includes/Parser.php

   1 <?php
   2 /**
   3  * File for Parser and related classes
   4  *
   5  * @package MediaWiki
   6  * @subpackage Parser
   7  */
   8
   9 /**
  10  * Update this version number when the ParserOutput format
  11  * changes in an incompatible way, so the parser cache
  12  * can automatically discard old data.
  13  */
  14 define( 'MW_PARSER_VERSION', '1.6.1' );
  15
  16 /**
  17  * Variable substitution O(N^2) attack
  18  *
  19  * Without countermeasures, it would be possible to attack the parser by saving
  20  * a page filled with a large number of inclusions of large pages. The size of
  21  * the generated page would be proportional to the square of the input size.
  22  * Hence, we limit the number of inclusions of any given page, thus bringing any
  23  * attack back to O(N).
  24  */
  25
  26 define( 'MAX_INCLUDE_REPEAT', 100 );
  27 define( 'MAX_INCLUDE_SIZE', 1000000 ); // 1 Million
  28
  29 define( 'RLH_FOR_UPDATE', 1 );
  30
  31 # Allowed values for $mOutputType
  32 define( 'OT_HTML', 1 );
  33 define( 'OT_WIKI', 2 );
  34 define( 'OT_MSG' , 3 );
  35
  36 # string parameter for extractTags which will cause it
  37 # to strip HTML comments in addition to regular
  38 # <XML>-style tags. This should not be anything we
  39 # may want to use in wikisyntax
  40 define( 'STRIP_COMMENTS', 'HTMLCommentStrip' );
  41
  42 # Constants needed for external link processing
  43 define( 'HTTP_PROTOCOLS', 'http:\/\/|https:\/\/' );
  44 # Everything except bracket, space, or control characters
  45 define( 'EXT_LINK_URL_CLASS', '[^][<>"\\x00-\\x20\\x7F]' );
  46 # Including space
  47 define( 'EXT_LINK_TEXT_CLASS', '[^\]\\x00-\\x1F\\x7F]' );
  48 define( 'EXT_IMAGE_FNAME_CLASS', '[A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]' );
  49 define( 'EXT_IMAGE_EXTENSIONS', 'gif|png|jpg|jpeg' );
  50 define( 'EXT_LINK_BRACKETED',  '/\[(\b(' . wfUrlProtocols() . ')'.EXT_LINK_URL_CLASS.'+) *('.EXT_LINK_TEXT_CLASS.'*?)\]/S' );
  51 define( 'EXT_IMAGE_REGEX',
  52         '/^('.HTTP_PROTOCOLS.')'.  # Protocol
  53         '('.EXT_LINK_URL_CLASS.'+)\\/'.  # Hostname and path
  54         '('.EXT_IMAGE_FNAME_CLASS.'+)\\.((?i)'.EXT_IMAGE_EXTENSIONS.')$/S' # Filename
  55 );
  56
  57 /**
  58  * PHP Parser
  59  *
  60  * Processes wiki markup
  61  *
  62  * <pre>
  63  * There are three main entry points into the Parser class:
  64  * parse()
  65  *   produces HTML output
  66  * preSaveTransform().
  67  *   produces altered wiki markup.
  68  * transformMsg()
  69  *   performs brace substitution on MediaWiki messages
  70  *
  71  * Globals used:
  72  *    objects:   $wgLang, $wgContLang
  73  *
  74  * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
  75  *
  76  * settings:
  77  *  $wgUseTex*, $wgUseDynamicDates*, $wgInterwikiMagic*,
  78  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  79  *  $wgLocaltimezone, $wgAllowSpecialInclusion*
  80  *
  81  *  * only within ParserOptions
  82  * </pre>
  83  *
  84  * @package MediaWiki
  85  */
  86 class Parser
  87 {
  88         /**#@+
  89          * @private
  90          */
  91         # Persistent:
  92         var $mTagHooks, $mFunctionHooks;
  93
  94         # Cleared with clearState():
  95         var $mOutput, $mAutonumber, $mDTopen, $mStripState = array();
  96         var $mVariables, $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
  97         var $mInterwikiLinkHolders, $mLinkHolders, $mUniqPrefix;
  98         var $mTemplates,        // cache of already loaded templates, avoids
  99                                 // multiple SQL queries for the same string
 100             $mTemplatePath;     // stores an unsorted hash of all the templates already loaded
 101                                 // in this path. Used for loop detection.
 102
 103         # Temporary
 104         # These are variables reset at least once per parse regardless of $clearState
 105         var $mOptions,      // ParserOptions object
 106                 $mTitle,        // Title context, used for self-link rendering and similar things
 107                 $mOutputType,   // Output type, one of the OT_xxx constants
 108                 $mRevisionId;   // ID to display in {{REVISIONID}} tags
 109
 110         /**#@-*/
 111
 112         /**
 113          * Constructor
 114          *
 115          * @public
 116          */
 117         function Parser() {
 118                 $this->mTagHooks = array();
 119                 $this->mFunctionHooks = array();
 120                 $this->clearState();
 121         }
 122
 123         /**
 124          * Clear Parser state
 125          *
 126          * @private
 127          */
 128         function clearState() {
 129                 $this->mOutput = new ParserOutput;
 130                 $this->mAutonumber = 0;
 131                 $this->mLastSection = '';
 132                 $this->mDTopen = false;
 133                 $this->mVariables = false;
 134                 $this->mIncludeCount = array();
 135                 $this->mStripState = array();
 136                 $this->mArgStack = array();
 137                 $this->mInPre = false;
 138                 $this->mInterwikiLinkHolders = array(
 139                         'texts' => array(),
 140                         'titles' => array()
 141                 );
 142                 $this->mLinkHolders = array(
 143                         'namespaces' => array(),
 144                         'dbkeys' => array(),
 145                         'queries' => array(),
 146                         'texts' => array(),
 147                         'titles' => array()
 148                 );
 149                 $this->mRevisionId = null;
 150
 151                 /**
 152                  * Prefix for temporary replacement strings for the multipass parser.
 153                  * \x07 should never appear in input as it's disallowed in XML.
 154                  * Using it at the front also gives us a little extra robustness
 155                  * since it shouldn't match when butted up against identifier-like
 156                  * string constructs.
 157                  */
 158                 $this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 159
 160                 # Clear these on every parse, bug 4549
 161                 $this->mTemplates = array();
 162                 $this->mTemplatePath = array();
 163
 164                 $this->mShowToc = true;
 165                 $this->mForceTocPosition = false;
 166
 167                 wfRunHooks( 'ParserClearState', array( &$this ) );
 168         }
 169
 170         /**
 171          * Accessor for mUniqPrefix.
 172          *
 173          * @public
 174          */
 175         function UniqPrefix() {
 176                 return $this->mUniqPrefix;
 177         }
 178
 179         /**
 180          * Convert wikitext to HTML
 181          * Do not call this function recursively.
 182          *
 183          * @private
 184          * @param string $text Text we want to parse
 185          * @param Title &$title A title object
 186          * @param array $options
 187          * @param boolean $linestart
 188          * @param boolean $clearState
 189          * @param int $revid number to pass in {{REVISIONID}}
 190          * @return ParserOutput a ParserOutput
 191          */
 192         function parse( $text, &$title, $options, $linestart = true, $clearState = true, $revid = null ) {
 193                 /**
 194                  * First pass--just handle <nowiki> sections, pass the rest off
 195                  * to internalParse() which does all the real work.
 196                  */
 197
 198                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
 199                 $fname = 'Parser::parse';
 200                 wfProfileIn( $fname );
 201
 202                 if ( $clearState ) {
 203                         $this->clearState();
 204                 }
 205
 206                 $this->mOptions = $options;
 207                 $this->mTitle =& $title;
 208                 $this->mRevisionId = $revid;
 209                 $this->mOutputType = OT_HTML;
 210
 211                 //$text = $this->strip( $text, $this->mStripState );
 212                 // VOODOO MAGIC FIX! Sometimes the above segfaults in PHP5.
 213                 $x =& $this->mStripState;
 214
 215                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$x ) );
 216                 $text = $this->strip( $text, $x );
 217                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$x ) );
 218
 219                 # Hook to suspend the parser in this state
 220                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$x ) ) ) {
 221                         wfProfileOut( $fname );
 222                         return $text ;
 223                 }
 224
 225                 $text = $this->internalParse( $text );
 226
 227                 $text = $this->unstrip( $text, $this->mStripState );
 228
 229                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 230                 $fixtags = array(
 231                         # french spaces, last one Guillemet-left
 232                         # only if there is something before the space
 233                         '/(.) (?=\\?|:|;|!|\\302\\273)/' => '\\1&nbsp;\\2',
 234                         # french spaces, Guillemet-right
 235                         '/(\\302\\253) /' => '\\1&nbsp;',
 236                         '/<center *>(.*)<\\/center *>/i' => '<div class="center">\\1</div>',
 237                 );
 238                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 239
 240                 # only once and last
 241                 $text = $this->doBlockLevels( $text, $linestart );
 242
 243                 $this->replaceLinkHolders( $text );
 244
 245                 # the position of the parserConvert() call should not be changed. it
 246                 # assumes that the links are all replaced and the only thing left
 247                 # is the <nowiki> mark.
