dl><dd>', $indent_level ) .
 750                                         '<table' . Sanitizer::fixTagAttributes ( $attributes, 'table' ) . '>' ;
 751                                 array_push ( $td , false ) ;
 752                                 array_push ( $ltd , '' ) ;
 753                                 array_push ( $tr , false ) ;
 754                                 array_push ( $ltr , '' ) ;
 755                                 array_push ( $has_opened_tr, false );
 756                         }
 757                         else if ( count ( $td ) == 0 ) { } # Don't do any of the following
 758                         else if ( '|}' == substr ( $x , 0 , 2 ) ) {
 759                                 $z = "</table>" . substr ( $x , 2);
 760                                 $l = array_pop ( $ltd ) ;
 761                                 if ( !array_pop ( $has_opened_tr ) ) $z = "<tr><td></td></tr>" . $z ;
 762                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 763                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 764                                 array_pop ( $ltr ) ;
 765                                 $t[$k] = $z . str_repeat( '</dd></dl>', $indent_level );
 766                         }
 767                         else if ( '|-' == substr ( $x , 0 , 2 ) ) { # Allows for |---------------
 768                                 $x = substr ( $x , 1 ) ;
 769                                 while ( $x != '' && substr ( $x , 0 , 1 ) == '-' ) $x = substr ( $x , 1 ) ;
 770                                 $z = '' ;
 771                                 $l = array_pop ( $ltd ) ;
 772                                 array_pop ( $has_opened_tr );
 773                                 array_push ( $has_opened_tr , true ) ;
 774                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 775                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 776                                 array_pop ( $ltr ) ;
 777                                 $t[$k] = $z ;
 778                                 array_push ( $tr , false ) ;
 779                                 array_push ( $td , false ) ;
 780                                 array_push ( $ltd , '' ) ;
 781                                 $attributes = $this->unstripForHTML( $x );
 782                                 array_push ( $ltr , Sanitizer::fixTagAttributes ( $attributes, 'tr' ) ) ;
 783                         }
 784                         else if ( '|' == $fc || '!' == $fc || '|+' == substr ( $x , 0 , 2 ) ) { # Caption
 785                                 # $x is a table row
 786                                 if ( '|+' == substr ( $x , 0 , 2 ) ) {
 787                                         $fc = '+' ;
 788                                         $x = substr ( $x , 1 ) ;
 789                                 }
 790                                 $after = substr ( $x , 1 ) ;
 791                                 if ( $fc == '!' ) $after = str_replace ( '!!' , '||' , $after ) ;
 792
 793                                 // Split up multiple cells on the same line.
 794                                 // FIXME: This can result in improper nesting of tags processed
 795                                 // by earlier parser steps, but should avoid splitting up eg
 796                                 // attribute values containing literal "||".
 797                                 $after = wfExplodeMarkup( '||', $after );
 798
 799                                 $t[$k] = '' ;
 800
 801                                 # Loop through each table cell
 802                                 foreach ( $after AS $theline )
 803                                 {
 804                                         $z = '' ;
 805                                         if ( $fc != '+' )
 806                                         {
 807                                                 $tra = array_pop ( $ltr ) ;
 808                                                 if ( !array_pop ( $tr ) ) $z = '<tr'.$tra.">\n" ;
 809                                                 array_push ( $tr , true ) ;
 810                                                 array_push ( $ltr , '' ) ;
 811                                                 array_pop ( $has_opened_tr );
 812                                                 array_push ( $has_opened_tr , true ) ;
 813                                         }
 814
 815                                         $l = array_pop ( $ltd ) ;
 816                                         if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 817                                         if ( $fc == '|' ) $l = 'td' ;
 818                                         else if ( $fc == '!' ) $l = 'th' ;
 819                                         else if ( $fc == '+' ) $l = 'caption' ;
 820                                         else $l = '' ;
 821                                         array_push ( $ltd , $l ) ;
 822
 823                                         # Cell parameters
 824                                         $y = explode ( '|' , $theline , 2 ) ;
 825                                         # Note that a '|' inside an invalid link should not
 826                                         # be mistaken as delimiting cell parameters
 827                                         if ( strpos( $y[0], '[[' ) !== false ) {
 828                                                 $y = array ($theline);
 829                                         }
 830                                         if ( count ( $y ) == 1 )
 831                                                 $y = "{$z}<{$l}>{$y[0]}" ;
 832                                         else {
 833                                                 $attributes = $this->unstripForHTML( $y[0] );
 834                                                 $y = "{$z}<{$l}".Sanitizer::fixTagAttributes($attributes, $l).">{$y[1]}" ;
 835                                         }
 836                                         $t[$k] .= $y ;
 837                                         array_push ( $td , true ) ;
 838                                 }
 839                         }
 840                 }
 841
 842                 # Closing open td, tr && table
 843                 while ( count ( $td ) > 0 )
 844                 {
 845                         $l = array_pop ( $ltd ) ;
 846                         if ( array_pop ( $td ) ) $t[] = '</td>' ;
 847                         if ( array_pop ( $tr ) ) $t[] = '</tr>' ;
 848                         if ( !array_pop ( $has_opened_tr ) ) $t[] = "<tr><td></td></tr>" ;
 849                         $t[] = '</table>' ;
 850                 }
 851
 852                 $t = implode ( "\n" , $t ) ;
 853                 # special case: don't return empty table
 854                 if($t == "<table>\n<tr><td></td></tr>\n</table>")
 855                         $t = '';
 856                 wfProfileOut( $fname );
 857                 return $t ;
 858         }
 859
 860         /**
 861          * Helper function for parse() that transforms wiki markup into
 862          * HTML. Only called for $mOutputType == OT_HTML.
 863          *
 864          * @private
 865          */
 866         function internalParse( $text ) {
 867                 $args = array();
 868                 $isMain = true;
 869                 $fname = 'Parser::internalParse';
 870                 wfProfileIn( $fname );
 871
 872                 # Remove <noinclude> tags and <includeonly> sections
 873                 $text = strtr( $text, array( '<onlyinclude>' => '' , '</onlyinclude>' => '' ) );
 874                 $text = strtr( $text, array( '<noinclude>' => '', '</noinclude>' => '') );
 875                 $text = preg_replace( '/<includeonly>.*?<\/includeonly>/s', '', $text );
 876
 877                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ) );
 878                 $text = $this->replaceVariables( $text, $args );
 879
 880                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 881
 882                 $text = $this->doHeadings( $text );
 883                 if($this->mOptions->getUseDynamicDates()) {
 884                         $df =& DateFormatter::getInstance();
 885                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 886                 }
 887                 $text = $this->doAllQuotes( $text );
 888                 $text = $this->replaceInternalLinks( $text );
 889                 $text = $this->replaceExternalLinks( $text );
 890
 891                 # replaceInternalLinks may sometimes leave behind
 892                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 893                 $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
 894
 895                 $text = $this->doMagicLinks( $text );
 896                 $text = $this->doTableStuff( $text );
 897                 $text = $this->formatHeadings( $text, $isMain );
 898
 899                 wfProfileOut( $fname );
 900                 return $text;
 901         }
 902
 903         /**
 904          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 905          * magic external links.
 906          *
 907          * @private
 908          */
 909         function &doMagicLinks( &$text ) {
 910                 $text = $this->magicISBN( $text );
 911                 $text = $this->magicRFC( $text, 'RFC ', 'rfcurl' );
 912                 $text = $this->magicRFC( $text, 'PMID ', 'pubmedurl' );
 913                 return $text;
 914         }
 915
 916         /**
 917          * Parse headers and return html
 918          *
 919          * @private
 920          */
 921         function doHeadings( $text ) {
 922                 $fname = 'Parser::doHeadings';
 923                 wfProfileIn( $fname );
 924                 for ( $i = 6; $i >= 1; --$i ) {
 925                         $h = str_repeat( '=', $i );
 926                         $text = preg_replace( "/^{$h}(.+){$h}(\\s|$)/m",
 927                           "<h{$i}>\\1</h{$i}>\\2", $text );
 928                 }
 929                 wfProfileOut( $fname );
 930                 return $text;
 931         }
 932
 933         /**
 934          * Replace single quotes with HTML markup
 935          * @private
 936          * @return string the altered text
 937          */
 938         function doAllQuotes( $text ) {
 939                 $fname = 'Parser::doAllQuotes';
 940                 wfProfileIn( $fname );
 941                 $outtext = '';
 942                 $lines = explode( "\n", $text );
 943                 foreach ( $lines as $line ) {
 944                         $outtext .= $this->doQuotes ( $line ) . "\n";
 945                 }
 946                 $outtext = substr($outtext, 0,-1);
 947                 wfProfileOut( $fname );
 948                 return $outtext;
 949         }
 950
 951         /**
 952          * Helper function for doAllQuotes()
 953          * @private
 954          */
 955         function doQuotes( $text ) {
 956                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
 957                 if ( count( $arr ) == 1 )
 958                         return $text;
 959                 else
 960                 {
 961                         # First, do some preliminary work. This may shift some apostrophes from
 962                         # being mark-up to being text. It also counts the number of occurrences
 963                         # of bold and italics mark-ups.
 964                         $i = 0;
 965                         $numbold = 0;
 966                         $numitalics = 0;
 967                         foreach ( $arr as $r )
 968                         {
 969                                 if ( ( $i % 2 ) == 1 )
 970                                 {
 971                                         # If there are ever four apostrophes, assume the first is supposed to
 972                                         # be text, and the remaining three constitute mark-up for bold text.
 973                                         if ( strlen( $arr[$i] ) == 4 )
 974                                         {
 975                                                 $arr[$i-1] .= "'";
 976                                                 $arr[$i] = "'''";
 977                                         }
 978                                         # If there are more than 5 apostrophes in a row, assume they're all
 979                                         # text except for the last 5.
 980                                         else if ( strlen( $arr[$i] ) > 5 )
 981                                         {
 982                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
 983                                                 $arr[$i] = "'''''";
 984                                         }
 985                                         # Count the number of occurrences of bold and italics mark-ups.
 986                                         # We are not counting sequences of five apostrophes.
 987                                         if ( strlen( $arr[$i] ) == 2 ) $numitalics++;  else
 988                                         if ( strlen( $arr[$i] ) == 3 ) $numbold++;     else
 989                                         if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
 990                                 }
 991                                 $i++;
 992                         }
 993
 994                         # If there is an odd number of both bold and italics, it is likely
 995                         # that one of the bold ones was meant to be an apostrophe followed
 996                         # by italics. Which one we cannot know for certain, but it is more
 997                         # likely to be one that has a single-letter word before it.
 998                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
 999                         {
1000                                 $i = 0;
1001                                 $firstsingleletterword = -1;
1002                                 $firstmultiletterword = -1;
1003                                 $firstspace = -1;
1004                                 foreach ( $arr as $r )
1005                                 {
1006                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1007                                         {
1008                                                 $x1 = substr ($arr[$i-1], -1);
1009                                                 $x2 = substr ($arr[$i-1], -2, 1);
1010                                                 if ($x1 == ' ') {
1011                                                         if ($firstspace == -1) $firstspace = $i;
1012                                                 } else if ($x2 == ' ') {
1013                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1014                                                 } else {
1015                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1016                                                 }
1017                                         }
1018                                         $i++;
1019                                 }
1020
1021                                 # If there is a single-letter word, use it!
1022                                 if ($firstsingleletterword > -1)
1023                                 {
1024                                         $arr [ $firstsingleletterword ] = "''";
1025                                         $arr [ $firstsingleletterword-1 ] .= "'";
1026                                 }
1027                                 # If not, but there's a multi-letter word, use that one.
1028                                 else if ($firstmultiletterword > -1)
1029                                 {
1030                                         $arr [ $firstmultiletterword ] = "''";
1031                                         $arr [ $firstmultiletterword-1 ] .= "'";
1032                                 }
1033                                 # ... otherwise use the first one that has neither.
1034                                 # (notice that it is possible for all three to be -1 if, for example,
1035                                 # there is only one pentuple-apostrophe in the line)
1036                                 else if ($firstspace > -1)
1037                                 {
1038                                         $arr [ $firstspace ] = "''";
1039                                         $arr [ $firstspace-1 ] .= "'";
1040                                 }
1041                         }
1042
1043                         # Now let's actually convert our apostrophic mush to HTML!
1044                         $output = '';
1045                         $buffer = '';
1046                         $state = '';
1047                         $i = 0;
1048                         foreach ($arr as $r)
1049                         {
1050                                 if (($i % 2) == 0)
1051                                 {
1052                                         if ($state == 'both')
1053                                                 $buffer .= $r;
1054                                         else
1055                                                 $output .= $r;
1056                                 }
1057                                 else
1058                                 {
1059                                         if (strlen ($r) == 2)
1060                                         {
1061                                                 if ($state == 'i')
1062                                                 { $output .= '</i>'; $state = ''; }
1063                                                 else if ($state == 'bi')
1064                                                 { $output .= '</i>'; $state = 'b'; }
1065                                                 else if ($state == 'ib')
1066                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1067                                                 else if ($state == 'both')
1068                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1069                                                 else # $state can be 'b' or ''
1070                                                 { $output .= '<i>'; $state .= 'i'; }
1071                                         }
1072                                         else if (strlen ($r) == 3)
1073                                         {
1074                                                 if ($state == 'b')
1075                                                 { $output .= '</b>'; $state = ''; }
1076                                                 else if ($state == 'bi')
1077                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1078                                                 else if ($state == 'ib')
1079                                                 { $output .= '</b>'; $state = 'i'; }
1080                                                 else if ($state == 'both')
1081                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1082                                                 else # $state can be 'i' or ''
1083                                                 { $output .= '<b>'; $state .= 'b'; }
1084                                         }
1085                                         else if (strlen ($r) == 5)
1086                                         {
1087                                                 if ($state == 'b')
1088                                                 { $output .= '</b><i>'; $state = 'i'; }
1089                                                 else if ($state == 'i')
1090                                                 { $output .= '</i><b>'; $state = 'b'; }
1091                                                 else if ($state == 'bi')
1092                                                 { $output .= '</i></b>'; $state = ''; }
1093                                                 else if ($state == 'ib')
1094                                                 { $output .= '</b></i>'; $state = ''; }
1095                                                 else if ($state == 'both')
1096                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1097                                                 else # ($state == '')
1098                                                 { $buffer = ''; $state = 'both'; }
1099                                         }
1100                                 }
1101                                 $i++;
1102                         }
1103                         # Now close all remaining tags.  Notice that the order is important.
1104                         if ($state == 'b' || $state == 'ib')
1105                                 $output .= '</b>';
1106                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
1107                                 $output .= '</i>';
1108                         if ($state == 'bi')
1109                                 $output .= '</b>';
1110                         if ($state == 'both')
1111                                 $output .= '<b><i>'.$buffer.'</i></b>';
1112                         return $output;
1113                 }
1114         }
1115
1116         /**
1117          * Replace external links
1118          *
1119          * Note: this is all very hackish and the order of execution matters a lot.
1120          * Make sure to run maintenance/parserTests.php if you change this code.
1121          *
1122          * @private
1123          */
1124         function replaceExternalLinks( $text ) {
1125                 global $wgContLang;
1126                 $fname = 'Parser::replaceExternalLinks';
1127                 wfProfileIn( $fname );
1128
1129                 $sk =& $this->mOptions->getSkin();
1130
1131                 $bits = preg_split( EXT_LINK_BRACKETED, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1132
1133                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
1134
1135                 $i = 0;
1136                 while ( $i<count( $bits ) ) {
1137                         $url = $bits[$i++];
1138                         $protocol = $bits[$i++];
1139                         $text = $bits[$i++];
1140                         $trail = $bits[$i++];
1141
1142                         # The characters '<' and '>' (which were escaped by
1143                         # removeHTMLtags()) should not be included in
1144                         # URLs, per RFC 2396.
1145                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1146                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1147                                 $url = substr($url, 0, $m2[0][1]);
1148                         }
1149
1150                         # If the link text is an image URL, replace it with an <img> tag
1151                         # This happened by accident in the original parser, but some people used it extensively
1152                         $img = $this->maybeMakeExternalImage( $text );
1153                         if ( $img !== false ) {
1154                                 $text = $img;
1155                         }
1156
1157                         $dtrail = '';
1158
1159                         # Set linktype for CSS - if URL==text, link is essentially free
1160                         $linktype = ($text == $url) ? 'free' : 'text';
1161
1162                         # No link text, e.g. [http://domain.tld/some.link]
1163                         if ( $text == '' ) {
1164                                 # Autonumber if allowed
1165                                 if ( strpos( HTTP_PROTOCOLS, str_replace('/','\/', $protocol) ) !== false ) {
1166                                         $text = '[' . ++$this->mAutonumber . ']';
1167                                         $linktype = 'autonumber';
1168                                 } else {
1169                                         # Otherwise just use the URL
1170                                         $text = htmlspecialchars( $url );
1171                                         $linktype = 'free';
1172                                 }
1173                         } else {
1174                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1175                                 # Check for trail
1176                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1177                         }
1178
1179                         $text = $wgContLang->markNoConversion($text);
1180
1181                         # Replace &amp; from obsolete syntax with &.