 248                 # Side-effects: this calls $this->mOutput->setTitleText()
 249                 $text = $wgContLang->parserConvert( $text, $this );
 250
 251                 $text = $this->unstripNoWiki( $text, $this->mStripState );
 252
 253                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 254
 255                 $text = Sanitizer::normalizeCharReferences( $text );
 256
 257                 if (($wgUseTidy and $this->mOptions->mTidy) or $wgAlwaysUseTidy) {
 258                         $text = Parser::tidy($text);
 259                 } else {
 260                         # attempt to sanitize at least some nesting problems
 261                         # (bug #2702 and quite a few others)
 262                         $tidyregs = array(
 263                                 # ''Something [http://www.cool.com cool''] -->
 264                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 265                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 266                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 267                                 # fix up an anchor inside another anchor, only
 268                                 # at least for a single single nested link (bug 3695)
 269                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 270                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 271                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 272                                 # contains a div, so fix it up here; replace
 273                                 # div with escaped text
 274                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 275                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 276                                 # remove empty italic or bold tag pairs, some
 277                                 # introduced by rules above
 278                                 '/<([bi])><\/\\1>/' => ''
 279                         );
 280
 281                         $text = preg_replace(
 282                                 array_keys( $tidyregs ),
 283                                 array_values( $tidyregs ),
 284                                 $text );
 285                 }
 286
 287                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 288
 289                 $this->mOutput->setText( $text );
 290                 wfProfileOut( $fname );
 291
 292                 return $this->mOutput;
 293         }
 294
 295         /**
 296          * Get a random string
 297          *
 298          * @private
 299          * @static
 300          */
 301         function getRandomString() {
 302                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 303         }
 304
 305         function &getTitle() { return $this->mTitle; }
 306         function getOptions() { return $this->mOptions; }
 307
 308         /**
 309          * Replaces all occurrences of <$tag>content</$tag> in the text
 310          * with a random marker and returns the new text. the output parameter
 311          * $content will be an associative array filled with data on the form
 312          * $unique_marker => content.
 313          *
 314          * If $content is already set, the additional entries will be appended
 315          * If $tag is set to STRIP_COMMENTS, the function will extract
 316          * <!-- HTML comments -->
 317          *
 318          * $output: array( 'UNIQ-xxxxx' => array(
 319          *    'element',
 320          *    'tag content',
 321          *    array( 'param' => 'x' ),
 322          *    '<element param="x">' ) )
 323          * @private
 324          * @static
 325          */
 326         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 327                 $rand = Parser::getRandomString();
 328                 $n = 1;
 329                 $stripped = '';
 330                 $matches = array();
 331
 332                 if( $elements == STRIP_COMMENTS ) {
 333                         $start = '/<!--()()/';
 334                 } else {
 335                         $taglist = implode( '|', $elements );
 336                         $start = "/<($taglist)(\\s+[^>]*|\\s*\/?)>/i";
 337                 }
 338
 339                 while ( '' != $text ) {
 340                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 341                         $stripped .= $p[0];
 342                         if( count( $p ) < 4 ) {
 343                                 break;
 344                         }
 345                         $element    = $p[1];
 346                         $attributes = $p[2];
 347                         $inside     = $p[3];
 348
 349                         // If $attributes ends with '/', we have an empty element tag, <tag />
 350                         if( $element != '' && substr( $attributes, -1 ) == '/' ) {
 351                                 $attributes = substr( $attributes, 0, -1);
 352                                 $empty = '/';
 353                         } else {
 354                                 $empty = '';
 355                         }
 356
 357                         $marker = "$uniq_prefix-$element-$rand" . sprintf('%08X', $n++);
 358                         $stripped .= $marker;
 359
 360                         if ( $empty === '/' ) {
 361                                 // Empty element tag, <tag />
 362                                 $content = null;
 363                                 $text = $inside;
 364                         } else {
 365                                 if( $element ) {
 366                                         $end = "/<\\/$element\\s*>/i";
 367                                 } else {
 368                                         $end = '/-->/';
 369                                 }
 370                                 $q = preg_split( $end, $inside, 2 );
 371                                 $content = $q[0];
 372                                 if( count( $q ) < 2 ) {
 373                                         # No end tag -- let it run out to the end of the text.
 374                                         break;
 375                                 } else {
 376                                         $text = $q[1];
 377                                 }
 378                         }
 379
 380                         $matches[$marker] = array( $element,
 381                                 $content,
 382                                 Sanitizer::decodeTagAttributes( $attributes ),
 383                                 "<$element$attributes$empty>" );
 384                 }
 385                 return $stripped;
 386         }
 387
 388         /**
 389          * Strips and renders nowiki, pre, math, hiero
 390          * If $render is set, performs necessary rendering operations on plugins
 391          * Returns the text, and fills an array with data needed in unstrip()
 392          * If the $state is already a valid strip state, it adds to the state
 393          *
 394          * @param bool $stripcomments when set, HTML comments <!-- like this -->
 395          *  will be stripped in addition to other tags. This is important
 396          *  for section editing, where these comments cause confusion when
 397          *  counting the sections in the wikisource
 398          *
 399          * @private
 400          */
 401         function strip( $text, &$state, $stripcomments = false ) {
 402                 $render = ($this->mOutputType == OT_HTML);
 403
 404                 # Replace any instances of the placeholders
 405                 $uniq_prefix = $this->mUniqPrefix;
 406                 #$text = str_replace( $uniq_prefix, wfHtmlEscapeFirst( $uniq_prefix ), $text );
 407
 408                 $elements = array_merge(
 409                         array( 'nowiki', 'pre', 'gallery' ),
 410                         array_keys( $this->mTagHooks ) );
 411                 global $wgRawHtml;
 412                 if( $wgRawHtml ) {
 413                         $elements[] = 'html';
 414                 }
 415                 if( $this->mOptions->getUseTeX() ) {
 416                         $elements[] = 'math';
 417                 }
 418
 419
 420                 // Strip comments in a first pass.
 421                 // This saves us from needlessly rendering extensions in comment text
 422                 $text = Parser::extractTagsAndParams(STRIP_COMMENTS, $text, $comment_matches, $uniq_prefix);
 423                 $commentState = array();
 424                 foreach( $comment_matches as $marker => $data ){
 425                         list( $element, $content, $params, $tag ) = $data;
 426                         $commentState[$marker] = '<!--' . $content . '-->';
 427                 }
 428
 429                 $matches = array();
 430                 $text = Parser::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 431
 432                 foreach( $matches as $marker => $data ) {
 433                         list( $element, $content, $params, $tag ) = $data;
 434                         // Restore any comments; the extension can deal with them.
 435                         if( $content !== null) {
 436                                 $content = strtr( $content, $commentState );
 437                         }
 438                         if( $render ) {
 439                                 switch( $element ) {
 440                                 case 'html':
 441                                         if( $wgRawHtml ) {
 442                                                 $output = $content;
 443                                                 break;
 444                                         }
 445                                         // Shouldn't happen otherwise. :)
 446                                 case 'nowiki':
 447                                         $output = wfEscapeHTMLTagsOnly( $content );
 448                                         break;
 449                                 case 'math':
 450                                         $output = MathRenderer::renderMath( $content );
 451                                         break;
 452                                 case 'pre':
 453                                         // Backwards-compatibility hack
 454                                         $content = preg_replace( '!<nowiki>(.*?)</nowiki>!is', '\\1', $content );
 455                                         $output = '<pre>' . wfEscapeHTMLTagsOnly( $content ) . '</pre>';
 456                                         break;
 457                                 case 'gallery':
 458                                         $output = $this->renderImageGallery( $content );
 459                                         break;
 460                                 default:
 461                                         $tagName = strtolower( $element );
 462                                         if( isset( $this->mTagHooks[$tagName] ) ) {
 463                                                 $output = call_user_func_array( $this->mTagHooks[$tagName],
 464                                                         array( $content, $params, $this ) );
 465                                         } else {
 466                                                 wfDebugDieBacktrace( "Invalid call hook $element" );
 467                                         }
 468                                 }
 469                         } else {
 470                                 // Just stripping tags; keep the source
 471                                 if( $content === null ) {
 472                                         $output = $tag;
 473                                 } else {
 474                                         $output = "$tag$content</$element>";
 475                                 }
 476                         }
 477                         $state[$element][$marker] = $output;
 478                 }
 479
 480                 # Unstrip comments unless explicitly told otherwise.
 481                 # (The comments are always stripped prior to this point, so as to
 482                 # not invoke any extension tags / parser hooks contained within
 483                 # a comment.)