1182                         # All HTML entities will be escaped by makeExternalLink()
1183                         $url = str_replace( '&amp;', '&', $url );
1184
1185                         # Process the trail (i.e. everything after this link up until start of the next link),
1186                         # replacing any non-bracketed links
1187                         $trail = $this->replaceFreeExternalLinks( $trail );
1188
1189                         # Use the encoded URL
1190                         # This means that users can paste URLs directly into the text
1191                         # Funny characters like &ouml; aren't valid in URLs anyway
1192                         # This was changed in August 2004
1193                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype ) . $dtrail . $trail;
1194
1195                         # Register link in the output object.
1196                         # Replace unnecessary URL escape codes with the referenced character
1197                         # This prevents spammers from hiding links from the filters
1198                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1199                         $this->mOutput->addExternalLink( $pasteurized );
1200                 }
1201
1202                 wfProfileOut( $fname );
1203                 return $s;
1204         }
1205
1206         /**
1207          * Replace anything that looks like a URL with a link
1208          * @private
1209          */
1210         function replaceFreeExternalLinks( $text ) {
1211                 global $wgContLang;
1212                 $fname = 'Parser::replaceFreeExternalLinks';
1213                 wfProfileIn( $fname );
1214
1215                 $bits = preg_split( '/(\b(?:' . wfUrlProtocols() . '))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1216                 $s = array_shift( $bits );
1217                 $i = 0;
1218
1219                 $sk =& $this->mOptions->getSkin();
1220
1221                 while ( $i < count( $bits ) ){
1222                         $protocol = $bits[$i++];
1223                         $remainder = $bits[$i++];
1224
1225                         if ( preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1226                                 # Found some characters after the protocol that look promising
1227                                 $url = $protocol . $m[1];
1228                                 $trail = $m[2];
1229
1230                                 # special case: handle urls as url args:
1231                                 # http://www.example.com/foo?=http://www.example.com/bar
1232                                 if(strlen($trail) == 0 &&
1233                                         isset($bits[$i]) &&
1234                                         preg_match('/^'. wfUrlProtocols() . '$/S', $bits[$i]) &&
1235                                         preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $bits[$i + 1], $m ))
1236                                 {
1237                                         # add protocol, arg
1238                                         $url .= $bits[$i] . $bits[$i + 1]; # protocol, url as arg to previous link
1239                                         $i += 2;
1240                                         $trail = $m[2];
1241                                 }
1242
1243                                 # The characters '<' and '>' (which were escaped by
1244                                 # removeHTMLtags()) should not be included in
1245                                 # URLs, per RFC 2396.
1246                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1247                                         $trail = substr($url, $m2[0][1]) . $trail;
1248                                         $url = substr($url, 0, $m2[0][1]);
1249                                 }
1250
1251                                 # Move trailing punctuation to $trail
1252                                 $sep = ',;\.:!?';
1253                                 # If there is no left bracket, then consider right brackets fair game too
1254                                 if ( strpos( $url, '(' ) === false ) {
1255                                         $sep .= ')';
1256                                 }
1257
1258                                 $numSepChars = strspn( strrev( $url ), $sep );
1259                                 if ( $numSepChars ) {
1260                                         $trail = substr( $url, -$numSepChars ) . $trail;
1261                                         $url = substr( $url, 0, -$numSepChars );
1262                                 }
1263
1264                                 # Replace &amp; from obsolete syntax with &.
1265                                 # All HTML entities will be escaped by makeExternalLink()
1266                                 # or maybeMakeExternalImage()
1267                                 $url = str_replace( '&amp;', '&', $url );
1268
1269                                 # Is this an external image?
1270                                 $text = $this->maybeMakeExternalImage( $url );
1271                                 if ( $text === false ) {
1272                                         # Not an image, make a link
1273                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free' );
1274                                         # Register it in the output object...
1275                                         # Replace unnecessary URL escape codes with their equivalent characters
1276                                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1277                                         $this->mOutput->addExternalLink( $pasteurized );
1278                                 }
1279                                 $s .= $text . $trail;
1280                         } else {
1281                                 $s .= $protocol . $remainder;
1282                         }
1283                 }
1284                 wfProfileOut( $fname );
1285                 return $s;
1286         }
1287
1288         /**
1289          * Replace unusual URL escape codes with their equivalent characters
1290          * @param string
1291          * @return string
1292          * @static
1293          * @fixme This can merge genuinely required bits in the path or query string,
1294          *        breaking legit URLs. A proper fix would treat the various parts of
1295          *        the URL differently; as a workaround, just use the output for
1296          *        statistical records, not for actual linking/output.
1297          */
1298         function replaceUnusualEscapes( $url ) {
1299                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1300                         array( 'Parser', 'replaceUnusualEscapesCallback' ), $url );
1301         }
1302
1303         /**
1304          * Callback function used in replaceUnusualEscapes().
1305          * Replaces unusual URL escape codes with their equivalent character
1306          * @static
1307          * @private
1308          */
1309         function replaceUnusualEscapesCallback( $matches ) {
1310                 $char = urldecode( $matches[0] );
1311                 $ord = ord( $char );
1312                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1313                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1314                         // No, shouldn't be escaped
1315                         return $char;
1316                 } else {
1317                         // Yes, leave it escaped
1318                         return $matches[0];
1319                 }
1320         }
1321
1322         /**
1323          * make an image if it's allowed, either through the global
1324          * option or through the exception
1325          * @private
1326          */
1327         function maybeMakeExternalImage( $url ) {
1328                 $sk =& $this->mOptions->getSkin();
1329                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1330                 $imagesexception = !empty($imagesfrom);
1331                 $text = false;
1332                 if ( $this->mOptions->getAllowExternalImages()
1333                      || ( $imagesexception && strpos( $url, $imagesfrom ) === 0 ) ) {
1334                         if ( preg_match( EXT_IMAGE_REGEX, $url ) ) {
1335                                 # Image found
1336                                 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
1337                         }
1338                 }
1339                 return $text;
1340         }
1341
1342         /**
1343          * Process [[ ]] wikilinks
1344          *
1345          * @private
1346          */
1347         function replaceInternalLinks( $s ) {
1348                 global $wgContLang;
1349                 static $fname = 'Parser::replaceInternalLinks' ;
1350
1351                 wfProfileIn( $fname );
1352
1353                 wfProfileIn( $fname.'-setup' );
1354                 static $tc = FALSE;
1355                 # the % is needed to support urlencoded titles as well
1356                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1357
1358                 $sk =& $this->mOptions->getSkin();
1359
1360                 #split the entire text string on occurences of [[
1361                 $a = explode( '[[', ' ' . $s );
1362                 #get the first element (all text up to first [[), and remove the space we added
1363                 $s = array_shift( $a );
1364                 $s = substr( $s, 1 );
1365
1366                 # Match a link having the form [[namespace:link|alternate]]trail
1367                 static $e1 = FALSE;
1368                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1369                 # Match cases where there is no "]]", which might still be images
1370                 static $e1_img = FALSE;
1371                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1372                 # Match the end of a line for a word that's not followed by whitespace,
1373                 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1374                 $e2 = wfMsgForContent( 'linkprefix' );
1375
1376                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1377
1378                 if( is_null( $this->mTitle ) ) {
1379                         wfDebugDieBacktrace( 'nooo' );
1380                 }
1381                 $nottalk = !$this->mTitle->isTalkPage();
1382
1383                 if ( $useLinkPrefixExtension ) {
1384                         if ( preg_match( $e2, $s, $m ) ) {
1385                                 $first_prefix = $m[2];
1386                         } else {
1387                                 $first_prefix = false;
1388                         }
1389                 } else {
1390                         $prefix = '';
1391                 }
1392
1393                 $selflink = $this->mTitle->getPrefixedText();
1394                 wfProfileOut( $fname.'-setup' );
1395
1396                 $checkVariantLink = sizeof($wgContLang->getVariants())>1;
1397                 $useSubpages = $this->areSubpagesAllowed();
1398
1399                 # Loop for each link
1400                 for ($k = 0; isset( $a[$k] ); $k++) {
1401                         $line = $a[$k];
1402                         if ( $useLinkPrefixExtension ) {
1403                                 wfProfileIn( $fname.'-prefixhandling' );
1404                                 if ( preg_match( $e2, $s, $m ) ) {
1405                                         $prefix = $m[2];
1406                                         $s = $m[1];
1407                                 } else {
1408                                         $prefix='';
1409                                 }
1410                                 # first link
1411                                 if($first_prefix) {
1412                                         $prefix = $first_prefix;
1413                                         $first_prefix = false;
1414                                 }
1415                                 wfProfileOut( $fname.'-prefixhandling' );
1416                         }
1417
1418                         $might_be_img = false;
1419
1420                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1421                                 $text = $m[2];
1422                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1423                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1424                                 # the real problem is with the $e1 regex
1425                                 # See bug 1300.
1426                                 #
1427                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1428                                 # and no image is in sight. See bug 2095.
1429                                 #
1430                                 if( $text !== '' &&
1431                                         preg_match( "/^\](.*)/s", $m[3], $n ) &&
1432                                         strpos($text, '[') !== false
1433                                 )
1434                                 {
1435                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1436                                         $m[3] = $n[1];
1437                                 }
1438                                 # fix up urlencoded title texts
1439                                 if(preg_match('/%/', $m[1] ))
1440                                         # Should anchors '#' also be rejected?
1441                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1442                                 $trail = $m[3];
1443                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1444                                 $might_be_img = true;
1445                                 $text = $m[2];
1446                                 if(preg_match('/%/', $m[1] )) $m[1] = urldecode($m[1]);
1447                                 $trail = "";
1448                         } else { # Invalid form; output directly
1449                                 $s .= $prefix . '[[' . $line ;
1450                                 continue;
1451                         }
1452
1453                         # Don't allow internal links to pages containing
1454                         # PROTO: where PROTO is a valid URL protocol; these
1455                         # should be external links.
1456                         if (preg_match('/^(\b(?:' . wfUrlProtocols() . '))/', $m[1])) {
1457                                 $s .= $prefix . '[[' . $line ;
1458                                 continue;
1459                         }
1460
1461                         # Make subpage if necessary
1462                         if( $useSubpages ) {
1463                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1464                         } else {
1465                                 $link = $m[1];
1466                         }
1467
1468                         $noforce = (substr($m[1], 0, 1) != ':');
1469                         if (!$noforce) {
1470                                 # Strip off leading ':'
1471                                 $link = substr($link, 1);
1472                         }
1473
1474                         $nt = Title::newFromText( $this->unstripNoWiki($link, $this->mStripState) );
1475                         if( !$nt ) {
1476                                 $s .= $prefix . '[[' . $line;
1477                                 continue;
1478                         }
1479
1480                         #check other language variants of the link
1481                         #if the article does not exist
1482                         if( $checkVariantLink
1483                             && $nt->getArticleID() == 0 ) {
1484                                 $wgContLang->findVariantLink($link, $nt);
1485                         }
1486
1487                         $ns = $nt->getNamespace();
1488                         $iw = $nt->getInterWiki();
1489
1490                         if ($might_be_img) { # if this is actually an invalid link
1491                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1492                                         $found = false;
1493                                         while (isset ($a[$k+1]) ) {
1494                                                 #look at the next 'line' to see if we can close it there
1495                                                 $spliced = array_splice( $a, $k + 1, 1 );
1496                                                 $next_line = array_shift( $spliced );
1497                                                 if( preg_match("/^(.*?]].*?)]](.*)$/sD", $next_line, $m) ) {
1498                                                 # the first ]] closes the inner link, the second the image
1499                                                         $found = true;
1500                                                         $text .= '[[' . $m[1];
1501                                                         $trail = $m[2];
1502                                                         break;
1503                                                 } elseif( preg_match("/^.*?]].*$/sD", $next_line, $m) ) {
1504                                                         #if there's exactly one ]] that's fine, we'll keep looking
1505                                                         $text .= '[[' . $m[0];
1506                                                 } else {
1507                                                         #if $next_line is invalid too, we need look no further
1508                                                         $text .= '[[' . $next_line;
1509                                                         break;
1510                                                 }
1511                                         }
1512                                         if ( !$found ) {
1513                                                 # we couldn't find the end of this imageLink, so output it raw
1514                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1515                                                 $text = $this->replaceInternalLinks($text);
1516                                                 $s .= $prefix . '[[' . $link . '|' . $text;
1517                                                 # note: no $trail, because without an end, there *is* no trail
1518                                                 continue;
1519                                         }
1520                                 } else { #it's not an image, so output it raw
1521                                         $s .= $prefix . '[[' . $link . '|' . $text;
1522                                         # note: no $trail, because without an end, there *is* no trail
1523                                         continue;
1524                                 }
1525                         }
1526
1527                         $wasblank = ( '' == $text );
1528                         if( $wasblank ) $text = $link;
1529
1530
1531                         # Link not escaped by : , create the various objects
1532                         if( $noforce ) {
1533
1534                                 # Interwikis
1535                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1536                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1537                                         $s = rtrim($s . "\n");
1538                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1539                                         continue;
1540                                 }
1541
1542                                 if ( $ns == NS_IMAGE ) {
1543                                         wfProfileIn( "$fname-image" );
1544                                         if ( !wfIsBadImage( $nt->getDBkey() ) ) {
1545                                                 # recursively parse links inside the image caption
1546                                                 # actually, this will parse them in any other parameters, too,
1547                                                 # but it might be hard to fix that, and it doesn't matter ATM
1548                                                 $text = $this->replaceExternalLinks($text);
1549                                                 $text = $this->replaceInternalLinks($text);
1550
1551                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1552                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text ) ) . $trail;
1553                                                 $this->mOutput->addImage( $nt->getDBkey() );
1554
1555                                                 wfProfileOut( "$fname-image" );
1556                                                 continue;
1557                                         }
1558                                         wfProfileOut( "$fname-image" );
1559
1560                                 }
1561
1562                                 if ( $ns == NS_CATEGORY ) {
1563                                         wfProfileIn( "$fname-category" );
1564                                         $s = rtrim($s . "\n"); # bug 87
1565
1566                                         if ( $wasblank ) {
1567                                                 if ( $this->mTitle->getNamespace() == NS_CATEGORY ) {
1568                                                         $sortkey = $this->mTitle->getText();
1569                                                 } else {
1570                                                         $sortkey = $this->mTitle->getPrefixedText();
1571                                                 }
1572                                         } else {
1573                                                 $sortkey = $text;
1574                                         }
1575                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1576                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1577                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1578
1579                                         /**
1580                                          * Strip the whitespace Category links produce, see bug 87
1581                                          * @todo We might want to use trim($tmp, "\n") here.
1582                                          */
1583                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1584
1585                                         wfProfileOut( "$fname-category" );
1586                                         continue;
1587                                 }
1588                         }
1589
1590                         if( ( $nt->getPrefixedText() === $selflink ) &&
1591                             ( $nt->getFragment() === '' ) ) {
1592                                 # Self-links are handled specially; generally de-link and change to bold.
1593                                 $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1594                                 continue;
1595                         }
1596
1597                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1598                         if( $ns == NS_MEDIA ) {
1599                                 $link = $sk->makeMediaLinkObj( $nt, $text );
1600                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1601                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1602                                 $this->mOutput->addImage( $nt->getDBkey() );
1603                                 continue;
1604                         } elseif( $ns == NS_SPECIAL ) {
1605                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1606                                 continue;
1607                         } elseif( $ns == NS_IMAGE ) {
1608                                 $img = Image::newFromTitle( $nt );
1609                                 if( $img->exists() ) {
1610                                         // Force a blue link if the file exists; may be a remote
1611                                         // upload on the shared repository, and we want to see its
1612                                         // auto-generated page.
1613                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1614                                         continue;
1615                                 }
1616                         }
1617                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1618                 }
1619                 wfProfileOut( $fname );
1620                 return $s;
1621         }
1622
1623         /**
1624          * Make a link placeholder. The text returned can be later resolved to a real link with
1625          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1626          * parsing of interwiki links, and secondly to allow all extistence checks and
1627          * article length checks (for stub links) to be bundled into a single query.
1628          *
1629          */
1630         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1631                 if ( ! is_object($nt) ) {
1632                         # Fail gracefully
1633                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1634                 } else {
1635                         # Separate the link trail from the rest of the link
1636                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1637
1638                         if ( $nt->isExternal() ) {
1639                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1640                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1641                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1642                         } else {
1643                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1644                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1645                                 $this->mLinkHolders['queries'][] = $query;
1646                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1647                                 $this->mLinkHolders['titles'][] = $nt;
1648
1649                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1650                         }
1651                 }
1652                 return $retVal;
1653         }
1654
1655         /**
1656          * Render a forced-blue link inline; protect against double expansion of
1657          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1658          * Since this little disaster has to split off the trail text to avoid
1659          * breaking URLs in the following text without breaking trails on the
1660          * wiki links, it's been made into a horrible function.
1661          *
1662          * @param Title $nt
1663          * @param string $text
1664          * @param string $query
1665          * @param string $trail
1666          * @param string $prefix
1667          * @return string HTML-wikitext mix oh yuck
1668          */
1669         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1670                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1671                 $sk =& $this->mOptions->getSkin();
1672                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1673                 return $this->armorLinks( $link ) . $trail;
1674         }
1675
1676         /**
1677          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1678          * going to go through further parsing steps before inline URL expansion.
1679          *
1680          * In particular this is important when using action=render, which causes
1681          * full URLs to be included.
1682          *
1683          * Oh man I hate our multi-layer parser!
1684          *
1685          * @param string more-or-less HTML
1686          * @return string less-or-more HTML with NOPARSE bits
1687          */
1688         function armorLinks( $text ) {
1689                 return preg_replace( "/\b(" . wfUrlProtocols() . ')/',
1690                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1691         }
1692
1693         /**
1694          * Return true if subpage links should be expanded on this page.
1695          * @return bool
1696          */
1697         function areSubpagesAllowed() {
1698                 # Some namespaces don't allow subpages
1699                 global $wgNamespacesWithSubpages;
1700                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1701         }
1702
1703         /**
1704          * Handle link to subpage if necessary
1705          * @param string $target the source of the link
1706          * @param string &$text the link text, modified as necessary
1707          * @return string the full name of the link
1708          * @private
1709          */
1710         function maybeDoSubpageLink($target, &$text) {
1711                 # Valid link forms:
1712                 # Foobar -- normal
1713                 # :Foobar -- override special treatment of prefix (images, language links)
1714                 # /Foobar -- convert to CurrentPage/Foobar
1715                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1716                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1717                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1718
1719                 $fname = 'Parser::maybeDoSubpageLink';
1720                 wfProfileIn( $fname );
1721                 $ret = $target; # default return value is no change
1722
1723                 # Some namespaces don't allow subpages,
1724                 # so only perform processing if subpages are allowed
1725                 if( $this->areSubpagesAllowed() ) {
1726                         # Look at the first character
1727                         if( $target != '' && $target{0} == '/' ) {
1728                                 # / at end means we don't want the slash to be shown
1729                                 if( substr( $target, -1, 1 ) == '/' ) {
1730                                         $target = substr( $target, 1, -1 );
1731                                         $noslash = $target;
1732                                 } else {
1733                                         $noslash = substr( $target, 1 );
1734                                 }
1735
1736                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash);
1737                                 if( '' === $text ) {
1738                                         $text = $target;
1739                                 } # this might be changed for ugliness reasons
1740                         } else {
1741                                 # check for .. subpage backlinks
1742                                 $dotdotcount = 0;
1743                                 $nodotdot = $target;
1744                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1745                                         ++$dotdotcount;
1746                                         $nodotdot = substr( $nodotdot, 3 );
1747                                 }
1748                                 if($dotdotcount > 0) {
1749                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1750                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1751                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1752                                                 # / at the end means don't show full path
1753                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1754                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1755                                                         if( '' === $text ) {
1756                                                                 $text = $nodotdot;
1757                                                         }
1758                                                 }
1759                                                 $nodotdot = trim( $nodotdot );
1760                                                 if( $nodotdot != '' ) {
1761                                                         $ret .= '/' . $nodotdot;
1762                                                 }
1763                                         }
1764                                 }
1765                         }
1766                 }
1767
1768                 wfProfileOut( $fname );
1769                 return $ret;
1770         }
1771
1772         /**#@+
1773          * Used by doBlockLevels()
1774          * @private
1775          */
1776         /* private */ function closeParagraph() {
1777                 $result = '';
1778                 if ( '' != $this->mLastSection ) {
1779                         $result = '</' . $this->mLastSection  . ">\n";
1780                 }
1781                 $this->mInPre = false;
1782                 $this->mLastSection = '';
1783                 return $result;
1784         }
1785         # getCommon() returns the length of the longest common substring
1786         # of both arguments, starting at the beginning of both.
1787         #
1788         /* private */ function getCommon( $st1, $st2 ) {
1789                 $fl = strlen( $st1 );
1790                 $shorter = strlen( $st2 );
1791                 if ( $fl < $shorter ) { $shorter = $fl; }
1792
1793                 for ( $i = 0; $i < $shorter; ++$i ) {
1794                         if ( $st1{$i} != $st2{$i} ) { break; }
1795                 }
1796                 return $i;
1797         }
1798         # These next three functions open, continue, and close the list
1799         # element appropriate to the prefix character passed into them.
1800         #
1801         /* private */ function openList( $char ) {
1802                 $result = $this->closeParagraph();
1803
1804                 if ( '*' == $char ) { $result .= '<ul><li>'; }
1805                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
1806                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
1807                 else if ( ';' == $char ) {
1808                         $result .= '<dl><dt>';
1809                         $this->mDTopen = true;
1810                 }
1811                 else { $result = '<!-- ERR 1 -->'; }
1812
1813                 return $result;
1814         }
1815
1816         /* private */ function nextItem( $char ) {
1817                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
1818                 else if ( ':' == $char || ';' == $char ) {
1819                         $close = '</dd>';
1820                         if ( $this->mDTopen ) { $close = '</dt>'; }
1821                         if ( ';' == $char ) {
1822                                 $this->mDTopen = true;
1823                                 return $close . '<dt>';
1824                         } else {
1825                                 $this->mDTopen = false;
1826                                 return $close . '<dd>';
1827                         }
1828                 }
1829                 return '<!-- ERR 2 -->';
1830         }
1831
1832         /* private */ function closeList( $char ) {
1833                 if ( '*' == $char ) { $text = '</li></ul>'; }
1834                 else if ( '#' == $char ) { $text = '</li></ol>'; }
1835                 else if ( ':' == $char ) {
1836                         if ( $this->mDTopen ) {
1837                                 $this->mDTopen = false;
1838                                 $text = '</dt></dl>';
1839                         } else {
1840                                 $text = '</dd></dl>';
1841                         }
1842                 }
1843                 else {  return '<!-- ERR 3 -->'; }
1844                 return $text."\n";
1845         }
1846         /**#@-*/
1847
1848         /**
1849          * Make lists from lines starting with ':', '*', '#', etc.
1850          *
1851          * @private
1852          * @return string the lists rendered as HTML
1853          */
1854         function doBlockLevels( $text, $linestart ) {
1855                 $fname = 'Parser::doBlockLevels';
1856                 wfProfileIn( $fname );
1857
1858                 # Parsing through the text line by line.  The main thing
1859                 # happening here is handling of block-level elements p, pre,
1860                 # and making lists from lines starting with * # : etc.
1861                 #
1862                 $textLines = explode( "\n", $text );
1863
1864                 $lastPrefix = $output = '';
1865                 $this->mDTopen = $inBlockElem = false;
1866                 $prefixLength = 0;
1867                 $paragraphStack = false;
1868
1869                 if ( !$linestart ) {
1870                         $output .= array_shift( $textLines );
1871                 }
1872                 foreach ( $textLines as $oLine ) {
1873                         $lastPrefixLength = strlen( $lastPrefix );
1874                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
1875                         $preOpenMatch = preg_match('/<pre/i', $oLine );
1876                         if ( !$this->mInPre ) {
1877                                 # Multiple prefixes may abut each other for nested lists.
1878                                 $prefixLength = strspn( $oLine, '*#:;' );
1879                                 $pref = substr( $oLine, 0, $prefixLength );
1880
1881                                 # eh?
1882                                 $pref2 = str_replace( ';', ':', $pref );
1883                                 $t = substr( $oLine, $prefixLength );
1884                                 $this->mInPre = !empty($preOpenMatch);
1885                         } else {
1886                                 # Don't interpret any other prefixes in preformatted text
1887                                 $prefixLength = 0;
1888                                 $pref = $pref2 = '';
1889                                 $t = $oLine;
1890                         }
1891
1892                         # List generation
1893                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
1894                                 # Same as the last item, so no need to deal with nesting or opening stuff
1895                                 $output .= $this->nextItem( substr( $pref, -1 ) );
1896                                 $paragraphStack = false;
1897
1898                                 if ( substr( $pref, -1 ) == ';') {
1899                                         # The one nasty exception: definition lists work like this:
1900                                         # ; title : definition text
1901                                         # So we check for : in the remainder text to split up the
1902                                         # title and definition, without b0rking links.
1903                                         $term = $t2 = '';
1904                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
1905                                                 $t = $t2;
1906                                                 $output .= $term . $this->nextItem( ':' );
1907                                         }
1908                                 }
1909                         } elseif( $prefixLength || $lastPrefixLength ) {
1910                                 # Either open or close a level...
1911                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
1912                                 $paragraphStack = false;
1913
1914                                 while( $commonPrefixLength < $lastPrefixLength ) {
1915                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
1916                                         --$lastPrefixLength;
1917                                 }
1918                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
1919                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
1920                                 }
1921                                 while ( $prefixLength > $commonPrefixLength ) {
1922                                         $char = substr( $pref, $commonPrefixLength, 1 );
1923                                         $output .= $this->openList( $char );
1924
1925                                         if ( ';' == $char ) {
1926                                                 # FIXME: This is dupe of code above
1927                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
1928                                                         $t = $t2;
1929                                                         $output .= $term . $this->nextItem( ':' );
1930                                                 }
1931                                         }
1932                                         ++$commonPrefixLength;
1933                                 }
1934                                 $lastPrefix = $pref2;
1935                         }
1936                         if( 0 == $prefixLength ) {
1937                                 wfProfileIn( "$fname-paragraph" );
1938                                 # No prefix (not in list)--go to paragraph mode
1939                                 // XXX: use a stack for nestable elements like span, table and div
1940                                 $openmatch = preg_match('/(<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
1941                                 $closematch = preg_match(
1942                                         '/(<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
1943                                         '<td|<th|<div|<\\/div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul)/iS', $t );
1944                                 if ( $openmatch or $closematch ) {
1945                                         $paragraphStack = false;
1946                                         # TODO bug 5718: paragraph closed
1947                                         $output .= $this->closeParagraph();
1948                                         if ( $preOpenMatch and !$preCloseMatch ) {
1949                                                 $this->mInPre = true;
1950                                         }
1951                                         if ( $closematch ) {
1952                                                 $inBlockElem = false;
1953                                         } else {
1954                                                 $inBlockElem = true;
1955                                         }
1956                                 } else if ( !$inBlockElem && !$this->mInPre ) {
1957                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
1958                                                 // pre
1959                                                 if ($this->mLastSection != 'pre') {
1960                                                         $paragraphStack = false;
1961                                                         $output .= $this->closeParagraph().'<pre>';
1962                                                         $this->mLastSection = 'pre';
1963                                                 }
1964                                                 $t = substr( $t, 1 );
1965                                         } else {
1966                                                 // paragraph
1967                                                 if ( '' == trim($t) ) {
1968                                                         if ( $paragraphStack ) {
1969                                                                 $output .= $paragraphStack.'<br />';
1970                                                                 $paragraphStack = false;
1971                                                                 $this->mLastSection = 'p';
1972                                                         } else {
1973                                                                 if ($this->mLastSection != 'p' ) {
1974                                                                         $output .= $this->closeParagraph();
1975                                                                         $this->mLastSection = '';
1976                                                                         $paragraphStack = '<p>';
1977                                                                 } else {
1978                                                                         $paragraphStack = '</p><p>';
1979                                                                 }
1980                                                         }
1981                                                 } else {
1982                                                         if ( $paragraphStack ) {
1983                                                                 $output .= $paragraphStack;
1984                                                                 $paragraphStack = false;
1985                                                                 $this->mLastSection = 'p';
1986                                                         } else if ($this->mLastSection != 'p') {
1987                                                                 $output .= $this->closeParagraph().'<p>';
1988                                                                 $this->mLastSection = 'p';
1989                                                         }
1990                                                 }
1991                                         }
1992                                 }
1993                                 wfProfileOut( "$fname-paragraph" );
1994                         }
1995                         // somewhere above we forget to get out of pre block (bug 785)
1996                         if($preCloseMatch && $this->mInPre) {
1997                                 $this->mInPre = false;
1998                         }
1999                         if ($paragraphStack === false) {
2000                                 $output .= $t."\n";
2001                         }
2002                 }
2003                 while ( $prefixLength ) {
2004                         $output .= $this->closeList( $pref2{$prefixLength-1} );
2005                         --$prefixLength;
2006                 }
2007                 if ( '' != $this->mLastSection ) {
2008                         $output .= '</' . $this->mLastSection . '>';
2009                         $this->mLastSection = '';
2010                 }
2011
2012                 wfProfileOut( $fname );
2013                 return $output;
2014         }
2015
2016         /**
2017          * Split up a string on ':', ignoring any occurences inside
2018          * <a>..</a> or <span>...</span>
2019          * @param string $str the string to split
2020          * @param string &$before set to everything before the ':'
2021          * @param string &$after set to everything after the ':'
2022          * return string the position of the ':', or false if none found
2023          */
2024         function findColonNoLinks($str, &$before, &$after) {
2025                 # I wonder if we should make this count all tags, not just <a>
2026                 # and <span>. That would prevent us from matching a ':' that
2027                 # comes in the middle of italics other such formatting....
2028                 # -- Wil
2029                 $fname = 'Parser::findColonNoLinks';
2030                 wfProfileIn( $fname );
2031                 $pos = 0;
2032                 do {
2033                         $colon = strpos($str, ':', $pos);
2034
2035                         if ($colon !== false) {
2036                                 $before = substr($str, 0, $colon);
2037                                 $after = substr($str, $colon + 1);
2038
2039                                 # Skip any ':' within <a> or <span> pairs
2040                                 $a = substr_count($before, '<a');
2041                                 $s = substr_count($before, '<span');
2042                                 $ca = substr_count($before, '</a>');
2043                                 $cs = substr_count($before, '</span>');
2044
2045                                 if ($a <= $ca and $s <= $cs) {
2046                                         # Tags are balanced before ':'; ok
2047                                         break;
2048                                 }
2049                                 $pos = $colon + 1;
2050                         }
2051                 } while ($colon !== false);
2052                 wfProfileOut( $fname );
2053                 return $colon;
2054         }
2055
2056         /**
2057          * Return value of a magic variable (like PAGENAME)
2058          *
2059          * @private
2060          */
2061         function getVariableValue( $index ) {
2062                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2063
2064                 /**
2065                  * Some of these require message or data lookups and can be
2066                  * expensive to check many times.