 484                 if ( $stripcomments ) {
 485                         // Add remaining comments to the state array
 486                         foreach( $commentState as $marker => $content ) {
 487                                 $state['comment'][$marker] = $content;
 488                         }
 489                 } else {
 490                         // Put them all back and forget them
 491                         $text = strtr( $text, $commentState );
 492                 }
 493
 494                 return $text;
 495         }
 496
 497         /**
 498          * Restores pre, math, and other extensions removed by strip()
 499          *
 500          * always call unstripNoWiki() after this one
 501          * @private
 502          */
 503         function unstrip( $text, &$state ) {
 504                 if ( !is_array( $state ) ) {
 505                         return $text;
 506                 }
 507
 508                 $replacements = array();
 509                 foreach( $state as $tag => $contentDict ) {
 510                         if( $tag != 'nowiki' && $tag != 'html' ) {
 511                                 foreach( $contentDict as $uniq => $content ) {
 512                                         $replacements[$uniq] = $content;
 513                                 }
 514                         }
 515                 }
 516                 $text = strtr( $text, $replacements );
 517
 518                 return $text;
 519         }
 520
 521         /**
 522          * Always call this after unstrip() to preserve the order
 523          *
 524          * @private
 525          */
 526         function unstripNoWiki( $text, &$state ) {
 527                 if ( !is_array( $state ) ) {
 528                         return $text;
 529                 }
 530
 531                 $replacements = array();
 532                 foreach( $state as $tag => $contentDict ) {
 533                         if( $tag == 'nowiki' || $tag == 'html' ) {
 534                                 foreach( $contentDict as $uniq => $content ) {
 535                                         $replacements[$uniq] = $content;
 536                                 }
 537                         }
 538                 }
 539                 $text = strtr( $text, $replacements );
 540
 541                 return $text;
 542         }
 543
 544         /**
 545          * Add an item to the strip state
 546          * Returns the unique tag which must be inserted into the stripped text
 547          * The tag will be replaced with the original text in unstrip()
 548          *
 549          * @private
 550          */
 551         function insertStripItem( $text, &$state ) {
 552                 $rnd = $this->mUniqPrefix . '-item' . Parser::getRandomString();
 553                 if ( !$state ) {
 554                         $state = array();
 555                 }
 556                 $state['item'][$rnd] = $text;
 557                 return $rnd;
 558         }
 559
 560         /**
 561          * Interface with html tidy, used if $wgUseTidy = true.
 562          * If tidy isn't able to correct the markup, the original will be
 563          * returned in all its glory with a warning comment appended.
 564          *
 565          * Either the external tidy program or the in-process tidy extension
 566          * will be used depending on availability. Override the default
 567          * $wgTidyInternal setting to disable the internal if it's not working.
 568          *
 569          * @param string $text Hideous HTML input
 570          * @return string Corrected HTML output
 571          * @public
 572          * @static
 573          */
 574         function tidy( $text ) {
 575                 global $wgTidyInternal;
 576                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 577 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 578 '<head><title>test</title></head><body>'.$text.'</body></html>';
 579                 if( $wgTidyInternal ) {
 580                         $correctedtext = Parser::internalTidy( $wrappedtext );
 581                 } else {
 582                         $correctedtext = Parser::externalTidy( $wrappedtext );
 583                 }
 584                 if( is_null( $correctedtext ) ) {
 585                         wfDebug( "Tidy error detected!\n" );
 586                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 587                 }
 588                 return $correctedtext;
 589         }
 590
 591         /**
 592          * Spawn an external HTML tidy process and get corrected markup back from it.
 593          *
 594          * @private
 595          * @static
 596          */
 597         function externalTidy( $text ) {
 598                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 599                 $fname = 'Parser::externalTidy';
 600                 wfProfileIn( $fname );
 601
 602                 $cleansource = '';
 603                 $opts = ' -utf8';
 604
 605                 $descriptorspec = array(
 606                         0 => array('pipe', 'r'),
 607                         1 => array('pipe', 'w'),
 608                         2 => array('file', '/dev/null', 'a')
 609                 );
 610                 $pipes = array();
 611                 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 612                 if (is_resource($process)) {
 613                         // Theoretically, this style of communication could cause a deadlock
 614                         // here. If the stdout buffer fills up, then writes to stdin could
 615                         // block. This doesn't appear to happen with tidy, because tidy only
 616                         // writes to stdout after it's finished reading from stdin. Search
 617                         // for tidyParseStdin and tidySaveStdout in console/tidy.c
 618                         fwrite($pipes[0], $text);
 619                         fclose($pipes[0]);
 620                         while (!feof($pipes[1])) {
 621                                 $cleansource .= fgets($pipes[1], 1024);
 622                         }
 623                         fclose($pipes[1]);
 624                         proc_close($process);
 625                 }
 626
 627                 wfProfileOut( $fname );
 628
 629                 if( $cleansource == '' && $text != '') {
 630                         // Some kind of error happened, so we couldn't get the corrected text.
 631                         // Just give up; we'll use the source text and append a warning.
 632                         return null;
 633                 } else {
 634                         return $cleansource;
 635                 }
 636         }
 637
 638         /**
 639          * Use the HTML tidy PECL extension to use the tidy library in-process,
 640          * saving the overhead of spawning a new process. Currently written to
 641          * the PHP 4.3.x version of the extension, may not work on PHP 5.
 642          *
 643          * 'pear install tidy' should be able to compile the extension module.
 644          *
 645          * @private
 646          * @static
 647          */
 648         function internalTidy( $text ) {
 649                 global $wgTidyConf;
 650                 $fname = 'Parser::internalTidy';
 651                 wfProfileIn( $fname );
 652
 653                 tidy_load_config( $wgTidyConf );
 654                 tidy_set_encoding( 'utf8' );
 655                 tidy_parse_string( $text );
 656                 tidy_clean_repair();
 657                 if( tidy_get_status() == 2 ) {
 658                         // 2 is magic number for fatal error
 659                         // http://www.php.net/manual/en/function.tidy-get-status.php
 660                         $cleansource = null;
 661                 } else {
 662                         $cleansource = tidy_get_output();
 663                 }
 664                 wfProfileOut( $fname );
 665                 return $cleansource;
 666         }
 667
 668         /**
 669          * parse the wiki syntax used to render tables
 670          *
 671          * @private
 672          */
 673         function doTableStuff ( $t ) {
 674                 $fname = 'Parser::doTableStuff';
 675                 wfProfileIn( $fname );
 676
 677                 $t = explode ( "\n" , $t ) ;
 678                 $td = array () ; # Is currently a td tag open?
 679                 $ltd = array () ; # Was it TD or TH?
 680                 $tr = array () ; # Is currently a tr tag open?
 681                 $ltr = array () ; # tr attributes
 682                 $has_opened_tr = array(); # Did this table open a <tr> element?
 683                 $indent_level = 0; # indent level of the table
 684                 foreach ( $t AS $k => $x )
 685                 {
 686                         $x = trim ( $x ) ;
 687                         $fc = substr ( $x , 0 , 1 ) ;
 688                         if ( preg_match( '/^(:*)\{\|(.*)$/', $x, $matches ) ) {
 689                                 $indent_level = strlen( $matches[1] );
 690
 691                                 $attributes = $this->unstripForHTML( $matches[2] );
 692
 693                                 $t[$k] = str_repeat( '<dl><dd>', $indent_level ) .
 694                                         '<table' . Sanitizer::fixTagAttributes ( $attributes, 'table' ) . '>' ;
 695                                 array_push ( $td , false ) ;
 696                                 array_push ( $ltd , '' ) ;
 697                                 array_push ( $tr , false ) ;
 698                                 array_push ( $ltr , '' ) ;
 699                                 array_push ( $has_opened_tr, false );
 700                         }
 701                         else if ( count ( $td ) == 0 ) { } # Don't do any of the following
 702                         else if ( '|}' == substr ( $x , 0 , 2 ) ) {
 703                                 $z = "</table>" . substr ( $x , 2);
 704                                 $l = array_pop ( $ltd ) ;
 705                                 if ( !array_pop ( $has_opened_tr ) ) $z = "<tr><td></td></tr>" . $z ;
 706                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 707                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 708                                 array_pop ( $ltr ) ;
 709                                 $t[$k] = $z . str_repeat( '</dd></dl>', $indent_level );
 710                         }
 711                         else if ( '|-' == substr ( $x , 0 , 2 ) ) { # Allows for |---------------
 712                                 $x = substr ( $x , 1 ) ;
 713                                 while ( $x != '' && substr ( $x , 0 , 1 ) == '-' ) $x = substr ( $x , 1 ) ;
 714                                 $z = '' ;
 715                                 $l = array_pop ( $ltd ) ;
 716                                 array_pop ( $has_opened_tr );
 717                                 array_push ( $has_opened_tr , true ) ;
 718                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 719                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 720                                 array_pop ( $ltr ) ;
 721                                 $t[$k] = $z ;
 722                                 array_push ( $tr , false ) ;
 723                                 array_push ( $td , false ) ;
 724                                 array_push ( $ltd , '' ) ;
 725                                 $attributes = $this->unstripForHTML( $x );
 726                                 array_push ( $ltr , Sanitizer::fixTagAttributes ( $attributes, 'tr' ) ) ;
 727                         }
 728                         else if ( '|' == $fc || '!' == $fc || '|+' == substr ( $x , 0 , 2 ) ) { # Caption
 729                                 # $x is a table row
 730                                 if ( '|+' == substr ( $x , 0 , 2 ) ) {
 731                                         $fc = '+' ;
 732                                         $x = substr ( $x , 1 ) ;
 733                                 }
 734                                 $after = substr ( $x , 1 ) ;
 735                                 if ( $fc == '!' ) $after = str_replace ( '!!' , '||' , $after ) ;
 736
 737                                 // Split up multiple cells on the same line.
 738                                 // FIXME: This can result in improper nesting of tags processed
 739                                 // by earlier parser steps, but should avoid splitting up eg
 740                                 // attribute values containing literal "||".
 741                                 $after = wfExplodeMarkup( '||', $after );
 742
 743                                 $t[$k] = '' ;
 744
 745                                 # Loop through each table cell
 746                                 foreach ( $after AS $theline )
 747                                 {
 748                                         $z = '' ;
 749                                         if ( $fc != '+' )
 750                                         {
 751                                                 $tra = array_pop ( $ltr ) ;
 752                                                 if ( !array_pop ( $tr ) ) $z = '<tr'.$tra.">\n" ;
 753                                                 array_push ( $tr , true ) ;
 754                                                 array_push ( $ltr , '' ) ;
 755                                                 array_pop ( $has_opened_tr );
 756                                                 array_push ( $has_opened_tr , true ) ;
 757                                         }
 758
 759                                         $l = array_pop ( $ltd ) ;
 760                                         if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 761                                         if ( $fc == '|' ) $l = 'td' ;
 762                                         else if ( $fc == '!' ) $l = 'th' ;
 763                                         else if ( $fc == '+' ) $l = 'caption' ;
 764                                         else $l = '' ;
 765                                         array_push ( $ltd , $l ) ;
 766
 767                                         # Cell parameters
 768                                         $y = explode ( '|' , $theline , 2 ) ;
 769                                         # Note that a '|' inside an invalid link should not
 770                                         # be mistaken as delimiting cell parameters
 771                                         if ( strpos( $y[0], '[[' ) !== false ) {
 772                                                 $y = array ($theline);
 773                                         }
 774                                         if ( count ( $y ) == 1 )
 775                                                 $y = "{$z}<{$l}>{$y[0]}" ;
 776                                         else {
 777                                                 $attributes = $this->unstripForHTML( $y[0] );
 778                                                 $y = "{$z}<{$l}".Sanitizer::fixTagAttributes($attributes, $l).">{$y[1]}" ;
 779                                         }
 780                                         $t[$k] .= $y ;
 781                                         array_push ( $td , true ) ;
 782                                 }
 783                         }
 784                 }
 785
 786                 # Closing open td, tr && table
 787                 while ( count ( $td ) > 0 )
 788                 {
 789                         $l = array_pop ( $ltd ) ;
 790                         if ( array_pop ( $td ) ) $t[] = '</td>' ;
 791                         if ( array_pop ( $tr ) ) $t[] = '</tr>' ;
 792                         if ( !array_pop ( $has_opened_tr ) ) $t[] = "<tr><td></td></tr>" ;
 793                         $t[] = '</table>' ;
 794                 }
 795
 796                 $t = implode ( "\n" , $t ) ;
 797                 # special case: don't return empty table
 798                 if($t == "<table>\n<tr><td></td></tr>\n</table>")
 799                         $t = '';
 800                 wfProfileOut( $fname );
 801                 return $t ;
 802         }
 803
 804         /**
 805          * Helper function for parse() that transforms wiki markup into
 806          * HTML. Only called for $mOutputType == OT_HTML.
 807          *
 808          * @private
 809          */
 810         function internalParse( $text ) {
 811                 $args = array();
 812                 $isMain = true;
 813                 $fname = 'Parser::internalParse';
 814                 wfProfileIn( $fname );
 815
 816                 # Remove <noinclude> tags and <includeonly> sections
 817                 $text = strtr( $text, array( '<onlyinclude>' => '' , '</onlyinclude>' => '' ) );
 818                 $text = strtr( $text, array( '<noinclude>' => '', '</noinclude>' => '') );
 819                 $text = preg_replace( '/<includeonly>.*?<\/includeonly>/s', '', $text );
 820
 821                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ) );
 822
 823                 $text = $this->replaceVariables( $text, $args );
 824
 825                 // Tables need to come after variable replacement for things to work
 826                 // properly; putting them before other transformations should keep
 827                 // exciting things like link expansions from showing up in surprising
 828                 // places.
 829                 $text = $this->doTableStuff( $text );
 830
 831                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 832
 833                 $text = $this->stripToc( $text );
 834                 $text = $this->doHeadings( $text );
 835                 if($this->mOptions->getUseDynamicDates()) {
 836                         $df =& DateFormatter::getInstance();
 837                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 838                 }
 839                 $text = $this->doAllQuotes( $text );
 840                 $text = $this->replaceInternalLinks( $text );
 841                 $text = $this->replaceExternalLinks( $text );
 842
 843                 # replaceInternalLinks may sometimes leave behind
 844                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 845                 $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
 846
 847                 $text = $this->doMagicLinks( $text );
 848                 $text = $this->formatHeadings( $text, $isMain );
 849
 850                 wfProfileOut( $fname );
 851                 return $text;
 852         }
 853
 854         /**
 855          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 856          * magic external links.
 857          *
 858          * @private
 859          */
 860         function &doMagicLinks( &$text ) {
 861                 $text = $this->magicISBN( $text );
 862                 $text = $this->magicRFC( $text, 'RFC ', 'rfcurl' );
 863                 $text = $this->magicRFC( $text, 'PMID ', 'pubmedurl' );
 864                 return $text;
 865         }
 866
 867         /**
 868          * Parse headers and return html
 869          *
 870          * @private
 871          */
 872         function doHeadings( $text ) {
 873                 $fname = 'Parser::doHeadings';
 874                 wfProfileIn( $fname );
 875                 for ( $i = 6; $i >= 1; --$i ) {
 876                         $h = str_repeat( '=', $i );
 877                         $text = preg_replace( "/^{$h}(.+){$h}(\\s|$)/m",
 878                           "<h{$i}>\\1</h{$i}>\\2", $text );
 879                 }
 880                 wfProfileOut( $fname );
 881                 return $text;
 882         }
 883
 884         /**
 885          * Replace single quotes with HTML markup
 886          * @private
 887          * @return string the altered text
 888          */
 889         function doAllQuotes( $text ) {
 890                 $fname = 'Parser::doAllQuotes';
 891                 wfProfileIn( $fname );
 892                 $outtext = '';
 893                 $lines = explode( "\n", $text );
 894                 foreach ( $lines as $line ) {
 895                         $outtext .= $this->doQuotes ( $line ) . "\n";
 896                 }
 897                 $outtext = substr($outtext, 0,-1);
 898                 wfProfileOut( $fname );
 899                 return $outtext;
 900         }
 901
 902         /**
 903          * Helper function for doAllQuotes()
 904          * @private
 905          */
 906         function doQuotes( $text ) {
 907                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
 908                 if ( count( $arr ) == 1 )
 909                         return $text;
 910                 else
 911                 {
 912                         # First, do some preliminary work. This may shift some apostrophes from
 913                         # being mark-up to being text. It also counts the number of occurrences
 914                         # of bold and italics mark-ups.
 915                         $i = 0;
 916                         $numbold = 0;
 917                         $numitalics = 0;
 918                         foreach ( $arr as $r )
 919                         {
 920                                 if ( ( $i % 2 ) == 1 )
 921                                 {
 922                                         # If there are ever four apostrophes, assume the first is supposed to
 923                                         # be text, and the remaining three constitute mark-up for bold text.
 924                                         if ( strlen( $arr[$i] ) == 4 )
 925                                         {
 926                                                 $arr[$i-1] .= "'";
 927                                                 $arr[$i] = "'''";
 928                                         }
 929                                         # If there are more than 5 apostrophes in a row, assume they're all
 930                                         # text except for the last 5.
 931                                         else if ( strlen( $arr[$i] ) > 5 )
 932                                         {
 933                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
 934                                                 $arr[$i] = "'''''";
 935                                         }
 936                                         # Count the number of occurrences of bold and italics mark-ups.
 937                                         # We are not counting sequences of five apostrophes.
 938                                         if ( strlen( $arr[$i] ) == 2 ) $numitalics++;  else
 939                                         if ( strlen( $arr[$i] ) == 3 ) $numbold++;     else
 940                                         if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
 941                                 }
 942                                 $i++;
 943                         }
 944
 945                         # If there is an odd number of both bold and italics, it is likely
 946                         # that one of the bold ones was meant to be an apostrophe followed
 947                         # by italics. Which one we cannot know for certain, but it is more
 948                         # likely to be one that has a single-letter word before it.