2067                  */
2068                 static $varCache = array();
2069                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$varCache ) ) )
2070                         if ( isset( $varCache[$index] ) )
2071                                 return $varCache[$index];
2072
2073                 $ts = time();
2074                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2075
2076                 switch ( $index ) {
2077                         case MAG_CURRENTMONTH:
2078                                 return $varCache[$index] = $wgContLang->formatNum( date( 'm', $ts ) );
2079                         case MAG_CURRENTMONTHNAME:
2080                                 return $varCache[$index] = $wgContLang->getMonthName( date( 'n', $ts ) );
2081                         case MAG_CURRENTMONTHNAMEGEN:
2082                                 return $varCache[$index] = $wgContLang->getMonthNameGen( date( 'n', $ts ) );
2083                         case MAG_CURRENTMONTHABBREV:
2084                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( date( 'n', $ts ) );
2085                         case MAG_CURRENTDAY:
2086                                 return $varCache[$index] = $wgContLang->formatNum( date( 'j', $ts ) );
2087                         case MAG_CURRENTDAY2:
2088                                 return $varCache[$index] = $wgContLang->formatNum( date( 'd', $ts ) );
2089                         case MAG_PAGENAME:
2090                                 return $this->mTitle->getText();
2091                         case MAG_PAGENAMEE:
2092                                 return $this->mTitle->getPartialURL();
2093                         case MAG_FULLPAGENAME:
2094                                 return $this->mTitle->getPrefixedText();
2095                         case MAG_FULLPAGENAMEE:
2096                                 return $this->mTitle->getPrefixedURL();
2097                         case MAG_SUBPAGENAME:
2098                                 return $this->mTitle->getSubpageText();
2099                         case MAG_SUBPAGENAMEE:
2100                                 return $this->mTitle->getSubpageUrlForm();
2101                         case MAG_TALKPAGENAME:
2102                                 if( $this->mTitle->canTalk() ) {
2103                                         $talkPage = $this->mTitle->getTalkPage();
2104                                         return $talkPage->getPrefixedText();
2105                                 } else {
2106                                         return '';
2107                                 }
2108                         case MAG_TALKPAGENAMEE:
2109                                 if( $this->mTitle->canTalk() ) {
2110                                         $talkPage = $this->mTitle->getTalkPage();
2111                                         return $talkPage->getPrefixedUrl();
2112                                 } else {
2113                                         return '';
2114                                 }
2115                         case MAG_SUBJECTPAGENAME:
2116                                 $subjPage = $this->mTitle->getSubjectPage();
2117                                 return $subjPage->getPrefixedText();
2118                         case MAG_SUBJECTPAGENAMEE:
2119                                 $subjPage = $this->mTitle->getSubjectPage();
2120                                 return $subjPage->getPrefixedUrl();
2121                         case MAG_REVISIONID:
2122                                 return $this->mRevisionId;
2123                         case MAG_NAMESPACE:
2124                                 return $wgContLang->getNsText( $this->mTitle->getNamespace() );
2125                         case MAG_NAMESPACEE:
2126                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2127                         case MAG_TALKSPACE:
2128                                 return $this->mTitle->canTalk() ? $this->mTitle->getTalkNsText() : '';
2129                         case MAG_TALKSPACEE:
2130                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2131                         case MAG_SUBJECTSPACE:
2132                                 return $this->mTitle->getSubjectNsText();
2133                         case MAG_SUBJECTSPACEE:
2134                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2135                         case MAG_CURRENTDAYNAME:
2136                                 return $varCache[$index] = $wgContLang->getWeekdayName( date( 'w', $ts ) + 1 );
2137                         case MAG_CURRENTYEAR:
2138                                 return $varCache[$index] = $wgContLang->formatNum( date( 'Y', $ts ), true );
2139                         case MAG_CURRENTTIME:
2140                                 return $varCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2141                         case MAG_CURRENTWEEK:
2142                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2143                                 // int to remove the padding
2144                                 return $varCache[$index] = $wgContLang->formatNum( (int)date( 'W', $ts ) );
2145                         case MAG_CURRENTDOW:
2146                                 return $varCache[$index] = $wgContLang->formatNum( date( 'w', $ts ) );
2147                         case MAG_NUMBEROFARTICLES:
2148                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfArticles() );
2149                         case MAG_NUMBEROFFILES:
2150                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfFiles() );
2151                         case MAG_SITENAME:
2152                                 return $wgSitename;
2153                         case MAG_SERVER:
2154                                 return $wgServer;
2155                         case MAG_SERVERNAME:
2156                                 return $wgServerName;
2157                         case MAG_SCRIPTPATH:
2158                                 return $wgScriptPath;
2159                         default:
2160                                 $ret = null;
2161                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$varCache, &$index, &$ret ) ) )
2162                                         return $ret;
2163                                 else
2164                                         return null;
2165                 }
2166         }
2167
2168         /**
2169          * initialise the magic variables (like CURRENTMONTHNAME)
2170          *
2171          * @private
2172          */
2173         function initialiseVariables() {
2174                 $fname = 'Parser::initialiseVariables';
2175                 wfProfileIn( $fname );
2176                 global $wgVariableIDs;
2177                 $this->mVariables = array();
2178                 foreach ( $wgVariableIDs as $id ) {
2179                         $mw =& MagicWord::get( $id );
2180                         $mw->addToArray( $this->mVariables, $id );
2181                 }
2182                 wfProfileOut( $fname );
2183         }
2184
2185         /**
2186          * parse any parentheses in format ((title|part|part))
2187          * and call callbacks to get a replacement text for any found piece
2188          *
2189          * @param string $text The text to parse
2190          * @param array $callbacks rules in form:
2191          *     '{' => array(                            # opening parentheses
2192          *                                      'end' => '}',   # closing parentheses
2193          *                                      'cb' => array(2 => callback,    # replacement callback to call if {{..}} is found
2194          *                                                                4 => callback         # replacement callback to call if {{{{..}}}} is found
2195          *                                                                )
2196          *                                      )
2197          * @private
2198          */
2199         function replace_callback ($text, $callbacks) {
2200                 $openingBraceStack = array();   # this array will hold a stack of parentheses which are not closed yet
2201                 $lastOpeningBrace = -1;                 # last not closed parentheses
2202
2203                 for ($i = 0; $i < strlen($text); $i++) {
2204                         # check for any opening brace
2205                         $rule = null;
2206                         $nextPos = -1;
2207                         foreach ($callbacks as $key => $value) {
2208                                 $pos = strpos ($text, $key, $i);
2209                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)) {
2210                                         $rule = $value;
2211                                         $nextPos = $pos;
2212                                 }
2213                         }
2214
2215                         if ($lastOpeningBrace >= 0) {
2216                                 $pos = strpos ($text, $openingBraceStack[$lastOpeningBrace]['braceEnd'], $i);
2217
2218                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)){
2219                                         $rule = null;
2220                                         $nextPos = $pos;
2221                                 }
2222
2223                                 $pos = strpos ($text, '|', $i);
2224
2225                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)){
2226                                         $rule = null;
2227                                         $nextPos = $pos;
2228                                 }
2229                         }
2230
2231                         if ($nextPos == -1)
2232                                 break;
2233
2234                         $i = $nextPos;
2235
2236                         # found openning brace, lets add it to parentheses stack
2237                         if (null != $rule) {
2238                                 $piece = array('brace' => $text[$i],
2239                                                            'braceEnd' => $rule['end'],
2240                                                            'count' => 1,
2241                                                            'title' => '',
2242                                                            'parts' => null);
2243
2244                                 # count openning brace characters
2245                                 while ($i+1 < strlen($text) && $text[$i+1] == $piece['brace']) {
2246                                         $piece['count']++;
2247                                         $i++;
2248                                 }
2249
2250                                 $piece['startAt'] = $i+1;
2251                                 $piece['partStart'] = $i+1;
2252
2253                                 # we need to add to stack only if openning brace count is enough for any given rule
2254                                 foreach ($rule['cb'] as $cnt => $fn) {
2255                                         if ($piece['count'] >= $cnt) {
2256                                                 $lastOpeningBrace ++;
2257                                                 $openingBraceStack[$lastOpeningBrace] = $piece;
2258                                                 break;
2259                                         }
2260                                 }
2261
2262                                 continue;
2263                         }
2264                         else if ($lastOpeningBrace >= 0) {
2265                                 # first check if it is a closing brace
2266                                 if ($openingBraceStack[$lastOpeningBrace]['braceEnd'] == $text[$i]) {
2267                                         # lets check if it is enough characters for closing brace
2268                                         $count = 1;
2269                                         while ($i+$count < strlen($text) && $text[$i+$count] == $text[$i])
2270                                                 $count++;
2271
2272                                         # if there are more closing parentheses than opening ones, we parse less
2273                                         if ($openingBraceStack[$lastOpeningBrace]['count'] < $count)
2274                                                 $count = $openingBraceStack[$lastOpeningBrace]['count'];
2275
2276                                         # check for maximum matching characters (if there are 5 closing characters, we will probably need only 3 - depending on the rules)
2277                                         $matchingCount = 0;
2278                                         $matchingCallback = null;
2279                                         foreach ($callbacks[$openingBraceStack[$lastOpeningBrace]['brace']]['cb'] as $cnt => $fn) {
2280                                                 if ($count >= $cnt && $matchingCount < $cnt) {
2281                                                         $matchingCount = $cnt;
2282                                                         $matchingCallback = $fn;
2283                                                 }
2284                                         }
2285
2286                                         if ($matchingCount == 0) {
2287                                                 $i += $count - 1;
2288                                                 continue;
2289                                         }
2290
2291                                         # lets set a title or last part (if '|' was found)
2292                                         if (null === $openingBraceStack[$lastOpeningBrace]['parts'])
2293                                                 $openingBraceStack[$lastOpeningBrace]['title'] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2294                                         else
2295                                                 $openingBraceStack[$lastOpeningBrace]['parts'][] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2296
2297                                         $pieceStart = $openingBraceStack[$lastOpeningBrace]['startAt'] - $matchingCount;
2298                                         $pieceEnd = $i + $matchingCount;
2299
2300                                         if( is_callable( $matchingCallback ) ) {
2301                                                 $cbArgs = array (
2302                                                                                  'text' => substr($text, $pieceStart, $pieceEnd - $pieceStart),
2303                                                                                  'title' => trim($openingBraceStack[$lastOpeningBrace]['title']),
2304                                                                                  'parts' => $openingBraceStack[$lastOpeningBrace]['parts'],
2305                                                                                  'lineStart' => (($pieceStart > 0) && ($text[$pieceStart-1] == '\n')),
2306                                                                                  );
2307                                                 # finally we can call a user callback and replace piece of text
2308                                                 $replaceWith = call_user_func( $matchingCallback, $cbArgs );
2309                                                 $text = substr($text, 0, $pieceStart) . $replaceWith . substr($text, $pieceEnd);
2310                                                 $i = $pieceStart + strlen($replaceWith) - 1;
2311                                         }
2312                                         else {
2313                                                 # null value for callback means that parentheses should be parsed, but not replaced
2314                                                 $i += $matchingCount - 1;
2315                                         }
2316
2317                                         # reset last openning parentheses, but keep it in case there are unused characters
2318                                         $piece = array('brace' => $openingBraceStack[$lastOpeningBrace]['brace'],
2319                                                                    'braceEnd' => $openingBraceStack[$lastOpeningBrace]['braceEnd'],
2320                                                                    'count' => $openingBraceStack[$lastOpeningBrace]['count'],
2321                                                                    'title' => '',
2322                                                                    'parts' => null,
2323                                                                    'startAt' => $openingBraceStack[$lastOpeningBrace]['startAt']);
2324                                         $openingBraceStack[$lastOpeningBrace--] = null;
2325
2326                                         if ($matchingCount < $piece['count']) {
2327                                                 $piece['count'] -= $matchingCount;
2328                                                 $piece['startAt'] -= $matchingCount;
2329                                                 $piece['partStart'] = $piece['startAt'];
2330                                                 # do we still qualify for any callback with remaining count?
2331                                                 foreach ($callbacks[$piece['brace']]['cb'] as $cnt => $fn) {
2332                                                         if ($piece['count'] >= $cnt) {
2333                                                                 $lastOpeningBrace ++;
2334                                                                 $openingBraceStack[$lastOpeningBrace] = $piece;
2335                                                                 break;
2336                                                         }
2337                                                 }
2338                                         }
2339                                         continue;
2340                                 }
2341
2342                                 # lets set a title if it is a first separator, or next part otherwise
2343                                 if ($text[$i] == '|') {
2344                                         if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2345                                                 $openingBraceStack[$lastOpeningBrace]['title'] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2346                                                 $openingBraceStack[$lastOpeningBrace]['parts'] = array();
2347                                         }
2348                                         else
2349                                                 $openingBraceStack[$lastOpeningBrace]['parts'][] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2350
2351                                         $openingBraceStack[$lastOpeningBrace]['partStart'] = $i + 1;
2352                                 }
2353                         }
2354                 }
2355
2356                 return $text;
2357         }
2358
2359         /**
2360          * Replace magic variables, templates, and template arguments
2361          * with the appropriate text. Templates are substituted recursively,
2362          * taking care to avoid infinite loops.
2363          *
2364          * Note that the substitution depends on value of $mOutputType:
2365          *  OT_WIKI: only {{subst:}} templates
2366          *  OT_MSG: only magic variables
2367          *  OT_HTML: all templates and magic variables
2368          *
2369          * @param string $tex The text to transform
2370          * @param array $args Key-value pairs representing template parameters to substitute
2371          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2372          * @private
2373          */
2374         function replaceVariables( $text, $args = array(), $argsOnly = false ) {
2375                 # Prevent too big inclusions
2376                 if( strlen( $text ) > MAX_INCLUDE_SIZE ) {
2377                         return $text;
2378                 }
2379
2380                 $fname = 'Parser::replaceVariables';
2381                 wfProfileIn( $fname );
2382
2383                 # This function is called recursively. To keep track of arguments we need a stack:
2384                 array_push( $this->mArgStack, $args );
2385
2386                 $braceCallbacks = array();
2387                 if ( !$argsOnly ) {
2388                         $braceCallbacks[2] = array( &$this, 'braceSubstitution' );
2389                 }
2390                 if ( $this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI ) {
2391                         $braceCallbacks[3] = array( &$this, 'argSubstitution' );
2392                 }
2393                 $callbacks = array();
2394                 $callbacks['{'] = array('end' => '}', 'cb' => $braceCallbacks);
2395                 $callbacks['['] = array('end' => ']', 'cb' => array(2=>null));
2396                 $text = $this->replace_callback ($text, $callbacks);
2397
2398                 array_pop( $this->mArgStack );
2399
2400                 wfProfileOut( $fname );
2401                 return $text;
2402         }
2403
2404         /**
2405          * Replace magic variables
2406          * @private
2407          */
2408         function variableSubstitution( $matches ) {
2409                 $fname = 'Parser::variableSubstitution';
2410                 $varname = $matches[1];
2411                 wfProfileIn( $fname );
2412                 if ( !$this->mVariables ) {
2413                         $this->initialiseVariables();
2414                 }
2415                 $skip = false;
2416                 if ( $this->mOutputType == OT_WIKI ) {
2417                         # Do only magic variables prefixed by SUBST
2418                         $mwSubst =& MagicWord::get( MAG_SUBST );
2419                         if (!$mwSubst->matchStartAndRemove( $varname ))
2420                                 $skip = true;
2421                         # Note that if we don't substitute the variable below,
2422                         # we don't remove the {{subst:}} magic word, in case
2423                         # it is a template rather than a magic variable.
2424                 }
2425                 if ( !$skip && array_key_exists( $varname, $this->mVariables ) ) {
2426                         $id = $this->mVariables[$varname];
2427                         $text = $this->getVariableValue( $id );
2428                         $this->mOutput->mContainsOldMagic = true;
2429                 } else {
2430                         $text = $matches[0];
2431                 }
2432                 wfProfileOut( $fname );
2433                 return $text;
2434         }
2435
2436         # Split template arguments
2437         function getTemplateArgs( $argsString ) {
2438                 if ( $argsString === '' ) {
2439                         return array();
2440                 }
2441
2442                 $args = explode( '|', substr( $argsString, 1 ) );
2443
2444                 # If any of the arguments contains a '[[' but no ']]', it needs to be
2445                 # merged with the next arg because the '|' character between belongs
2446                 # to the link syntax and not the template parameter syntax.
2447                 $argc = count($args);
2448
2449                 for ( $i = 0; $i < $argc-1; $i++ ) {
2450                         if ( substr_count ( $args[$i], '[[' ) != substr_count ( $args[$i], ']]' ) ) {
2451                                 $args[$i] .= '|'.$args[$i+1];
2452                                 array_splice($args, $i+1, 1);
2453                                 $i--;
2454                                 $argc--;
2455                         }
2456                 }
2457
2458                 return $args;
2459         }
2460
2461         /**
2462          * Return the text of a template, after recursively
2463          * replacing any variables or templates within the template.
2464          *
2465          * @param array $piece The parts of the template
2466          *  $piece['text']: matched text
2467          *  $piece['title']: the title, i.e. the part before the |
2468          *  $piece['parts']: the parameter array
2469          * @return string the text of the template
2470          * @private
2471          */
2472         function braceSubstitution( $piece ) {
2473                 global $wgContLang, $wgAllowDisplayTitle, $action;
2474                 $fname = 'Parser::braceSubstitution';
2475                 wfProfileIn( $fname );
2476
2477                 # Flags
2478                 $found = false;             # $text has been filled
2479                 $nowiki = false;            # wiki markup in $text should be escaped
2480                 $noparse = false;           # Unsafe HTML tags should not be stripped, etc.