 949                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
 950                         {
 951                                 $i = 0;
 952                                 $firstsingleletterword = -1;
 953                                 $firstmultiletterword = -1;
 954                                 $firstspace = -1;
 955                                 foreach ( $arr as $r )
 956                                 {
 957                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
 958                                         {
 959                                                 $x1 = substr ($arr[$i-1], -1);
 960                                                 $x2 = substr ($arr[$i-1], -2, 1);
 961                                                 if ($x1 == ' ') {
 962                                                         if ($firstspace == -1) $firstspace = $i;
 963                                                 } else if ($x2 == ' ') {
 964                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
 965                                                 } else {
 966                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
 967                                                 }
 968                                         }
 969                                         $i++;
 970                                 }
 971
 972                                 # If there is a single-letter word, use it!
 973                                 if ($firstsingleletterword > -1)
 974                                 {
 975                                         $arr [ $firstsingleletterword ] = "''";
 976                                         $arr [ $firstsingleletterword-1 ] .= "'";
 977                                 }
 978                                 # If not, but there's a multi-letter word, use that one.
 979                                 else if ($firstmultiletterword > -1)
 980                                 {
 981                                         $arr [ $firstmultiletterword ] = "''";
 982                                         $arr [ $firstmultiletterword-1 ] .= "'";
 983                                 }
 984                                 # ... otherwise use the first one that has neither.
 985                                 # (notice that it is possible for all three to be -1 if, for example,
 986                                 # there is only one pentuple-apostrophe in the line)
 987                                 else if ($firstspace > -1)
 988                                 {
 989                                         $arr [ $firstspace ] = "''";
 990                                         $arr [ $firstspace-1 ] .= "'";
 991                                 }
 992                         }
 993
 994                         # Now let's actually convert our apostrophic mush to HTML!
 995                         $output = '';
 996                         $buffer = '';
 997                         $state = '';
 998                         $i = 0;
 999                         foreach ($arr as $r)
1000                         {
1001                                 if (($i % 2) == 0)
1002                                 {
1003                                         if ($state == 'both')
1004                                                 $buffer .= $r;
1005                                         else
1006                                                 $output .= $r;
1007                                 }
1008                                 else
1009                                 {
1010                                         if (strlen ($r) == 2)
1011                                         {
1012                                                 if ($state == 'i')
1013                                                 { $output .= '</i>'; $state = ''; }
1014                                                 else if ($state == 'bi')
1015                                                 { $output .= '</i>'; $state = 'b'; }
1016                                                 else if ($state == 'ib')
1017                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1018                                                 else if ($state == 'both')
1019                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1020                                                 else # $state can be 'b' or ''
1021                                                 { $output .= '<i>'; $state .= 'i'; }
1022                                         }
1023                                         else if (strlen ($r) == 3)
1024                                         {
1025                                                 if ($state == 'b')
1026                                                 { $output .= '</b>'; $state = ''; }
1027                                                 else if ($state == 'bi')
1028                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1029                                                 else if ($state == 'ib')
1030                                                 { $output .= '</b>'; $state = 'i'; }
1031                                                 else if ($state == 'both')
1032                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1033                                                 else # $state can be 'i' or ''
1034                                                 { $output .= '<b>'; $state .= 'b'; }
1035                                         }
1036                                         else if (strlen ($r) == 5)
1037                                         {
1038                                                 if ($state == 'b')
1039                                                 { $output .= '</b><i>'; $state = 'i'; }
1040                                                 else if ($state == 'i')
1041                                                 { $output .= '</i><b>'; $state = 'b'; }
1042                                                 else if ($state == 'bi')
1043                                                 { $output .= '</i></b>'; $state = ''; }
1044                                                 else if ($state == 'ib')
1045                                                 { $output .= '</b></i>'; $state = ''; }
1046                                                 else if ($state == 'both')
1047                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1048                                                 else # ($state == '')
1049                                                 { $buffer = ''; $state = 'both'; }
1050                                         }
1051                                 }
1052                                 $i++;
1053                         }
1054                         # Now close all remaining tags.  Notice that the order is important.
1055                         if ($state == 'b' || $state == 'ib')
1056                                 $output .= '</b>';
1057                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
1058                                 $output .= '</i>';
1059                         if ($state == 'bi')
1060                                 $output .= '</b>';
1061                         if ($state == 'both')
1062                                 $output .= '<b><i>'.$buffer.'</i></b>';
1063                         return $output;
1064                 }
1065         }
1066
1067         /**
1068          * Replace external links
1069          *
1070          * Note: this is all very hackish and the order of execution matters a lot.
1071          * Make sure to run maintenance/parserTests.php if you change this code.
1072          *
1073          * @private
1074          */
1075         function replaceExternalLinks( $text ) {
1076                 global $wgContLang;
1077                 $fname = 'Parser::replaceExternalLinks';
1078                 wfProfileIn( $fname );
1079
1080                 $sk =& $this->mOptions->getSkin();
1081
1082                 $bits = preg_split( EXT_LINK_BRACKETED, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1083
1084                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
1085
1086                 $i = 0;
1087                 while ( $i<count( $bits ) ) {
1088                         $url = $bits[$i++];
1089                         $protocol = $bits[$i++];
1090                         $text = $bits[$i++];
1091                         $trail = $bits[$i++];
1092
1093                         # The characters '<' and '>' (which were escaped by
1094                         # removeHTMLtags()) should not be included in
1095                         # URLs, per RFC 2396.
1096                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1097                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1098                                 $url = substr($url, 0, $m2[0][1]);
1099                         }
1100
1101                         # If the link text is an image URL, replace it with an <img> tag
1102                         # This happened by accident in the original parser, but some people used it extensively
1103                         $img = $this->maybeMakeExternalImage( $text );
1104                         if ( $img !== false ) {
1105                                 $text = $img;
1106                         }
1107
1108                         $dtrail = '';
1109
1110                         # Set linktype for CSS - if URL==text, link is essentially free
1111                         $linktype = ($text == $url) ? 'free' : 'text';
1112
1113                         # No link text, e.g. [http://domain.tld/some.link]
1114                         if ( $text == '' ) {
1115                                 # Autonumber if allowed. See bug #5918
1116                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1117                                         $text = '[' . ++$this->mAutonumber . ']';
1118                                         $linktype = 'autonumber';
1119                                 } else {
1120                                         # Otherwise just use the URL
1121                                         $text = htmlspecialchars( $url );
1122                                         $linktype = 'free';
1123                                 }
1124                         } else {
1125                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1126                                 # Check for trail
1127                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1128                         }
1129
1130                         $text = $wgContLang->markNoConversion($text);
1131
1132                         # Normalize any HTML entities in input. They will be
1133                         # re-escaped by makeExternalLink().
1134                         $url = Sanitizer::decodeCharReferences( $url );
1135
1136                         # Process the trail (i.e. everything after this link up until start of the next link),
1137                         # replacing any non-bracketed links
1138                         $trail = $this->replaceFreeExternalLinks( $trail );
1139
1140                         # Use the encoded URL
1141                         # This means that users can paste URLs directly into the text
1142                         # Funny characters like &ouml; aren't valid in URLs anyway
1143                         # This was changed in August 2004
1144                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype, $this->mTitle->getNamespace() ) . $dtrail . $trail;
1145
1146                         # Register link in the output object.
1147                         # Replace unnecessary URL escape codes with the referenced character
1148                         # This prevents spammers from hiding links from the filters
1149                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1150                         $this->mOutput->addExternalLink( $pasteurized );
1151                 }
1152
1153                 wfProfileOut( $fname );
1154                 return $s;
1155         }
1156
1157         /**
1158          * Replace anything that looks like a URL with a link
1159          * @private
1160          */
1161         function replaceFreeExternalLinks( $text ) {
1162                 global $wgContLang;
1163                 $fname = 'Parser::replaceFreeExternalLinks';
1164                 wfProfileIn( $fname );
1165
1166                 $bits = preg_split( '/(\b(?:' . wfUrlProtocols() . '))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1167                 $s = array_shift( $bits );
1168                 $i = 0;
1169
1170                 $sk =& $this->mOptions->getSkin();
1171
1172                 while ( $i < count( $bits ) ){
1173                         $protocol = $bits[$i++];
1174                         $remainder = $bits[$i++];
1175
1176                         if ( preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1177                                 # Found some characters after the protocol that look promising
1178                                 $url = $protocol . $m[1];
1179                                 $trail = $m[2];
1180
1181                                 # special case: handle urls as url args:
1182                                 # http://www.example.com/foo?=http://www.example.com/bar
1183                                 if(strlen($trail) == 0 &&
1184                                         isset($bits[$i]) &&
1185                                         preg_match('/^'. wfUrlProtocols() . '$/S', $bits[$i]) &&
1186                                         preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $bits[$i + 1], $m ))
1187                                 {
1188                                         # add protocol, arg
1189                                         $url .= $bits[$i] . $m[1]; # protocol, url as arg to previous link
1190                                         $i += 2;
1191                                         $trail = $m[2];
1192                                 }
1193
1194                                 # The characters '<' and '>' (which were escaped by
1195                                 # removeHTMLtags()) should not be included in
1196                                 # URLs, per RFC 2396.