2481                 $noargs = false;            # Don't replace triple-brace arguments in $text
2482                 $replaceHeadings = false;   # Make the edit section links go to the template not the article
2483                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2484                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2485
2486                 # Title object, where $text came from
2487                 $title = NULL;
2488
2489                 $linestart = '';
2490
2491                 # $part1 is the bit before the first |, and must contain only title characters
2492                 # $args is a list of arguments, starting from index 0, not including $part1
2493
2494                 $part1 = $piece['title'];
2495                 # If the third subpattern matched anything, it will start with |
2496
2497                 if (null == $piece['parts']) {
2498                         $replaceWith = $this->variableSubstitution (array ($piece['text'], $piece['title']));
2499                         if ($replaceWith != $piece['text']) {
2500                                 $text = $replaceWith;
2501                                 $found = true;
2502                                 $noparse = true;
2503                                 $noargs = true;
2504                         }
2505                 }
2506
2507                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2508                 $argc = count( $args );
2509
2510                 # SUBST
2511                 if ( !$found ) {
2512                         $mwSubst =& MagicWord::get( MAG_SUBST );
2513                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor ($this->mOutputType == OT_WIKI) ) {
2514                                 # One of two possibilities is true:
2515                                 # 1) Found SUBST but not in the PST phase
2516                                 # 2) Didn't find SUBST and in the PST phase
2517                                 # In either case, return without further processing
2518                                 $text = $piece['text'];
2519                                 $found = true;
2520                                 $noparse = true;
2521                                 $noargs = true;
2522                         }
2523                 }
2524
2525                 # MSG, MSGNW, INT and RAW
2526                 if ( !$found ) {
2527                         # Check for MSGNW:
2528                         $mwMsgnw =& MagicWord::get( MAG_MSGNW );
2529                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2530                                 $nowiki = true;
2531                         } else {
2532                                 # Remove obsolete MSG:
2533                                 $mwMsg =& MagicWord::get( MAG_MSG );
2534                                 $mwMsg->matchStartAndRemove( $part1 );
2535                         }
2536
2537                         # Check for RAW:
2538                         $mwRaw =& MagicWord::get( MAG_RAW );
2539                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2540                                 $forceRawInterwiki = true;
2541                         }
2542
2543                         # Check if it is an internal message
2544                         $mwInt =& MagicWord::get( MAG_INT );
2545                         if ( $mwInt->matchStartAndRemove( $part1 ) ) {
2546                                 if ( $this->incrementIncludeCount( 'int:'.$part1 ) ) {
2547                                         $text = $linestart . wfMsgReal( $part1, $args, true );
2548                                         $found = true;
2549                                 }
2550                         }
2551                 }
2552
2553                 # NS
2554                 if ( !$found ) {
2555                         # Check for NS: (namespace expansion)
2556                         $mwNs = MagicWord::get( MAG_NS );
2557                         if ( $mwNs->matchStartAndRemove( $part1 ) ) {
2558                                 if ( intval( $part1 ) || $part1 == "0" ) {
2559                                         $text = $linestart . $wgContLang->getNsText( intval( $part1 ) );
2560                                         $found = true;
2561                                 } else {
2562                                         $index = Namespace::getCanonicalIndex( strtolower( $part1 ) );
2563                                         if ( !is_null( $index ) ) {
2564                                                 $text = $linestart . $wgContLang->getNsText( $index );
2565                                                 $found = true;
2566                                         }
2567                                 }
2568                         }
2569                 }
2570
2571                 # LCFIRST, UCFIRST, LC and UC
2572                 if ( !$found ) {
2573                         $lcfirst =& MagicWord::get( MAG_LCFIRST );
2574                         $ucfirst =& MagicWord::get( MAG_UCFIRST );
2575                         $lc =& MagicWord::get( MAG_LC );
2576                         $uc =& MagicWord::get( MAG_UC );
2577                         if ( $lcfirst->matchStartAndRemove( $part1 ) ) {
2578                                 $text = $linestart . $wgContLang->lcfirst( $part1 );
2579                                 $found = true;
2580                         } else if ( $ucfirst->matchStartAndRemove( $part1 ) ) {
2581                                 $text = $linestart . $wgContLang->ucfirst( $part1 );
2582                                 $found = true;
2583                         } else if ( $lc->matchStartAndRemove( $part1 ) ) {
2584                                 $text = $linestart . $wgContLang->lc( $part1 );
2585                                 $found = true;
2586                         } else if ( $uc->matchStartAndRemove( $part1 ) ) {
2587                                  $text = $linestart . $wgContLang->uc( $part1 );
2588                                  $found = true;
2589                         }
2590                 }
2591
2592                 # LOCALURL and FULLURL
2593                 if ( !$found ) {
2594                         $mwLocal =& MagicWord::get( MAG_LOCALURL );
2595                         $mwLocalE =& MagicWord::get( MAG_LOCALURLE );
2596                         $mwFull =& MagicWord::get( MAG_FULLURL );
2597                         $mwFullE =& MagicWord::get( MAG_FULLURLE );
2598
2599
2600                         if ( $mwLocal->matchStartAndRemove( $part1 ) ) {
2601                                 $func = 'getLocalURL';
2602                         } elseif ( $mwLocalE->matchStartAndRemove( $part1 ) ) {
2603                                 $func = 'escapeLocalURL';
2604                         } elseif ( $mwFull->matchStartAndRemove( $part1 ) ) {
2605                                 $func = 'getFullURL';
2606                         } elseif ( $mwFullE->matchStartAndRemove( $part1 ) ) {
2607                                 $func = 'escapeFullURL';
2608                         } else {
2609                                 $func = false;
2610                         }
2611
2612                         if ( $func !== false ) {
2613                                 $title = Title::newFromText( $part1 );
2614                                 if ( !is_null( $title ) ) {
2615                                         if ( $argc > 0 ) {
2616                                                 $text = $linestart . $title->$func( $args[0] );
2617                                         } else {
2618                                                 $text = $linestart . $title->$func();
2619                                         }
2620                                         $found = true;
2621                                 }
2622                         }
2623                 }
2624
2625                 # GRAMMAR
2626                 if ( !$found && $argc == 1 ) {
2627                         $mwGrammar =& MagicWord::get( MAG_GRAMMAR );
2628                         if ( $mwGrammar->matchStartAndRemove( $part1 ) ) {
2629                                 $text = $linestart . $wgContLang->convertGrammar( $args[0], $part1 );
2630                                 $found = true;
2631                         }
2632                 }
2633
2634                 # PLURAL
2635                 if ( !$found && $argc >= 2 ) {
2636                         $mwPluralForm =& MagicWord::get( MAG_PLURAL );
2637                         if ( $mwPluralForm->matchStartAndRemove( $part1 ) ) {
2638                                 if ($argc==2) {$args[2]=$args[1];}
2639                                 $text = $linestart . $wgContLang->convertPlural( $part1, $args[0], $args[1], $args[2]);
2640                                 $found = true;
2641                         }
2642                 }
2643
2644                 # DISPLAYTITLE
2645                 if ( !$found && $argc == 1 && $wgAllowDisplayTitle ) {
2646                         $mwDT =& MagicWord::get( MAG_DISPLAYTITLE );
2647                         if ( $mwDT->matchStartAndRemove( $part1 ) ) {
2648
2649                                 # Set title in parser output object
2650                                 $param = $args[0];
2651                                 $parserOptions = new ParserOptions;
2652                                 $local_parser = new Parser ();
2653                                 $t2 = $local_parser->parse ( $param, $this->mTitle, $parserOptions, false );
2654                                 $this->mOutput->mHTMLtitle = $t2->GetText();
2655
2656                                 # Add subtitle
2657                                 $t = $this->mTitle->getPrefixedText();
2658                                 $this->mOutput->mSubtitle .= wfMsg('displaytitle', $t);
2659                                 $text = "" ;
2660                                 $found = true ;
2661                         }
2662                 }
2663
2664                 # Extensions
2665                 if ( !$found && substr( $part1, 0, 1 ) == '#' ) {
2666                         $colonPos = strpos( $part1, ':' );
2667                         if ( $colonPos !== false ) {
2668                                 $function = strtolower( substr( $part1, 1, $colonPos - 1 ) );
2669                                 if ( isset( $this->mFunctionHooks[$function] ) ) {
2670                                         $funcArgs = array_map( 'trim', $args );
2671                                         $funcArgs = array_merge( array( &$this, trim( substr( $part1, $colonPos + 1 ) ) ), $funcArgs );
2672                                         $result = call_user_func_array( $this->mFunctionHooks[$function], $funcArgs );
2673                                         $found = true;
2674
2675                                         // The text is usually already parsed, doesn't need triple-brace tags expanded, etc.
2676                                         //$noargs = true;
2677                                         //$noparse = true;
2678
2679                                         if ( is_array( $result ) ) {
2680                                                 $text = $linestart . $result[0];
2681                                                 unset( $result[0] );
2682
2683                                                 // Extract flags into the local scope
2684                                                 // This allows callers to set flags such as nowiki, noparse, found, etc.
2685                                                 extract( $result );
2686                                         } else {
2687                                                 $text = $linestart . $result;
2688                                         }
2689                                 }
2690                         }
2691                 }
2692
2693                 # Template table test
2694
2695                 # Did we encounter this template already? If yes, it is in the cache
2696                 # and we need to check for loops.
2697                 if ( !$found && isset( $this->mTemplates[$piece['title']] ) ) {
2698                         $found = true;
2699
2700                         # Infinite loop test
2701                         if ( isset( $this->mTemplatePath[$part1] ) ) {
2702                                 $noparse = true;
2703                                 $noargs = true;
2704                                 $found = true;
2705                                 $text = $linestart .
2706                                         '{{' . $part1 . '}}' .
2707                                         '<!-- WARNING: template loop detected -->';
2708                                 wfDebug( "$fname: template loop broken at '$part1'\n" );
2709                         } else {
2710                                 # set $text to cached message.
2711                                 $text = $linestart . $this->mTemplates[$piece['title']];
2712                         }
2713                 }
2714
2715                 # Load from database
2716                 $lastPathLevel = $this->mTemplatePath;
2717                 if ( !$found ) {
2718                         $ns = NS_TEMPLATE;
2719                         # declaring $subpage directly in the function call
2720                         # does not work correctly with references and breaks
2721                         # {{/subpage}}-style inclusions
2722                         $subpage = '';
2723                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
2724                         if ($subpage !== '') {
2725                                 $ns = $this->mTitle->getNamespace();
2726                         }
2727                         $title = Title::newFromText( $part1, $ns );
2728
2729                         if ( !is_null( $title ) ) {
2730                                 if ( !$title->isExternal() ) {
2731                                         # Check for excessive inclusion
2732                                         $dbk = $title->getPrefixedDBkey();
2733                                         if ( $this->incrementIncludeCount( $dbk ) ) {
2734                                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() ) {
2735                                                         # Capture special page output
2736                                                         $text = SpecialPage::capturePath( $title );
2737                                                         if ( is_string( $text ) ) {
2738                                                                 $found = true;
2739                                                                 $noparse = true;
2740                                                                 $noargs = true;
2741                                                                 $isHTML = true;
2742                                                                 $this->disableCache();
2743                                                         }
2744                                                 } else {
2745                                                         $articleContent = $this->fetchTemplate( $title );
2746                                                         if ( $articleContent !== false ) {
2747                                                                 $found = true;
2748                                                                 $text = $articleContent;
2749                                                                 $replaceHeadings = true;
2750                                                         }
2751                                                 }
2752                                         }
2753
2754                                         # If the title is valid but undisplayable, make a link to it
2755                                         if ( $this->mOutputType == OT_HTML && !$found ) {
2756                                                 $text = '[['.$title->getPrefixedText().']]';
2757                                                 $found = true;
2758                                         }
2759                                 } elseif ( $title->isTrans() ) {
2760                                         // Interwiki transclusion
2761                                         if ( $this->mOutputType == OT_HTML && !$forceRawInterwiki ) {
2762                                                 $text = $this->interwikiTransclude( $title, 'render' );
2763                                                 $isHTML = true;
2764                                                 $noparse = true;
2765                                         } else {
2766                                                 $text = $this->interwikiTransclude( $title, 'raw' );
2767                                                 $replaceHeadings = true;
2768                                         }
2769                                         $found = true;
2770                                 }
2771
2772                                 # Template cache array insertion
2773                                 # Use the original $piece['title'] not the mangled $part1, so that
2774                                 # modifiers such as RAW: produce separate cache entries
2775                                 if( $found ) {
2776                                         $this->mTemplates[$piece['title']] = $text;
2777                                         $text = $linestart . $text;
2778                                 }
2779                         }
2780                 }
2781
2782                 # Recursive parsing, escaping and link table handling
2783                 # Only for HTML output
2784                 if ( $nowiki && $found && $this->mOutputType == OT_HTML ) {
2785                         $text = wfEscapeWikiText( $text );
2786                 } elseif ( ($this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI) && $found ) {
2787                         if ( $noargs ) {
2788                                 $assocArgs = array();
2789                         } else {
2790                                 # Clean up argument array
2791                                 $assocArgs = array();
2792                                 $index = 1;
2793                                 foreach( $args as $arg ) {
2794                                         $eqpos = strpos( $arg, '=' );
2795                                         if ( $eqpos === false ) {
2796                                                 $assocArgs[$index++] = $arg;
2797                                         } else {
2798                                                 $name = trim( substr( $arg, 0, $eqpos ) );
2799                                                 $value = trim( substr( $arg, $eqpos+1 ) );
2800                                                 if ( $value === false ) {
2801                                                         $value = '';
2802                                                 }
2803                                                 if ( $name !== false ) {
2804                                                         $assocArgs[$name] = $value;
2805                                                 }
2806                                         }
2807                                 }
2808
2809                                 # Add a new element to the templace recursion path
2810                                 $this->mTemplatePath[$part1] = 1;
2811                         }
2812
2813                         if ( !$noparse ) {
2814                                 # If there are any <onlyinclude> tags, only include them
2815                                 if ( in_string( '<onlyinclude>', $text ) && in_string( '</onlyinclude>', $text ) ) {
2816                                         preg_match_all( '/<onlyinclude>(.*?)\n?<\/onlyinclude>/s', $text, $m );
2817                                         $text = '';
2818                                         foreach ($m[1] as $piece)
2819                                                 $text .= $piece;
2820                                 }
2821                                 # Remove <noinclude> sections and <includeonly> tags
2822                                 $text = preg_replace( '/<noinclude>.*?<\/noinclude>/s', '', $text );
2823                                 $text = strtr( $text, array( '<includeonly>' => '' , '</includeonly>' => '' ) );
2824
2825                                 if( $this->mOutputType == OT_HTML ) {
2826                                         # Strip <nowiki>, <pre>, etc.
2827                                         $text = $this->strip( $text, $this->mStripState );
2828                                         $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'replaceVariables' ), $assocArgs );
2829                                 }
2830                                 $text = $this->replaceVariables( $text, $assocArgs );
2831
2832                                 # If the template begins with a table or block-level
2833                                 # element, it should be treated as beginning a new line.
2834                                 if (!$piece['lineStart'] && preg_match('/^({\\||:|;|#|\*)/', $text)) {
2835                                         $text = "\n" . $text;
2836                                 }
2837                         } elseif ( !$noargs ) {
2838                                 # $noparse and !$noargs
2839                                 # Just replace the arguments, not any double-brace items
2840                                 # This is used for rendered interwiki transclusion
2841                                 $text = $this->replaceVariables( $text, $assocArgs, true );
2842                         }
2843                 }
2844                 # Prune lower levels off the recursion check path
2845                 $this->mTemplatePath = $lastPathLevel;
2846
2847                 if ( !$found ) {
2848                         wfProfileOut( $fname );
2849                         return $piece['text'];
2850                 } else {
2851                         if ( $isHTML ) {
2852                                 # Replace raw HTML by a placeholder
2853                                 # Add a blank line preceding, to prevent it from mucking up
2854                                 # immediately preceding headings
2855                                 $text = "\n\n" . $this->insertStripItem( $text, $this->mStripState );
2856                         } else {
2857                                 # replace ==section headers==
2858                                 # XXX this needs to go away once we have a better parser.
2859                                 if ( $this->mOutputType != OT_WIKI && $replaceHeadings ) {
2860                                         if( !is_null( $title ) )
2861                                                 $encodedname = base64_encode($title->getPrefixedDBkey());
2862                                         else
2863                                                 $encodedname = base64_encode("");
2864                                         $m = preg_split('/(^={1,6}.*?={1,6}\s*?$)/m', $text, -1,
2865                                                 PREG_SPLIT_DELIM_CAPTURE);
2866                                         $text = '';
2867                                         $nsec = 0;
2868                                         for( $i = 0; $i < count($m); $i += 2 ) {
2869                                                 $text .= $m[$i];
2870                                                 if (!isset($m[$i + 1]) || $m[$i + 1] == "") continue;
2871                                                 $hl = $m[$i + 1];
2872                                                 if( strstr($hl, "<!--MWTEMPLATESECTION") ) {
2873                                                         $text .= $hl;
2874                                                         continue;
2875                                                 }
2876                                                 preg_match('/^(={1,6})(.*?)(={1,6})\s*?$/m', $hl, $m2);
2877                                                 $text .= $m2[1] . $m2[2] . "<!--MWTEMPLATESECTION="
2878                                                         . $encodedname . "&" . base64_encode("$nsec") . "-->" . $m2[3];
2879
2880                                                 $nsec++;
2881                                         }
2882                                 }
2883                         }
2884                 }
2885
2886                 # Prune lower levels off the recursion check path
2887                 $this->mTemplatePath = $lastPathLevel;
2888
2889                 if ( !$found ) {
2890                         wfProfileOut( $fname );
2891                         return $piece['text'];
2892                 } else {
2893                         wfProfileOut( $fname );
2894                         return $text;
2895                 }
2896         }
2897
2898         /**
2899          * Fetch the unparsed text of a template and register a reference to it.
2900          */
2901         function fetchTemplate( $title ) {
2902                 $text = false;
2903                 // Loop to fetch the article, with up to 1 redirect
2904                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
2905                         $rev = Revision::newFromTitle( $title );
2906                         $this->mOutput->addTemplate( $title, $title->getArticleID() );
2907                         if ( !$rev ) {
2908                                 break;
2909                         }
2910                         $text = $rev->getText();
2911                         if ( $text === false ) {
2912                                 break;
2913                         }
2914                         // Redirect?
2915                         $title = Title::newFromRedirect( $text );
2916                 }
2917                 return $text;
2918         }
2919
2920         /**
2921          * Transclude an interwiki link.
2922          */
2923         function interwikiTransclude( $title, $action ) {
2924                 global $wgEnableScaryTranscluding, $wgCanonicalNamespaceNames;
2925
2926                 if (!$wgEnableScaryTranscluding)
2927                         return wfMsg('scarytranscludedisabled');
2928
2929                 // The namespace will actually only be 0 or 10, depending on whether there was a leading :
2930                 // But we'll handle it generally anyway
2931                 if ( $title->getNamespace() ) {
2932                         // Use the canonical namespace, which should work anywhere
2933                         $articleName = $wgCanonicalNamespaceNames[$title->getNamespace()] . ':' . $title->getDBkey();
2934                 } else {
2935                         $articleName = $title->getDBkey();
2936                 }
2937
2938                 $url = str_replace('$1', urlencode($articleName), Title::getInterwikiLink($title->getInterwiki()));
2939                 $url .= "?action=$action";
2940                 if (strlen($url) > 255)
2941                         return wfMsg('scarytranscludetoolong');
2942                 return $this->fetchScaryTemplateMaybeFromCache($url);
2943         }
2944
2945         function fetchScaryTemplateMaybeFromCache($url) {
2946                 global $wgTranscludeCacheExpiry;
2947                 $dbr =& wfGetDB(DB_SLAVE);
2948                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
2949                                 array('tc_url' => $url));
2950                 if ($obj) {
2951                         $time = $obj->tc_time;
2952                         $text = $obj->tc_contents;
2953                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
2954                                 return $text;
2955                         }
2956                 }
2957
2958                 $text = wfGetHTTP($url);
2959                 if (!$text)
2960                         return wfMsg('scarytranscludefailed', $url);
2961
2962                 $dbw =& wfGetDB(DB_MASTER);
2963                 $dbw->replace('transcache', array('tc_url'), array(
2964                         'tc_url' => $url,
2965                         'tc_time' => time(),
2966                         'tc_contents' => $text));
2967                 return $text;
2968         }
2969
2970
2971         /**
2972          * Triple brace replacement -- used for template arguments
2973          * @private
2974          */
2975         function argSubstitution( $matches ) {
2976                 $arg = trim( $matches['title'] );
2977                 $text = $matches['text'];
2978                 $inputArgs = end( $this->mArgStack );
2979
2980                 if ( array_key_exists( $arg, $inputArgs ) ) {
2981                         $text = $inputArgs[$arg];
2982                 } else if ($this->mOutputType == OT_HTML && null != $matches['parts'] && count($matches['parts']) > 0) {
2983                         $text = $matches['parts'][0];
2984                 }
2985
2986                 return $text;
2987         }
2988
2989         /**
2990          * Returns true if the function is allowed to include this entity
2991          * @private
2992          */
2993         function incrementIncludeCount( $dbk ) {
2994                 if ( !array_key_exists( $dbk, $this->mIncludeCount ) ) {
2995                         $this->mIncludeCount[$dbk] = 0;
2996                 }
2997                 if ( ++$this->mIncludeCount[$dbk] <= MAX_INCLUDE_REPEAT ) {
2998                         return true;
2999                 } else {
3000                         return false;
3001                 }
3002         }
3003
3004         /**
3005          * This function accomplishes several tasks:
3006          * 1) Auto-number headings if that option is enabled
3007          * 2) Add an [edit] link to sections for logged in users who have enabled the option
3008          * 3) Add a Table of contents on the top for users who have enabled the option
3009          * 4) Auto-anchor headings
3010          *
3011          * It loops through all headlines, collects the necessary data, then splits up the
3012          * string and re-inserts the newly formatted headlines.