1197                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1198                                         $trail = substr($url, $m2[0][1]) . $trail;
1199                                         $url = substr($url, 0, $m2[0][1]);
1200                                 }
1201
1202                                 # Move trailing punctuation to $trail
1203                                 $sep = ',;\.:!?';
1204                                 # If there is no left bracket, then consider right brackets fair game too
1205                                 if ( strpos( $url, '(' ) === false ) {
1206                                         $sep .= ')';
1207                                 }
1208
1209                                 $numSepChars = strspn( strrev( $url ), $sep );
1210                                 if ( $numSepChars ) {
1211                                         $trail = substr( $url, -$numSepChars ) . $trail;
1212                                         $url = substr( $url, 0, -$numSepChars );
1213                                 }
1214
1215                                 # Normalize any HTML entities in input. They will be
1216                                 # re-escaped by makeExternalLink() or maybeMakeExternalImage()
1217                                 $url = Sanitizer::decodeCharReferences( $url );
1218
1219                                 # Is this an external image?
1220                                 $text = $this->maybeMakeExternalImage( $url );
1221                                 if ( $text === false ) {
1222                                         # Not an image, make a link
1223                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free', $this->mTitle->getNamespace() );
1224                                         # Register it in the output object...
1225                                         # Replace unnecessary URL escape codes with their equivalent characters
1226                                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1227                                         $this->mOutput->addExternalLink( $pasteurized );
1228                                 }
1229                                 $s .= $text . $trail;
1230                         } else {
1231                                 $s .= $protocol . $remainder;
1232                         }
1233                 }
1234                 wfProfileOut( $fname );
1235                 return $s;
1236         }
1237
1238         /**
1239          * Replace unusual URL escape codes with their equivalent characters
1240          * @param string
1241          * @return string
1242          * @static
1243          * @fixme This can merge genuinely required bits in the path or query string,
1244          *        breaking legit URLs. A proper fix would treat the various parts of
1245          *        the URL differently; as a workaround, just use the output for
1246          *        statistical records, not for actual linking/output.
1247          */
1248         function replaceUnusualEscapes( $url ) {
1249                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1250                         array( 'Parser', 'replaceUnusualEscapesCallback' ), $url );
1251         }
1252
1253         /**
1254          * Callback function used in replaceUnusualEscapes().
1255          * Replaces unusual URL escape codes with their equivalent character
1256          * @static
1257          * @private
1258          */
1259         function replaceUnusualEscapesCallback( $matches ) {
1260                 $char = urldecode( $matches[0] );
1261                 $ord = ord( $char );
1262                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1263                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1264                         // No, shouldn't be escaped
1265                         return $char;
1266                 } else {
1267                         // Yes, leave it escaped
1268                         return $matches[0];
1269                 }
1270         }
1271
1272         /**
1273          * make an image if it's allowed, either through the global
1274          * option or through the exception
1275          * @private
1276          */
1277         function maybeMakeExternalImage( $url ) {
1278                 $sk =& $this->mOptions->getSkin();
1279                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1280                 $imagesexception = !empty($imagesfrom);
1281                 $text = false;
1282                 if ( $this->mOptions->getAllowExternalImages()
1283                      || ( $imagesexception && strpos( $url, $imagesfrom ) === 0 ) ) {
1284                         if ( preg_match( EXT_IMAGE_REGEX, $url ) ) {
1285                                 # Image found
1286                                 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
1287                         }
1288                 }
1289                 return $text;
1290         }
1291
1292         /**
1293          * Process [[ ]] wikilinks
1294          *
1295          * @private
1296          */
1297         function replaceInternalLinks( $s ) {
1298                 global $wgContLang;
1299                 static $fname = 'Parser::replaceInternalLinks' ;
1300
1301                 wfProfileIn( $fname );
1302
1303                 wfProfileIn( $fname.'-setup' );
1304                 static $tc = FALSE;
1305                 # the % is needed to support urlencoded titles as well
1306                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1307
1308                 $sk =& $this->mOptions->getSkin();
1309
1310                 #split the entire text string on occurences of [[
1311                 $a = explode( '[[', ' ' . $s );
1312                 #get the first element (all text up to first [[), and remove the space we added
1313                 $s = array_shift( $a );
1314                 $s = substr( $s, 1 );
1315
1316                 # Match a link having the form [[namespace:link|alternate]]trail
1317                 static $e1 = FALSE;
1318                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1319                 # Match cases where there is no "]]", which might still be images
1320                 static $e1_img = FALSE;
1321                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1322                 # Match the end of a line for a word that's not followed by whitespace,
1323                 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1324                 $e2 = wfMsgForContent( 'linkprefix' );
1325
1326                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1327
1328                 if( is_null( $this->mTitle ) ) {
1329                         wfDebugDieBacktrace( 'nooo' );
1330                 }
1331                 $nottalk = !$this->mTitle->isTalkPage();
1332
1333                 if ( $useLinkPrefixExtension ) {
1334                         if ( preg_match( $e2, $s, $m ) ) {
1335                                 $first_prefix = $m[2];
1336                         } else {
1337                                 $first_prefix = false;
1338                         }
1339                 } else {
1340                         $prefix = '';
1341                 }
1342
1343                 $selflink = $this->mTitle->getPrefixedText();
1344                 wfProfileOut( $fname.'-setup' );
1345
1346                 $checkVariantLink = sizeof($wgContLang->getVariants())>1;
1347                 $useSubpages = $this->areSubpagesAllowed();
1348
1349                 # Loop for each link
1350                 for ($k = 0; isset( $a[$k] ); $k++) {
1351                         $line = $a[$k];
1352                         if ( $useLinkPrefixExtension ) {
1353                                 wfProfileIn( $fname.'-prefixhandling' );
1354                                 if ( preg_match( $e2, $s, $m ) ) {
1355                                         $prefix = $m[2];
1356                                         $s = $m[1];
1357                                 } else {
1358                                         $prefix='';
1359                                 }
1360                                 # first link
1361                                 if($first_prefix) {
1362                                         $prefix = $first_prefix;
1363                                         $first_prefix = false;
1364                                 }
1365                                 wfProfileOut( $fname.'-prefixhandling' );
1366                         }
1367
1368                         $might_be_img = false;
1369
1370                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1371                                 $text = $m[2];
1372                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1373                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1374                                 # the real problem is with the $e1 regex
1375                                 # See bug 1300.
1376                                 #
1377                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1378                                 # and no image is in sight. See bug 2095.
1379                                 #
1380                                 if( $text !== '' &&
1381                                         preg_match( "/^\](.*)/s", $m[3], $n ) &&
1382                                         strpos($text, '[') !== false
1383                                 )
1384                                 {
1385                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1386                                         $m[3] = $n[1];
1387                                 }
1388                                 # fix up urlencoded title texts
1389                                 if(preg_match('/%/', $m[1] ))
1390                                         # Should anchors '#' also be rejected?
1391                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1392                                 $trail = $m[3];
1393                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1394                                 $might_be_img = true;
1395                                 $text = $m[2];
1396                                 if(preg_match('/%/', $m[1] )) $m[1] = urldecode($m[1]);
1397                                 $trail = "";
1398                         } else { # Invalid form; output directly
1399                                 $s .= $prefix . '[[' . $line ;
1400                                 continue;
1401                         }
1402
1403                         # Don't allow internal links to pages containing
1404                         # PROTO: where PROTO is a valid URL protocol; these
1405                         # should be external links.
1406                         if (preg_match('/^(\b(?:' . wfUrlProtocols() . '))/', $m[1])) {
1407                                 $s .= $prefix . '[[' . $line ;
1408                                 continue;
1409                         }
1410
1411                         # Make subpage if necessary
1412                         if( $useSubpages ) {
1413                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1414                         } else {
1415                                 $link = $m[1];
1416                         }
1417
1418                         $noforce = (substr($m[1], 0, 1) != ':');
1419                         if (!$noforce) {
1420                                 # Strip off leading ':'
1421                                 $link = substr($link, 1);
1422                         }
1423
1424                         $nt = Title::newFromText( $this->unstripNoWiki($link, $this->mStripState) );
1425                         if( !$nt ) {
1426                                 $s .= $prefix . '[[' . $line;
1427                                 continue;
1428                         }
1429
1430                         #check other language variants of the link
1431                         #if the article does not exist
1432                         if( $checkVariantLink
1433                             && $nt->getArticleID() == 0 ) {
1434                                 $wgContLang->findVariantLink($link, $nt);
1435                         }
1436
1437                         $ns = $nt->getNamespace();
1438                         $iw = $nt->getInterWiki();
1439
1440                         if ($might_be_img) { # if this is actually an invalid link
1441                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1442                                         $found = false;
1443                                         while (isset ($a[$k+1]) ) {
1444                                                 #look at the next 'line' to see if we can close it there
1445                                                 $spliced = array_splice( $a, $k + 1, 1 );
1446                                                 $next_line = array_shift( $spliced );
1447                                                 if( preg_match("/^(.*?]].*?)]](.*)$/sD", $next_line, $m) ) {
1448                                                 # the first ]] closes the inner link, the second the image
1449                                                         $found = true;
1450                                                         $text .= '[[' . $m[1];
1451                                                         $trail = $m[2];
1452                                                         break;
1453                                                 } elseif( preg_match("/^.*?]].*$/sD", $next_line, $m) ) {
1454                                                         #if there's exactly one ]] that's fine, we'll keep looking
1455                                                         $text .= '[[' . $m[0];
1456                                                 } else {
1457                                                         #if $next_line is invalid too, we need look no further
1458                                                         $text .= '[[' . $next_line;
1459                                                         break;
1460                                                 }
1461                                         }
1462                                         if ( !$found ) {
1463                                                 # we couldn't find the end of this imageLink, so output it raw
1464                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1465                                                 $text = $this->replaceInternalLinks($text);
1466                                                 $s .= $prefix . '[[' . $link . '|' . $text;
1467                                                 # note: no $trail, because without an end, there *is* no trail
1468                                                 continue;
1469                                         }
1470                                 } else { #it's not an image, so output it raw
1471                                         $s .= $prefix . '[[' . $link . '|' . $text;
1472                                         # note: no $trail, because without an end, there *is* no trail
1473                                         continue;
1474                                 }
1475                         }
1476
1477                         $wasblank = ( '' == $text );
1478                         if( $wasblank ) $text = $link;
1479
1480
1481                         # Link not escaped by : , create the various objects
1482                         if( $noforce ) {
1483
1484                                 # Interwikis
1485                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1486                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1487                                         $s = rtrim($s . "\n");
1488                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1489                                         continue;
1490                                 }
1491
1492                                 if ( $ns == NS_IMAGE ) {
1493                                         wfProfileIn( "$fname-image" );
1494                                         if ( !wfIsBadImage( $nt->getDBkey() ) ) {
1495                                                 # recursively parse links inside the image caption
1496                                                 # actually, this will parse them in any other parameters, too,
1497                                                 # but it might be hard to fix that, and it doesn't matter ATM
1498                                                 $text = $this->replaceExternalLinks($text);
1499                                                 $text = $this->replaceInternalLinks($text);
1500
1501                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1502                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text ) ) . $trail;
1503                                                 $this->mOutput->addImage( $nt->getDBkey() );
1504
1505                                                 wfProfileOut( "$fname-image" );
1506                                                 continue;
1507                                         }
1508                                         wfProfileOut( "$fname-image" );
1509
1510                                 }
1511
1512                                 if ( $ns == NS_CATEGORY ) {
1513                                         wfProfileIn( "$fname-category" );
1514                                         $s = rtrim($s . "\n"); # bug 87
1515
1516                                         if ( $wasblank ) {
1517                                                 if ( $this->mTitle->getNamespace() == NS_CATEGORY ) {
1518                                                         $sortkey = $this->mTitle->getText();
1519                                                 } else {
1520                                                         $sortkey = $this->mTitle->getPrefixedText();
1521                                                 }
1522                                         } else {
1523                                                 $sortkey = $text;
1524                                         }
1525                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1526                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1527                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1528
1529                                         /**
1530                                          * Strip the whitespace Category links produce, see bug 87
1531                                          * @todo We might want to use trim($tmp, "\n") here.
1532                                          */
1533                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1534
1535                                         wfProfileOut( "$fname-category" );
1536                                         continue;
1537                                 }
1538                         }
1539
1540                         if( ( $nt->getPrefixedText() === $selflink ) &&
1541                             ( $nt->getFragment() === '' ) ) {
1542                                 # Self-links are handled specially; generally de-link and change to bold.
1543                                 $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1544                                 continue;
1545                         }
1546
1547                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1548                         if( $ns == NS_MEDIA ) {
1549                                 $link = $sk->makeMediaLinkObj( $nt, $text );
1550                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1551                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1552                                 $this->mOutput->addImage( $nt->getDBkey() );
1553                                 continue;
1554                         } elseif( $ns == NS_SPECIAL ) {
1555                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1556                                 continue;
1557                         } elseif( $ns == NS_IMAGE ) {
1558                                 $img = Image::newFromTitle( $nt );
1559                                 if( $img->exists() ) {
1560                                         // Force a blue link if the file exists; may be a remote
1561                                         // upload on the shared repository, and we want to see its
1562                                         // auto-generated page.
1563                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1564                                         continue;
1565                                 }
1566                         }
1567                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1568                 }
1569                 wfProfileOut( $fname );
1570                 return $s;
1571         }
1572
1573         /**
1574          * Make a link placeholder. The text returned can be later resolved to a real link with
1575          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1576          * parsing of interwiki links, and secondly to allow all extistence checks and
1577          * article length checks (for stub links) to be bundled into a single query.
1578          *
1579          */
1580         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1581                 if ( ! is_object($nt) ) {
1582                         # Fail gracefully
1583                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1584                 } else {
1585                         # Separate the link trail from the rest of the link
1586                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1587
1588                         if ( $nt->isExternal() ) {
1589                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1590                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1591                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1592                         } else {
1593                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1594                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1595                                 $this->mLinkHolders['queries'][] = $query;
1596                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1597                                 $this->mLinkHolders['titles'][] = $nt;
1598
1599                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1600                         }
1601                 }
1602                 return $retVal;
1603         }
1604
1605         /**
1606          * Render a forced-blue link inline; protect against double expansion of
1607          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1608          * Since this little disaster has to split off the trail text to avoid
1609          * breaking URLs in the following text without breaking trails on the
1610          * wiki links, it's been made into a horrible function.
1611          *
1612          * @param Title $nt
1613          * @param string $text
1614          * @param string $query
1615          * @param string $trail
1616          * @param string $prefix
1617          * @return string HTML-wikitext mix oh yuck
1618          */
1619         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1620                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1621                 $sk =& $this->mOptions->getSkin();
1622                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1623                 return $this->armorLinks( $link ) . $trail;
1624         }
1625
1626         /**
1627          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1628          * going to go through further parsing steps before inline URL expansion.
1629          *
1630          * In particular this is important when using action=render, which causes
1631          * full URLs to be included.
1632          *
1633          * Oh man I hate our multi-layer parser!
1634          *
1635          * @param string more-or-less HTML
1636          * @return string less-or-more HTML with NOPARSE bits
1637          */
1638         function armorLinks( $text ) {
1639                 return preg_replace( "/\b(" . wfUrlProtocols() . ')/',
1640                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1641         }
1642
1643         /**
1644          * Return true if subpage links should be expanded on this page.
1645          * @return bool
1646          */
1647         function areSubpagesAllowed() {
1648                 # Some namespaces don't allow subpages
1649                 global $wgNamespacesWithSubpages;
1650                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1651         }
1652
1653         /**
1654          * Handle link to subpage if necessary
1655          * @param string $target the source of the link
1656          * @param string &$text the link text, modified as necessary
1657          * @return string the full name of the link
1658          * @private
1659          */
1660         function maybeDoSubpageLink($target, &$text) {
1661                 # Valid link forms:
1662                 # Foobar -- normal
1663                 # :Foobar -- override special treatment of prefix (images, language links)
1664                 # /Foobar -- convert to CurrentPage/Foobar
1665                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1666                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1667                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1668
1669                 $fname = 'Parser::maybeDoSubpageLink';
1670                 wfProfileIn( $fname );
1671                 $ret = $target; # default return value is no change
1672
1673                 # Some namespaces don't allow subpages,
1674                 # so only perform processing if subpages are allowed
1675                 if( $this->areSubpagesAllowed() ) {
1676                         # Look at the first character
1677                         if( $target != '' && $target{0} == '/' ) {
1678                                 # / at end means we don't want the slash to be shown
1679                                 if( substr( $target, -1, 1 ) == '/' ) {
1680                                         $target = substr( $target, 1, -1 );
1681                                         $noslash = $target;
1682                                 } else {
1683                                         $noslash = substr( $target, 1 );
1684                                 }
1685
1686                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash);
1687                                 if( '' === $text ) {
1688                                         $text = $target;
1689                                 } # this might be changed for ugliness reasons
1690                         } else {
1691                                 # check for .. subpage backlinks
1692                                 $dotdotcount = 0;
1693                                 $nodotdot = $target;
1694                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1695                                         ++$dotdotcount;
1696                                         $nodotdot = substr( $nodotdot, 3 );
1697                                 }
1698                                 if($dotdotcount > 0) {
1699                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1700                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1701                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1702                                                 # / at the end means don't show full path
1703                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1704                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1705                                                         if( '' === $text ) {
1706                                                                 $text = $nodotdot;
1707                                                         }
1708                                                 }
1709                                                 $nodotdot = trim( $nodotdot );
1710                                                 if( $nodotdot != '' ) {
1711                                                         $ret .= '/' . $nodotdot;
1712                                                 }
1713                                         }
1714                                 }
1715                         }
1716                 }
1717
1718                 wfProfileOut( $fname );
1719                 return $ret;
1720         }
1721
1722         /**#@+
1723          * Used by doBlockLevels()
1724          * @private
1725          */
1726         /* private */ function closeParagraph() {
1727                 $result = '';
1728                 if ( '' != $this->mLastSection ) {
1729                         $result = '</' . $this->mLastSection  . ">\n";
1730                 }
1731                 $this->mInPre = false;
1732                 $this->mLastSection = '';
1733                 return $result;
1734         }
1735         # getCommon() returns the length of the longest common substring
1736         # of both arguments, starting at the beginning of both.