3013          *
3014          * @param string $text
3015          * @param boolean $isMain
3016          * @private
3017          */
3018         function formatHeadings( $text, $isMain=true ) {
3019                 global $wgMaxTocLevel, $wgContLang;
3020
3021                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3022                 $doShowToc = true;
3023                 $forceTocHere = false;
3024                 if( !$this->mTitle->userCanEdit() ) {
3025                         $showEditLink = 0;
3026                 } else {
3027                         $showEditLink = $this->mOptions->getEditSection();
3028                 }
3029
3030                 # Inhibit editsection links if requested in the page
3031                 $esw =& MagicWord::get( MAG_NOEDITSECTION );
3032                 if( $esw->matchAndRemove( $text ) ) {
3033                         $showEditLink = 0;
3034                 }
3035                 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
3036                 # do not add TOC
3037                 $mw =& MagicWord::get( MAG_NOTOC );
3038                 if( $mw->matchAndRemove( $text ) ) {
3039                         $doShowToc = false;
3040                 }
3041
3042                 # Get all headlines for numbering them and adding funky stuff like [edit]
3043                 # links - this is for later, but we need the number of headlines right now
3044                 $numMatches = preg_match_all( '/<H([1-6])(.*?'.'>)(.*?)<\/H[1-6] *>/i', $text, $matches );
3045
3046                 # if there are fewer than 4 headlines in the article, do not show TOC
3047                 if( $numMatches < 4 ) {
3048                         $doShowToc = false;
3049                 }
3050
3051                 # if the string __TOC__ (not case-sensitive) occurs in the HTML,
3052                 # override above conditions and always show TOC at that place
3053
3054                 $mw =& MagicWord::get( MAG_TOC );
3055                 if($mw->match( $text ) ) {
3056                         $doShowToc = true;
3057                         $forceTocHere = true;
3058                 } else {
3059                         # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3060                         # override above conditions and always show TOC above first header
3061                         $mw =& MagicWord::get( MAG_FORCETOC );
3062                         if ($mw->matchAndRemove( $text ) ) {
3063                                 $doShowToc = true;
3064                         }
3065                 }
3066
3067                 # Never ever show TOC if no headers
3068                 if( $numMatches < 1 ) {
3069                         $doShowToc = false;
3070                 }
3071
3072                 # We need this to perform operations on the HTML
3073                 $sk =& $this->mOptions->getSkin();
3074
3075                 # headline counter
3076                 $headlineCount = 0;
3077                 $sectionCount = 0; # headlineCount excluding template sections
3078
3079                 # Ugh .. the TOC should have neat indentation levels which can be
3080                 # passed to the skin functions. These are determined here
3081                 $toc = '';
3082                 $full = '';
3083                 $head = array();
3084                 $sublevelCount = array();
3085                 $levelCount = array();
3086                 $toclevel = 0;
3087                 $level = 0;
3088                 $prevlevel = 0;
3089                 $toclevel = 0;
3090                 $prevtoclevel = 0;
3091
3092                 foreach( $matches[3] as $headline ) {
3093                         $istemplate = 0;
3094                         $templatetitle = '';
3095                         $templatesection = 0;
3096                         $numbering = '';
3097
3098                         if (preg_match("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", $headline, $mat)) {
3099                                 $istemplate = 1;
3100                                 $templatetitle = base64_decode($mat[1]);
3101                                 $templatesection = 1 + (int)base64_decode($mat[2]);
3102                                 $headline = preg_replace("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", "", $headline);
3103                         }
3104
3105                         if( $toclevel ) {
3106                                 $prevlevel = $level;
3107                                 $prevtoclevel = $toclevel;
3108                         }
3109                         $level = $matches[1][$headlineCount];
3110
3111                         if( $doNumberHeadings || $doShowToc ) {
3112
3113                                 if ( $level > $prevlevel ) {
3114                                         # Increase TOC level
3115                                         $toclevel++;
3116                                         $sublevelCount[$toclevel] = 0;
3117                                         if( $toclevel<$wgMaxTocLevel ) {
3118                                                 $toc .= $sk->tocIndent();
3119                                         }
3120                                 }
3121                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3122                                         # Decrease TOC level, find level to jump to
3123
3124                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
3125                                                 # Can only go down to level 1
3126                                                 $toclevel = 1;
3127                                         } else {
3128                                                 for ($i = $toclevel; $i > 0; $i--) {
3129                                                         if ( $levelCount[$i] == $level ) {
3130                                                                 # Found last matching level
3131                                                                 $toclevel = $i;
3132                                                                 break;
3133                                                         }
3134                                                         elseif ( $levelCount[$i] < $level ) {
3135                                                                 # Found first matching level below current level
3136                                                                 $toclevel = $i + 1;
3137                                                                 break;
3138                                                         }
3139                                                 }
3140                                         }
3141                                         if( $toclevel<$wgMaxTocLevel ) {
3142                                                 $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3143                                         }
3144                                 }
3145                                 else {
3146                                         # No change in level, end TOC line
3147                                         if( $toclevel<$wgMaxTocLevel ) {
3148                                                 $toc .= $sk->tocLineEnd();
3149                                         }
3150                                 }
3151
3152                                 $levelCount[$toclevel] = $level;
3153
3154                                 # count number of headlines for each level
3155                                 @$sublevelCount[$toclevel]++;
3156                                 $dot = 0;
3157                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3158                                         if( !empty( $sublevelCount[$i] ) ) {
3159                                                 if( $dot ) {
3160                                                         $numbering .= '.';
3161                                                 }
3162                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3163                                                 $dot = 1;
3164                                         }
3165                                 }
3166                         }
3167
3168                         # The canonized header is a version of the header text safe to use for links
3169                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3170                         $canonized_headline = $this->unstrip( $headline, $this->mStripState );
3171                         $canonized_headline = $this->unstripNoWiki( $canonized_headline, $this->mStripState );
3172
3173                         # Remove link placeholders by the link text.
3174                         #     <!--LINK number-->
3175                         # turns into
3176                         #     link text with suffix
3177                         $canonized_headline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
3178                                                             "\$this->mLinkHolders['texts'][\$1]",
3179                                                             $canonized_headline );
3180                         $canonized_headline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
3181                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
3182                                                             $canonized_headline );
3183
3184                         # strip out HTML
3185                         $canonized_headline = preg_replace( '/<.*?' . '>/','',$canonized_headline );
3186                         $tocline = trim( $canonized_headline );
3187                         # Save headline for section edit hint before it's escaped
3188                         $headline_hint = trim( $canonized_headline );
3189                         $canonized_headline = Sanitizer::escapeId( $tocline );
3190                         $refers[$headlineCount] = $canonized_headline;
3191
3192                         # count how many in assoc. array so we can track dupes in anchors
3193                         @$refers[$canonized_headline]++;
3194                         $refcount[$headlineCount]=$refers[$canonized_headline];
3195
3196                         # Don't number the heading if it is the only one (looks silly)
3197                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3198                                 # the two are different if the line contains a link
3199                                 $headline=$numbering . ' ' . $headline;
3200                         }
3201
3202                         # Create the anchor for linking from the TOC to the section
3203                         $anchor = $canonized_headline;
3204                         if($refcount[$headlineCount] > 1 ) {
3205                                 $anchor .= '_' . $refcount[$headlineCount];
3206                         }
3207                         if( $doShowToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3208                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3209                         }
3210                         if( $showEditLink && ( !$istemplate || $templatetitle !== "" ) ) {
3211                                 if ( empty( $head[$headlineCount] ) ) {
3212                                         $head[$headlineCount] = '';
3213                                 }
3214                                 if( $istemplate )
3215                                         $head[$headlineCount] .= $sk->editSectionLinkForOther($templatetitle, $templatesection);
3216                                 else
3217                                         $head[$headlineCount] .= $sk->editSectionLink($this->mTitle, $sectionCount+1, $headline_hint);
3218                         }
3219
3220                         # give headline the correct <h#> tag
3221                         @$head[$headlineCount] .= "<a name=\"$anchor\"></a><h".$level.$matches[2][$headlineCount] .$headline.'</h'.$level.'>';
3222
3223                         $headlineCount++;
3224                         if( !$istemplate )
3225                                 $sectionCount++;
3226                 }
3227
3228                 if( $doShowToc ) {
3229                         if( $toclevel<$wgMaxTocLevel ) {
3230                                 $toc .= $sk->tocUnindent( $toclevel - 1 );
3231                         }
3232                         $toc = $sk->tocList( $toc );
3233                 }
3234
3235                 # split up and insert constructed headlines
3236
3237                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3238                 $i = 0;
3239
3240                 foreach( $blocks as $block ) {
3241                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
3242                                 # This is the [edit] link that appears for the top block of text when
3243                                 # section editing is enabled
3244
3245                                 # Disabled because it broke block formatting
3246                                 # For example, a bullet point in the top line
3247                                 # $full .= $sk->editSectionLink(0);
3248                         }
3249                         $full .= $block;
3250                         if( $doShowToc && !$i && $isMain && !$forceTocHere) {
3251                         # Top anchor now in skin
3252                                 $full = $full.$toc;
3253                         }
3254
3255                         if( !empty( $head[$i] ) ) {
3256                                 $full .= $head[$i];
3257                         }
3258                         $i++;
3259                 }
3260                 if($forceTocHere) {
3261                         $mw =& MagicWord::get( MAG_TOC );
3262                         return $mw->replace( $toc, $full );
3263                 } else {
3264                         return $full;
3265                 }
3266         }
3267
3268         /**
3269          * Return an HTML link for the "ISBN 123456" text
3270          * @private
3271          */
3272         function magicISBN( $text ) {
3273                 $fname = 'Parser::magicISBN';
3274                 wfProfileIn( $fname );
3275
3276                 $a = split( 'ISBN ', ' '.$text );
3277                 if ( count ( $a ) < 2 ) {
3278                         wfProfileOut( $fname );
3279                         return $text;
3280                 }
3281                 $text = substr( array_shift( $a ), 1);
3282                 $valid = '0123456789-Xx';
3283
3284                 foreach ( $a as $x ) {
3285                         # hack: don't replace inside thumbnail title/alt
3286                         # attributes
3287                         if(preg_match('/<[^>]+(alt|title)="[^">]*$/', $text)) {
3288                                 $text .= "ISBN $x";
3289                                 continue;
3290                         }
3291
3292                         $isbn = $blank = '' ;
3293                         while ( ' ' == $x{0} ) {
3294                                 $blank .= ' ';
3295                                 $x = substr( $x, 1 );
3296                         }
3297                         if ( $x == '' ) { # blank isbn
3298                                 $text .= "ISBN $blank";
3299                                 continue;
3300                         }
3301                         while ( strstr( $valid, $x{0} ) != false ) {
3302                                 $isbn .= $x{0};
3303                                 $x = substr( $x, 1 );
3304                         }
3305                         $num = str_replace( '-', '', $isbn );
3306                         $num = str_replace( ' ', '', $num );
3307                         $num = str_replace( 'x', 'X', $num );
3308
3309                         if ( '' == $num ) {
3310                                 $text .= "ISBN $blank$x";
3311                         } else {
3312                                 $titleObj = Title::makeTitle( NS_SPECIAL, 'Booksources' );
3313                                 $text .= '<a href="' .
3314                                         $titleObj->escapeLocalUrl( 'isbn='.$num ) .
3315                                         "\" class=\"internal\">ISBN $isbn</a>";
3316                                 $text .= $x;
3317                         }
3318                 }
3319                 wfProfileOut( $fname );
3320                 return $text;
3321         }
3322
3323         /**
3324          * Return an HTML link for the "RFC 1234" text
3325          *
3326          * @private
3327          * @param string $text     Text to be processed
3328          * @param string $keyword  Magic keyword to use (default RFC)
3329          * @param string $urlmsg   Interface message to use (default rfcurl)
3330          * @return string
3331          */
3332         function magicRFC( $text, $keyword='RFC ', $urlmsg='rfcurl'  ) {
3333
3334                 $valid = '0123456789';
3335                 $internal = false;
3336
3337                 $a = split( $keyword, ' '.$text );
3338                 if ( count ( $a ) < 2 ) {
3339                         return $text;
3340                 }
3341                 $text = substr( array_shift( $a ), 1);
3342
3343                 /* Check if keyword is preceed by [[.
3344                  * This test is made here cause of the array_shift above
3345                  * that prevent the test to be done in the foreach.
3346                  */
3347                 if ( substr( $text, -2 ) == '[[' ) {
3348                         $internal = true;
3349                 }
3350
3351                 foreach ( $a as $x ) {
3352                         /* token might be empty if we have RFC RFC 1234 */
3353                         if ( $x=='' ) {
3354                                 $text.=$keyword;
3355                                 continue;
3356                                 }
3357
3358                         # hack: don't replace inside thumbnail title/alt
3359                         # attributes
3360                         if(preg_match('/<[^>]+(alt|title)="[^">]*$/', $text)) {
3361                                 $text .= $keyword . $x;
3362                                 continue;
3363                         }
3364
3365                         $id = $blank = '' ;
3366
3367                         /** remove and save whitespaces in $blank */
3368                         while ( $x{0} == ' ' ) {
3369                                 $blank .= ' ';
3370                                 $x = substr( $x, 1 );
3371                         }
3372
3373                         /** remove and save the rfc number in $id */
3374                         while ( strstr( $valid, $x{0} ) != false ) {
3375                                 $id .= $x{0};
3376                                 $x = substr( $x, 1 );
3377                         }
3378
3379                         if ( $id == '' ) {
3380                                 /* call back stripped spaces*/
3381                                 $text .= $keyword.$blank.$x;
3382                         } elseif( $internal ) {
3383                                 /* normal link */
3384                                 $text .= $keyword.$id.$x;
3385                         } else {
3386                                 /* build the external link*/
3387                                 $url = wfMsg( $urlmsg, $id);
3388                                 $sk =& $this->mOptions->getSkin();
3389                                 $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
3390                                 $text .= "<a href='{$url}'{$la}>{$keyword}{$id}</a>{$x}";
3391                         }
3392
3393                         /* Check if the next RFC keyword is preceed by [[ */
3394                         $internal = ( substr($x,-2) == '[[' );
3395                 }
3396                 return $text;
3397         }
3398
3399         /**
3400          * Transform wiki markup when saving a page by doing \r\n -> \n
3401          * conversion, substitting signatures, {{subst:}} templates, etc.
3402          *
3403          * @param string $text the text to transform
3404          * @param Title &$title the Title object for the current article
3405          * @param User &$user the User object describing the current user
3406          * @param ParserOptions $options parsing options
3407          * @param bool $clearState whether to clear the parser state first
3408          * @return string the altered wiki markup
3409          * @public
3410          */
3411         function preSaveTransform( $text, &$title, &$user, $options, $clearState = true ) {
3412                 $this->mOptions = $options;
3413                 $this->mTitle =& $title;
3414                 $this->mOutputType = OT_WIKI;
3415
3416                 if ( $clearState ) {
3417                         $this->clearState();
3418                 }
3419
3420                 $stripState = false;
3421                 $pairs = array(
3422                         "\r\n" => "\n",
3423                 );
3424                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3425                 $text = $this->strip( $text, $stripState, true );
3426                 $text = $this->pstPass2( $text, $user );
3427                 $text = $this->unstrip( $text, $stripState );
3428                 $text = $this->unstripNoWiki( $text, $stripState );
3429                 return $text;
3430         }
3431
3432         /**
3433          * Pre-save transform helper function
3434          * @private
3435          */
3436         function pstPass2( $text, &$user ) {
3437                 global $wgContLang, $wgLocaltimezone;
3438
3439                 /* Note: This is the timestamp saved as hardcoded wikitext to
3440                  * the database, we use $wgContLang here in order to give
3441                  * everyone the same signiture and use the default one rather
3442                  * than the one selected in each users preferences.
3443                  */
3444                 if ( isset( $wgLocaltimezone ) ) {
3445                         $oldtz = getenv( 'TZ' );
3446                         putenv( 'TZ='.$wgLocaltimezone );
3447                 }
3448                 $d = $wgContLang->timeanddate( date( 'YmdHis' ), false, false) .
3449                   ' (' . date( 'T' ) . ')';
3450                 if ( isset( $wgLocaltimezone ) ) {
3451                         putenv( 'TZ='.$oldtz );
3452                 }
3453
3454                 # Variable replacement
3455                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3456                 $text = $this->replaceVariables( $text );
3457
3458                 # Signatures
3459                 $sigText = $this->getUserSig( $user );
3460                 $text = strtr( $text, array(
3461                         '~~~~~' => $d,
3462                         '~~~~' => "$sigText $d",
3463                         '~~~' => $sigText
3464                 ) );
3465
3466                 # Context links: [[|name]] and [[name (context)|]]
3467                 #
3468                 global $wgLegalTitleChars;
3469                 $tc = "[$wgLegalTitleChars]";
3470                 $np = str_replace( array( '(', ')' ), array( '', '' ), $tc ); # No parens
3471
3472                 $namespacechar = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
3473                 $conpat = "/^({$np}+) \\(({$tc}+)\\)$/";
3474
3475                 $p1 = "/\[\[({$np}+) \\(({$np}+)\\)\\|]]/";             # [[page (context)|]]
3476                 $p2 = "/\[\[\\|({$tc}+)]]/";                                    # [[|page]]
3477                 $p3 = "/\[\[(:*$namespacechar+):({$np}+)\\|]]/";                # [[namespace:page|]] and [[:namespace:page|]]
3478                 $p4 = "/\[\[(:*$namespacechar+):({$np}+) \\(({$np}+)\\)\\|]]/"; # [[ns:page (cont)|]] and [[:ns:page (cont)|]]
3479                 $context = '';
3480                 $t = $this->mTitle->getText();
3481                 if ( preg_match( $conpat, $t, $m ) ) {
3482                         $context = $m[2];
3483                 }
3484                 $text = preg_replace( $p4, '[[\\1:\\2 (\\3)|\\2]]', $text );
3485                 $text = preg_replace( $p1, '[[\\1 (\\2)|\\1]]', $text );
3486                 $text = preg_replace( $p3, '[[\\1:\\2|\\2]]', $text );
3487
3488                 if ( '' == $context ) {
3489                         $text = preg_replace( $p2, '[[\\1]]', $text );
3490                 } else {
3491                         $text = preg_replace( $p2, "[[\\1 ({$context})|\\1]]", $text );
3492                 }
3493
3494                 # Trim trailing whitespace
3495                 # MAG_END (__END__) tag allows for trailing
3496                 # whitespace to be deliberately included
3497                 $text = rtrim( $text );
3498                 $mw =& MagicWord::get( MAG_END );
3499                 $mw->matchAndRemove( $text );
3500
3501                 return $text;
3502         }
3503
3504         /**
3505          * Fetch the user's signature text, if any, and normalize to
3506          * validated, ready-to-insert wikitext.
3507          *
3508          * @param User $user
3509          * @return string
3510          * @private
3511          */
3512         function getUserSig( &$user ) {
3513                 $username = $user->getName();
3514                 $nickname = $user->getOption( 'nickname' );
3515                 $nickname = $nickname === '' ? $username : $nickname;
3516
3517                 if( $user->getBoolOption( 'fancysig' ) !== false ) {
3518                         # Sig. might contain markup; validate this
3519                         if( $this->validateSig( $nickname ) !== false ) {
3520                                 # Validated; clean up (if needed) and return it
3521                                 return( $this->cleanSig( $nickname ) );
3522                         } else {
3523                                 # Failed to validate; fall back to the default
3524                                 $nickname = $username;
3525                                 wfDebug( "Parser::getUserSig: $username has bad XML tags in signature.\n" );
3526                         }
3527                 }
3528
3529                 # If we're still here, make it a link to the user page
3530                 $userpage = $user->getUserPage();
3531                 return( '[[' . $userpage->getPrefixedText() . '|' . wfEscapeWikiText( $nickname ) . ']]' );
3532         }
3533
3534         /**
3535          * Check that the user's signature contains no bad XML
3536          *
3537          * @param string $text
3538          * @return mixed An expanded string, or false if invalid.
3539          */
3540         function validateSig( $text ) {
3541                 return( wfIsWellFormedXmlFragment( $text ) ? $text : false );
3542         }
3543
3544         /**
3545          * Clean up signature text
3546          *
3547          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures
3548          * 2) Substitute all transclusions
3549          *
3550          * @param string $text
3551          * @return string Signature text
3552          */
3553         function cleanSig( $text ) {
3554                 $substWord = MagicWord::get( MAG_SUBST );
3555                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3556                 $substText = '{{' . $substWord->getSynonym( 0 );
3557
3558                 $text = preg_replace( $substRegex, $substText, $text );
3559                 $text = preg_replace( '/~{3,5}/', '', $text );
3560                 $text = $this->replaceVariables( $text );
3561
3562                 return $text;
3563         }
3564
3565         /**
3566          * Set up some variables which are usually set up in parse()
3567          * so that an external function can call some class members with confidence
3568          * @public
3569          */
3570         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
3571                 $this->mTitle =& $title;
3572                 $this->mOptions = $options;
3573                 $this->mOutputType = $outputType;
3574                 if ( $clearState ) {
3575                         $this->clearState();
3576                 }
3577         }
3578
3579         /**
3580          * Transform a MediaWiki message by replacing magic variables.
3581          *
3582          * @param string $text the text to transform
3583          * @param ParserOptions $options  options
3584          * @return string the text with variables substituted
3585          * @public
3586          */
3587         function transformMsg( $text, $options ) {
3588                 global $wgTitle;
3589                 static $executing = false;
3590
3591                 $fname = "Parser::transformMsg";
3592
3593                 # Guard against infinite recursion
3594                 if ( $executing ) {
3595                         return $text;
3596                 }
3597                 $executing = true;
3598
3599                 wfProfileIn($fname);
3600
3601                 $this->mTitle = $wgTitle;
3602                 $this->mOptions = $options;
3603                 $this->mOutputType = OT_MSG;
3604                 $this->clearState();
3605                 $text = $this->replaceVariables( $text );
3606
3607                 $executing = false;
3608                 wfProfileOut($fname);
3609                 return $text;
3610         }
3611
3612         /**
3613          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
3614          * The callback should have the following form:
3615          *    function myParserHook( $text, $params, &$parser ) { ... }
3616          *
3617          * Transform and return $text. Use $parser for any required context, e.g. use
3618          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
3619          *
3620          * @public
3621          *
3622          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
3623          * @param mixed $callback The callback function (and object) to use for the tag
3624          *
3625          * @return The old value of the mTagHooks array associated with the hook
3626          */
3627         function setHook( $tag, $callback ) {
3628                 $oldVal = @$this->mTagHooks[$tag];
3629                 $this->mTagHooks[$tag] = $callback;
3630
3631                 return $oldVal;
3632         }
3633
3634         /**
3635          * Create a function, e.g. {{sum:1|2|3}}
3636          * The callback function should have the form:
3637          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
3638          *
3639          * The callback may either return the text result of the function, or an array with the text
3640          * in element 0, and a number of flags in the other elements. The names of the flags are
3641          * specified in the keys. Valid flags are:
3642          *   found                     The text returned is valid, stop processing the template. This
3643          *                             is on by default.
3644          *   nowiki                    Wiki markup in the return value should be escaped
3645          *   noparse                   Unsafe HTML tags should not be stripped, etc.
3646          *   noargs                    Don't replace triple-brace arguments in the return value
3647          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
3648          *
3649          * @public
3650          *
3651          * @param string $name The function name. Function names are case-insensitive.
3652          * @param mixed $callback The callback function (and object) to use
3653          *
3654          * @return The old callback function for this name, if any
3655          */
3656         function setFunctionHook( $name, $callback ) {
3657                 $name = strtolower( $name );
3658                 $oldVal = @$this->mFunctionHooks[$name];
3659                 $this->mFunctionHooks[$name] = $callback;
3660                 return $oldVal;
3661         }
3662
3663         /**
3664          * Replace <!--LINK--> link placeholders with actual links, in the buffer
3665          * Placeholders created in Skin::makeLinkObj()
3666          * Returns an array of links found, indexed by PDBK:
3667          *  0 - broken
3668          *  1 - normal link
3669          *  2 - stub
3670          * $options is a bit field, RLH_FOR_UPDATE to select for update
3671          */
3672         function replaceLinkHolders( &$text, $options = 0 ) {
3673                 global $wgUser;
3674                 global $wgOutputReplace;
3675
3676                 $fname = 'Parser::replaceLinkHolders';
3677                 wfProfileIn( $fname );
3678
3679                 $pdbks = array();
3680                 $colours = array();
3681                 $sk =& $this->mOptions->getSkin();
3682                 $linkCache =& LinkCache::singleton();
3683
3684                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
3685                         wfProfileIn( $fname.'-check' );
3686                         $dbr =& wfGetDB( DB_SLAVE );
3687                         $page = $dbr->tableName( 'page' );
3688                         $threshold = $wgUser->getOption('stubthreshold');
3689
3690                         # Sort by namespace
3691                         asort( $this->mLinkHolders['namespaces'] );
3692
3693                         # Generate query
3694                         $query = false;
3695                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3696                                 # Make title object
3697                                 $title = $this->mLinkHolders['titles'][$key];
3698
3699                                 # Skip invalid entries.
3700                                 # Result will be ugly, but prevents crash.
3701                                 if ( is_null( $title ) ) {
3702                                         continue;
3703                                 }
3704                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
3705
3706                                 # Check if it's a static known link, e.g. interwiki
3707                                 if ( $title->isAlwaysKnown() ) {
3708                                         $colours[$pdbk] = 1;
3709                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
3710                                         $colours[$pdbk] = 1;
3711                                         $this->mOutput->addLink( $title, $id );
3712                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
3713                                         $colours[$pdbk] = 0;
3714                                 } else {
3715                                         # Not in the link cache, add it to the query
3716                                         if ( !isset( $current ) ) {
3717                                                 $current = $ns;
3718                                                 $query =  "SELECT page_id, page_namespace, page_title";
3719                                                 if ( $threshold > 0 ) {
3720                                                         $query .= ', page_len, page_is_redirect';
3721                                                 }
3722                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
3723                                         } elseif ( $current != $ns ) {
3724                                                 $current = $ns;
3725                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
3726                                         } else {
3727                                                 $query .= ', ';
3728                                         }
3729
3730                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
3731                                 }
3732                         }
3733                         if ( $query ) {
3734                                 $query .= '))';
3735                                 if ( $options & RLH_FOR_UPDATE ) {
3736                                         $query .= ' FOR UPDATE';
3737                                 }
3738
3739                                 $res = $dbr->query( $query, $fname );
3740
3741                                 # Fetch data and form into an associative array
3742                                 # non-existent = broken
3743                                 # 1 = known
3744                                 # 2 = stub
3745                                 while ( $s = $dbr->fetchObject($res) ) {
3746                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
3747                                         $pdbk = $title->getPrefixedDBkey();
3748                                         $linkCache->addGoodLinkObj( $s->page_id, $title );
3749                                         $this->mOutput->addLink( $title, $s->page_id );
3750
3751                                         if ( $threshold >  0 ) {
3752                                                 $size = $s->page_len;
3753                                                 if ( $s->page_is_redirect || $s->page_namespace != 0 || $size >= $threshold ) {
3754                                                         $colours[$pdbk] = 1;
3755                                                 } else {
3756                                                         $colours[$pdbk] = 2;
3757                                                 }
3758                                         } else {
3759                                                 $colours[$pdbk] = 1;
3760                                         }
3761                                 }
3762                         }
3763                         wfProfileOut( $fname.'-check' );
3764
3765                         # Construct search and replace arrays
3766                         wfProfileIn( $fname.'-construct' );
3767                         $wgOutputReplace = array();
3768                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3769                                 $pdbk = $pdbks[$key];
3770                                 $searchkey = "<!--LINK $key-->";
3771                                 $title = $this->mLinkHolders['titles'][$key];
3772                                 if ( empty( $colours[$pdbk] ) ) {
3773                                         $linkCache->addBadLinkObj( $title );
3774                                         $colours[$pdbk] = 0;
3775                                         $this->mOutput->addLink( $title, 0 );
3776                                         $wgOutputReplace[$searchkey] = $sk->makeBrokenLinkObj( $title,
3777                                                                         $this->mLinkHolders['texts'][$key],
3778                                                                         $this->mLinkHolders['queries'][$key] );
3779                                 } elseif ( $colours[$pdbk] == 1 ) {
3780                                         $wgOutputReplace[$searchkey] = $sk->makeKnownLinkObj( $title,
3781                                                                         $this->mLinkHolders['texts'][$key],
3782                                                                         $this->mLinkHolders['queries'][$key] );
3783                                 } elseif ( $colours[$pdbk] == 2 ) {
3784                                         $wgOutputReplace[$searchkey] = $sk->makeStubLinkObj( $title,
3785                                                                         $this->mLinkHolders['texts'][$key],
3786                                                                         $this->mLinkHolders['queries'][$key] );
3787                                 }
3788                         }
3789                         wfProfileOut( $fname.'-construct' );
3790
3791                         # Do the thing
3792                         wfProfileIn( $fname.'-replace' );
3793
3794                         $text = preg_replace_callback(
3795                                 '/(<!--LINK .*?-->)/',
3796                                 "wfOutputReplaceMatches",
3797                                 $text);
3798
3799                         wfProfileOut( $fname.'-replace' );
3800                 }
3801
3802                 # Now process interwiki link holders
3803                 # This is quite a bit simpler than internal links
3804                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
3805                         wfProfileIn( $fname.'-interwiki' );
3806                         # Make interwiki link HTML
3807                         $wgOutputReplace = array();
3808                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
3809                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
3810                                 $wgOutputReplace[$key] = $sk->makeLinkObj( $title, $link );
3811                         }
3812
3813                         $text = preg_replace_callback(
3814                                 '/<!--IWLINK (.*?)-->/',
3815                                 "wfOutputReplaceMatches",
3816                                 $text );
3817                         wfProfileOut( $fname.'-interwiki' );
3818                 }
3819
3820                 wfProfileOut( $fname );
3821                 return $colours;
3822         }
3823
3824         /**
3825          * Replace <!--LINK--> link placeholders with plain text of links
3826          * (not HTML-formatted).
3827          * @param string $text
3828          * @return string
3829          */
3830         function replaceLinkHoldersText( $text ) {
3831                 $fname = 'Parser::replaceLinkHoldersText';
3832                 wfProfileIn( $fname );
3833
3834                 $text = preg_replace_callback(
3835                         '/<!--(LINK|IWLINK) (.*?)-->/',
3836                         array( &$this, 'replaceLinkHoldersTextCallback' ),
3837                         $text );
3838
3839                 wfProfileOut( $fname );
3840                 return $text;
3841         }
3842
3843         /**
3844          * @param array $matches
3845          * @return string
3846          * @private
3847          */
3848         function replaceLinkHoldersTextCallback( $matches ) {
3849                 $type = $matches[1];
3850                 $key  = $matches[2];
3851                 if( $type == 'LINK' ) {
3852                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
3853                                 return $this->mLinkHolders['texts'][$key];
3854                         }
3855                 } elseif( $type == 'IWLINK' ) {
3856                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
3857                                 return $this->mInterwikiLinkHolders['texts'][$key];
3858                         }
3859                 }
3860                 return $matches[0];
3861         }
3862
3863         /**
3864          * Renders an image gallery from a text with one line per image.
3865          * text labels may be given by using |-style alternative text. E.g.
3866          *   Image:one.jpg|The number "1"
3867          *   Image:tree.jpg|A tree
3868          * given as text will return the HTML of a gallery with two images,
3869          * labeled 'The number "1"' and
3870          * 'A tree'.
3871          */
3872         function renderImageGallery( $text ) {
3873                 # Setup the parser
3874                 $parserOptions = new ParserOptions;
3875                 $localParser = new Parser();
3876
3877                 $ig = new ImageGallery();
3878                 $ig->setShowBytes( false );
3879                 $ig->setShowFilename( false );
3880                 $lines = explode( "\n", $text );
3881
3882                 foreach ( $lines as $line ) {
3883                         # match lines like these:
3884                         # Image:someimage.jpg|This is some image
3885                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
3886                         # Skip empty lines
3887                         if ( count( $matches ) == 0 ) {
3888                                 continue;
3889                         }
3890                         $nt =& Title::newFromText( $matches[1] );
3891                         if( is_null( $nt ) ) {
3892                                 # Bogus title. Ignore these so we don't bomb out later.
3893                                 continue;
3894                         }
3895                         if ( isset( $matches[3] ) ) {
3896                                 $label = $matches[3];
3897                         } else {
3898                                 $label = '';
3899                         }
3900
3901                         $pout = $localParser->parse( $label , $this->mTitle, $parserOptions );
3902                         $html = $pout->getText();
3903
3904                         $ig->add( new Image( $nt ), $html );
3905                         $this->mOutput->addImage( $nt->getDBkey() );
3906                 }
3907                 return $ig->toHTML();
3908         }
3909
3910         /**
3911          * Parse image options text and use it to make an image
3912          */
3913         function makeImage( &$nt, $options ) {
3914                 global $wgUseImageResize;
3915
3916                 $align = '';
3917
3918                 # Check if the options text is of the form "options|alt text"
3919                 # Options are:
3920                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
3921                 #  * left               no resizing, just left align. label is used for alt= only
3922                 #  * right              same, but right aligned
3923                 #  * none               same, but not aligned
3924                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
3925                 #  * center             center the image
3926                 #  * framed             Keep original image size, no magnify-button.
3927
3928                 $part = explode( '|', $options);
3929
3930                 $mwThumb  =& MagicWord::get( MAG_IMG_THUMBNAIL );
3931                 $mwManualThumb =& MagicWord::get( MAG_IMG_MANUALTHUMB );
3932                 $mwLeft   =& MagicWord::get( MAG_IMG_LEFT );
3933                 $mwRight  =& MagicWord::get( MAG_IMG_RIGHT );
3934                 $mwNone   =& MagicWord::get( MAG_IMG_NONE );
3935                 $mwWidth  =& MagicWord::get( MAG_IMG_WIDTH );
3936                 $mwCenter =& MagicWord::get( MAG_IMG_CENTER );
3937                 $mwFramed =& MagicWord::get( MAG_IMG_FRAMED );
3938                 $caption = '';
3939
3940                 $width = $height = $framed = $thumb = false;
3941                 $manual_thumb = '' ;
3942
3943                 foreach( $part as $key => $val ) {
3944                         if ( $wgUseImageResize && ! is_null( $mwThumb->matchVariableStartToEnd($val) ) ) {
3945                                 $thumb=true;
3946                         } elseif ( ! is_null( $match = $mwManualThumb->matchVariableStartToEnd($val) ) ) {
3947                                 # use manually specified thumbnail
3948                                 $thumb=true;
3949                                 $manual_thumb = $match;
3950                         } elseif ( ! is_null( $mwRight->matchVariableStartToEnd($val) ) ) {
3951                                 # remember to set an alignment, don't render immediately
3952                                 $align = 'right';
3953                         } elseif ( ! is_null( $mwLeft->matchVariableStartToEnd($val) ) ) {
3954                                 # remember to set an alignment, don't render immediately
3955                                 $align = 'left';
3956                         } elseif ( ! is_null( $mwCenter->matchVariableStartToEnd($val) ) ) {
3957                                 # remember to set an alignment, don't render immediately
3958                                 $align = 'center';
3959                         } elseif ( ! is_null( $mwNone->matchVariableStartToEnd($val) ) ) {
3960                                 # remember to set an alignment, don't render immediately
3961                                 $align = 'none';
3962                         } elseif ( $wgUseImageResize && ! is_null( $match = $mwWidth->matchVariableStartToEnd($val) ) ) {
3963                                 wfDebug( "MAG_IMG_WIDTH match: $match\n" );
3964                                 # $match is the image width in pixels
3965                                 if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $match, $m ) ) {
3966                                         $width = intval( $m[1] );
3967                                         $height = intval( $m[2] );
3968                                 } else {
3969                                         $width = intval($match);
3970                                 }
3971                         } elseif ( ! is_null( $mwFramed->matchVariableStartToEnd($val) ) ) {
3972                                 $framed=true;
3973                         } else {
3974                                 $caption = $val;
3975                         }
3976                 }
3977                 # Strip bad stuff out of the alt text
3978                 $alt = $this->replaceLinkHoldersText( $caption );
3979
3980                 # make sure there are no placeholders in thumbnail attributes
3981                 # that are later expanded to html- so expand them now and
3982                 # remove the tags
3983                 $alt = $this->unstrip($alt, $this->mStripState);
3984                 $alt = Sanitizer::stripAllTags( $alt );
3985
3986                 # Linker does the rest
3987                 $sk =& $this->mOptions->getSkin();
3988                 return $sk->makeImageLinkObj( $nt, $caption, $alt, $align, $width, $height, $framed, $thumb, $manual_thumb );
3989         }
3990
3991         /**
3992          * Set a flag in the output object indicating that the content is dynamic and
3993          * shouldn't be cached.
3994          */
3995         function disableCache() {
3996                 $this->mOutput->mCacheTime = -1;
3997         }
3998
3999         /**#@+
4000          * Callback from the Sanitizer for expanding items found in HTML attribute
4001          * values, so they can be safely tested and escaped.
4002          * @param string $text
4003          * @param array $args
4004          * @return string
4005          * @private
4006          */
4007         function attributeStripCallback( &$text, $args ) {
4008                 $text = $this->replaceVariables( $text, $args );
4009                 $text = $this->unstripForHTML( $text );
4010                 return $text;
4011         }
4012
4013         function unstripForHTML( $text ) {
4014                 $text = $this->unstrip( $text, $this->mStripState );
4015                 $text = $this->unstripNoWiki( $text, $this->mStripState );
4016                 return $text;
4017         }
4018         /**#@-*/
4019
4020         /**#@+
4021          * Accessor/mutator
4022          */
4023         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4024         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4025         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4026         /**#@-*/
4027
4028         /**#@+
4029          * Accessor
4030          */
4031         function getTags() { return array_keys( $this->mTagHooks ); }
4032         /**#@-*/
4033 }
4034
4035 /**
4036  * @todo document
4037  * @package MediaWiki
4038  */
4039 class ParserOutput
4040 {
4041         var $mText,             # The output text
4042                 $mLanguageLinks,    # List of the full text of language links, in the order they appear
4043                 $mCategories,       # Map of category names to sort keys
4044                 $mContainsOldMagic, # Boolean variable indicating if the input contained variables like {{CURRENTDAY}}
4045                 $mCacheTime,        # Time when this object was generated, or -1 for uncacheable. Used in ParserCache.
4046                 $mVersion,          # Compatibility check
4047                 $mTitleText,        # title text of the chosen language variant
4048                 $mLinks,            # 2-D map of NS/DBK to ID for the links in the document. ID=zero for broken.
4049                 $mTemplates,        # 2-D map of NS/DBK to ID for the template references. ID=zero for broken.
4050                 $mImages,           # DB keys of the images used, in the array key only
4051                 $mExternalLinks,    # External link URLs, in the key only
4052                 $mHTMLtitle,            # Display HTML title
4053                 $mSubtitle;                     # Additional subtitle
4054
4055         function ParserOutput( $text = '', $languageLinks = array(), $categoryLinks = array(),
4056                 $containsOldMagic = false, $titletext = '' )
4057         {
4058                 $this->mText = $text;
4059                 $this->mLanguageLinks = $languageLinks;
4060                 $this->mCategories = $categoryLinks;
4061                 $this->mContainsOldMagic = $containsOldMagic;
4062                 $this->mCacheTime = '';
4063                 $this->mVersion = MW_PARSER_VERSION;
4064                 $this->mTitleText = $titletext;
4065                 $this->mLinks = array();
4066                 $this->mTemplates = array();
4067                 $this->mImages = array();
4068                 $this->mExternalLinks = array();
4069                 $this->mHTMLtitle = "" ;
4070                 $this->mSubtitle = "" ;
4071         }
4072
4073         function getText()                   { return $this->mText; }
4074         function &getLanguageLinks()          { return $this->mLanguageLinks; }
4075         function getCategoryLinks()          { return array_keys( $this->mCategories ); }
4076         function &getCategories()            { return $this->mCategories; }
4077         function getCacheTime()              { return $this->mCacheTime; }
4078         function getTitleText()              { return $this->mTitleText; }
4079         function &getLinks()                 { return $this->mLinks; }
4080         function &getTemplates()             { return $this->mTemplates; }
4081         function &getImages()                { return $this->mImages; }
4082         function &getExternalLinks()         { return $this->mExternalLinks; }
4083
4084         function containsOldMagic()          { return $this->mContainsOldMagic; }
4085         function setText( $text )            { return wfSetVar( $this->mText, $text ); }
4086         function setLanguageLinks( $ll )     { return wfSetVar( $this->mLanguageLinks, $ll ); }
4087         function setCategoryLinks( $cl )     { return wfSetVar( $this->mCategories, $cl ); }
4088         function setContainsOldMagic( $com ) { return wfSetVar( $this->mContainsOldMagic, $com ); }
4089         function setCacheTime( $t )          { return wfSetVar( $this->mCacheTime, $t ); }
4090         function setTitleText( $t )          { return wfSetVar ($this->mTitleText, $t); }
4091
4092         function addCategory( $c, $sort )    { $this->mCategories[$c] = $sort; }
4093         function addImage( $name )           { $this->mImages[$name] = 1; }
4094         function addLanguageLink( $t )       { $this->mLanguageLinks[] = $t; }
4095         function addExternalLink( $url )     { $this->mExternalLinks[$url] = 1; }
4096
4097         function addLink( $title, $id ) {
4098                 $ns = $title->getNamespace();
4099                 $dbk = $title->getDBkey();
4100                 if ( !isset( $this->mLinks[$ns] ) ) {
4101                         $this->mLinks[$ns] = array();
4102                 }
4103                 $this->mLinks[$ns][$dbk] = $id;
4104         }
4105
4106         function addTemplate( $title, $id ) {
4107                 $ns = $title->getNamespace();
4108                 $dbk = $title->getDBkey();
4109                 if ( !isset( $this->mTemplates[$ns] ) ) {
4110                         $this->mTemplates[$ns] = array();
4111                 }
4112                 $this->mTemplates[$ns][$dbk] = $id;
4113         }
4114
4115         /**
4116          * Return true if this cached output object predates the global or
4117          * per-article cache invalidation timestamps, or if it comes from
4118          * an incompatible older version.
4119          *
4120          * @param string $touched the affected article's last touched timestamp
4121          * @return bool
4122          * @public
4123          */
4124         function expired( $touched ) {
4125                 global $wgCacheEpoch;
4126                 return $this->getCacheTime() == -1 || // parser says it's uncacheable
4127                        $this->getCacheTime() < $touched ||
4128                        $this->getCacheTime() <= $wgCacheEpoch ||
4129                        !isset( $this->mVersion ) ||
4130                        version_compare( $this->mVersion, MW_PARSER_VERSION, "lt" );
4131         }
4132 }
4133
4134 /**
4135  * Set options of the Parser
4136  * @todo document
4137  * @package MediaWiki
4138  */
4139 class ParserOptions
4140 {
4141         # All variables are private
4142         var $mUseTeX;                    # Use texvc to expand <math> tags
4143         var $mUseDynamicDates;           # Use DateFormatter to format dates
4144         var $mInterwikiMagic;            # Interlanguage links are removed and returned in an array
4145         var $mAllowExternalImages;       # Allow external images inline
4146         var $mAllowExternalImagesFrom;   # If not, any exception?
4147         var $mSkin;                      # Reference to the preferred skin
4148         var $mDateFormat;                # Date format index
4149         var $mEditSection;               # Create "edit section" links
4150         var $mNumberHeadings;            # Automatically number headings
4151         var $mAllowSpecialInclusion;     # Allow inclusion of special pages
4152         var $mTidy;                      # Ask for tidy cleanup
4153
4154         function getUseTeX()                        { return $this->mUseTeX; }
4155         function getUseDynamicDates()               { return $this->mUseDynamicDates; }
4156         function getInterwikiMagic()                { return $this->mInterwikiMagic; }
4157         function getAllowExternalImages()           { return $this->mAllowExternalImages; }
4158         function getAllowExternalImagesFrom()       { return $this->mAllowExternalImagesFrom; }
4159         function &getSkin()                         { return $this->mSkin; }
4160         function getDateFormat()                    { return $this->mDateFormat; }
4161         function getEditSection()                   { return $this->mEditSection; }
4162         function getNumberHeadings()                { return $this->mNumberHeadings; }
4163         function getAllowSpecialInclusion()         { return $this->mAllowSpecialInclusion; }
4164         function getTidy()                          { return $this->mTidy; }
4165
4166         function setUseTeX( $x )                    { return wfSetVar( $this->mUseTeX, $x ); }
4167         function setUseDynamicDates( $x )           { return wfSetVar( $this->mUseDynamicDates, $x ); }
4168         function setInterwikiMagic( $x )            { return wfSetVar( $this->mInterwikiMagic, $x ); }
4169         function setAllowExternalImages( $x )       { return wfSetVar( $this->mAllowExternalImages, $x ); }
4170         function setAllowExternalImagesFrom( $x )   { return wfSetVar( $this->mAllowExternalImagesFrom, $x ); }
4171         function setDateFormat( $x )                { return wfSetVar( $this->mDateFormat, $x ); }
4172         function setEditSection( $x )               { return wfSetVar( $this->mEditSection, $x ); }
4173         function setNumberHeadings( $x )            { return wfSetVar( $this->mNumberHeadings, $x ); }
4174         function setAllowSpecialInclusion( $x )     { return wfSetVar( $this->mAllowSpecialInclusion, $x ); }
4175         function setTidy( $x )                      { return wfSetVar( $this->mTidy, $x); }
4176         function setSkin( &$x ) { $this->mSkin =& $x; }
4177
4178         function ParserOptions() {
4179                 global $wgUser;
4180                 $this->initialiseFromUser( $wgUser );
4181         }
4182
4183         /**
4184          * Get parser options
4185          * @static
4186          */
4187         function newFromUser( &$user ) {
4188                 $popts = new ParserOptions;
4189                 $popts->initialiseFromUser( $user );
4190                 return $popts;
4191         }
4192
4193         /** Get user options */
4194         function initialiseFromUser( &$userInput ) {
4195                 global $wgUseTeX, $wgUseDynamicDates, $wgInterwikiMagic, $wgAllowExternalImages;
4196                 global $wgAllowExternalImagesFrom, $wgAllowSpecialInclusion;
4197                 $fname = 'ParserOptions::initialiseFromUser';
4198                 wfProfileIn( $fname );
4199                 if ( !$userInput ) {
4200                         $user = new User;
4201                         $user->setLoaded( true );
4202                 } else {
4203                         $user =& $userInput;
4204                 }
4205
4206                 $this->mUseTeX = $wgUseTeX;
4207                 $this->mUseDynamicDates = $wgUseDynamicDates;
4208                 $this->mInterwikiMagic = $wgInterwikiMagic;
4209                 $this->mAllowExternalImages = $wgAllowExternalImages;
4210                 $this->mAllowExternalImagesFrom = $wgAllowExternalImagesFrom;
4211                 wfProfileIn( $fname.'-skin' );
4212                 $this->mSkin =& $user->getSkin();
4213                 wfProfileOut( $fname.'-skin' );
4214                 $this->mDateFormat = $user->getOption( 'date' );
4215                 $this->mEditSection = true;
4216                 $this->mNumberHeadings = $user->getOption( 'numberheadings' );
4217                 $this->mAllowSpecialInclusion = $wgAllowSpecialInclusion;
4218                 $this->mTidy = false;
4219                 wfProfileOut( $fname );
4220         }
4221 }
4222
4223 /**
4224  * Callback function used by Parser::replaceLinkHolders()
4225  * to substitute link placeholders.
4226  */
4227 function &wfOutputReplaceMatches( $matches ) {
4228         global $wgOutputReplace;
4229         return $wgOutputReplace[$matches[1]];
4230 }
4231
4232 /**
4233  * Return the total number of articles
4234  */
4235 function wfNumberOfArticles() {
4236         global $wgNumberOfArticles;
4237
4238         wfLoadSiteStats();
4239         return $wgNumberOfArticles;
4240 }
4241
4242 /**
4243  * Return the number of files
4244  */
4245 function wfNumberOfFiles() {
4246         $fname = 'wfNumberOfFiles';
4247
4248         wfProfileIn( $fname );
4249         $dbr =& wfGetDB( DB_SLAVE );
4250         $numImages = $dbr->selectField('site_stats', 'ss_images', array(), $fname );
4251         wfProfileOut( $fname );
4252
4253         return $numImages;
4254 }
4255
4256 /**
4257  * Get various statistics from the database
4258  * @private
4259  */
4260 function wfLoadSiteStats() {
4261         global $wgNumberOfArticles, $wgTotalViews, $wgTotalEdits;
4262         $fname = 'wfLoadSiteStats';
4263
4264         if ( -1 != $wgNumberOfArticles ) return;
4265         $dbr =& wfGetDB( DB_SLAVE );
4266         $s = $dbr->selectRow( 'site_stats',
4267                 array( 'ss_total_views', 'ss_total_edits', 'ss_good_articles' ),
4268                 array( 'ss_row_id' => 1 ), $fname
4269         );
4270
4271         if ( $s === false ) {
4272                 return;
4273         } else {
4274                 $wgTotalViews = $s->ss_total_views;
4275                 $wgTotalEdits = $s->ss_total_edits;
4276                 $wgNumberOfArticles = $s->ss_good_articles;
4277         }
4278 }
4279
4280 /**
4281  * Escape html tags
4282  * Basically replacing " > and < with HTML entities ( &quot;, &gt;, &lt;)
4283  *
4284  * @param $in String: text that might contain HTML tags.
4285  * @return string Escaped string
4286  */
4287 function wfEscapeHTMLTagsOnly( $in ) {
4288         return str_replace(
4289                 array( '"', '>', '<' ),
4290                 array( '&quot;', '&gt;', '&lt;' ),
4291                 $in );
4292 }
4293
4294 ?>