1737         #
1738         /* private */ function getCommon( $st1, $st2 ) {
1739                 $fl = strlen( $st1 );
1740                 $shorter = strlen( $st2 );
1741                 if ( $fl < $shorter ) { $shorter = $fl; }
1742
1743                 for ( $i = 0; $i < $shorter; ++$i ) {
1744                         if ( $st1{$i} != $st2{$i} ) { break; }
1745                 }
1746                 return $i;
1747         }
1748         # These next three functions open, continue, and close the list
1749         # element appropriate to the prefix character passed into them.
1750         #
1751         /* private */ function openList( $char ) {
1752                 $result = $this->closeParagraph();
1753
1754                 if ( '*' == $char ) { $result .= '<ul><li>'; }
1755                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
1756                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
1757                 else if ( ';' == $char ) {
1758                         $result .= '<dl><dt>';
1759                         $this->mDTopen = true;
1760                 }
1761                 else { $result = '<!-- ERR 1 -->'; }
1762
1763                 return $result;
1764         }
1765
1766         /* private */ function nextItem( $char ) {
1767                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
1768                 else if ( ':' == $char || ';' == $char ) {
1769                         $close = '</dd>';
1770                         if ( $this->mDTopen ) { $close = '</dt>'; }
1771                         if ( ';' == $char ) {
1772                                 $this->mDTopen = true;
1773                                 return $close . '<dt>';
1774                         } else {
1775                                 $this->mDTopen = false;
1776                                 return $close . '<dd>';
1777                         }
1778                 }
1779                 return '<!-- ERR 2 -->';
1780         }
1781
1782         /* private */ function closeList( $char ) {
1783                 if ( '*' == $char ) { $text = '</li></ul>'; }
1784                 else if ( '#' == $char ) { $text = '</li></ol>'; }
1785                 else if ( ':' == $char ) {
1786                         if ( $this->mDTopen ) {
1787                                 $this->mDTopen = false;
1788                                 $text = '</dt></dl>';
1789                         } else {
1790                                 $text = '</dd></dl>';
1791                         }
1792                 }
1793                 else {  return '<!-- ERR 3 -->'; }
1794                 return $text."\n";
1795         }
1796         /**#@-*/
1797
1798         /**
1799          * Make lists from lines starting with ':', '*', '#', etc.
1800          *
1801          * @private
1802          * @return string the lists rendered as HTML
1803          */
1804         function doBlockLevels( $text, $linestart ) {
1805                 $fname = 'Parser::doBlockLevels';
1806                 wfProfileIn( $fname );
1807
1808                 # Parsing through the text line by line.  The main thing
1809                 # happening here is handling of block-level elements p, pre,
1810                 # and making lists from lines starting with * # : etc.
1811                 #
1812                 $textLines = explode( "\n", $text );
1813
1814                 $lastPrefix = $output = '';
1815                 $this->mDTopen = $inBlockElem = false;
1816                 $prefixLength = 0;
1817                 $paragraphStack = false;
1818
1819                 if ( !$linestart ) {
1820                         $output .= array_shift( $textLines );
1821                 }
1822                 foreach ( $textLines as $oLine ) {
1823                         $lastPrefixLength = strlen( $lastPrefix );
1824                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
1825                         $preOpenMatch = preg_match('/<pre/i', $oLine );
1826                         if ( !$this->mInPre ) {
1827                                 # Multiple prefixes may abut each other for nested lists.
1828                                 $prefixLength = strspn( $oLine, '*#:;' );
1829                                 $pref = substr( $oLine, 0, $prefixLength );
1830
1831                                 # eh?
1832                                 $pref2 = str_replace( ';', ':', $pref );
1833                                 $t = substr( $oLine, $prefixLength );
1834                                 $this->mInPre = !empty($preOpenMatch);
1835                         } else {
1836                                 # Don't interpret any other prefixes in preformatted text
1837                                 $prefixLength = 0;
1838                                 $pref = $pref2 = '';
1839                                 $t = $oLine;
1840                         }
1841
1842                         # List generation
1843                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
1844                                 # Same as the last item, so no need to deal with nesting or opening stuff
1845                                 $output .= $this->nextItem( substr( $pref, -1 ) );
1846                                 $paragraphStack = false;
1847
1848                                 if ( substr( $pref, -1 ) == ';') {
1849                                         # The one nasty exception: definition lists work like this:
1850                                         # ; title : definition text
1851                                         # So we check for : in the remainder text to split up the
1852                                         # title and definition, without b0rking links.
1853                                         $term = $t2 = '';
1854                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
1855                                                 $t = $t2;
1856                                                 $output .= $term . $this->nextItem( ':' );
1857                                         }
1858                                 }
1859                         } elseif( $prefixLength || $lastPrefixLength ) {
1860                                 # Either open or close a level...
1861                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
1862                                 $paragraphStack = false;
1863
1864                                 while( $commonPrefixLength < $lastPrefixLength ) {
1865                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
1866                                         --$lastPrefixLength;
1867                                 }
1868                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
1869                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
1870                                 }
1871                                 while ( $prefixLength > $commonPrefixLength ) {
1872                                         $char = substr( $pref, $commonPrefixLength, 1 );
1873                                         $output .= $this->openList( $char );
1874
1875                                         if ( ';' == $char ) {
1876                                                 # FIXME: This is dupe of code above
1877                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
1878                                                         $t = $t2;
1879                                                         $output .= $term . $this->nextItem( ':' );
1880                                                 }
1881                                         }
1882                                         ++$commonPrefixLength;
1883                                 }
1884                                 $lastPrefix = $pref2;
1885                         }
1886                         if( 0 == $prefixLength ) {
1887                                 wfProfileIn( "$fname-paragraph" );
1888                                 # No prefix (not in list)--go to paragraph mode
1889                                 // XXX: use a stack for nestable elements like span, table and div
1890                                 $openmatch = preg_match('/(<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
1891                                 $closematch = preg_match(
1892                                         '/(<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
1893                                         '<td|<th|<div|<\\/div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul)/iS', $t );
1894                                 if ( $openmatch or $closematch ) {
1895                                         $paragraphStack = false;
1896                                         # TODO bug 5718: paragraph closed
1897                                         $output .= $this->closeParagraph();
1898                                         if ( $preOpenMatch and !$preCloseMatch ) {
1899                                                 $this->mInPre = true;
1900                                         }
1901                                         if ( $closematch ) {
1902                                                 $inBlockElem = false;
1903                                         } else {
1904                                                 $inBlockElem = true;
1905                                         }
1906                                 } else if ( !$inBlockElem && !$this->mInPre ) {
1907                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
1908                                                 // pre
1909                                                 if ($this->mLastSection != 'pre') {
1910                                                         $paragraphStack = false;
1911                                                         $output .= $this->closeParagraph().'<pre>';
1912                                                         $this->mLastSection = 'pre';
1913                                                 }
1914                                                 $t = substr( $t, 1 );
1915                                         } else {
1916                                                 // paragraph
1917                                                 if ( '' == trim($t) ) {
1918                                                         if ( $paragraphStack ) {
1919                                                                 $output .= $paragraphStack.'<br />';
1920                                                                 $paragraphStack = false;
1921                                                                 $this->mLastSection = 'p';
1922                                                         } else {
1923                                                                 if ($this->mLastSection != 'p' ) {
1924                                                                         $output .= $this->closeParagraph();
1925                                                                         $this->mLastSection = '';
1926                                                                         $paragraphStack = '<p>';
1927                                                                 } else {
1928                                                                         $paragraphStack = '</p><p>';
1929                                                                 }
1930                                                         }
1931                                                 } else {
1932                                                         if ( $paragraphStack ) {
1933                                                                 $output .= $paragraphStack;
1934                                                                 $paragraphStack = false;
1935                                                                 $this->mLastSection = 'p';
1936                                                         } else if ($this->mLastSection != 'p') {
1937                                                                 $output .= $this->closeParagraph().'<p>';
1938                                                                 $this->mLastSection = 'p';
1939                                                         }
1940                                                 }
1941                                         }
1942                                 }
1943                                 wfProfileOut( "$fname-paragraph" );
1944                         }
1945                         // somewhere above we forget to get out of pre block (bug 785)
1946                         if($preCloseMatch && $this->mInPre) {
1947                                 $this->mInPre = false;
1948                         }
1949                         if ($paragraphStack === false) {
1950                                 $output .= $t."\n";
1951                         }
1952                 }
1953                 while ( $prefixLength ) {
1954                         $output .= $this->closeList( $pref2{$prefixLength-1} );
1955                         --$prefixLength;
1956                 }
1957                 if ( '' != $this->mLastSection ) {
1958                         $output .= '</' . $this->mLastSection . '>';
1959                         $this->mLastSection = '';
1960                 }
1961
1962                 wfProfileOut( $fname );
1963                 return $output;
1964         }
1965
1966         /**
1967          * Split up a string on ':', ignoring any occurences inside
1968          * <a>..</a> or <span>...</span>
1969          * @param string $str the string to split
1970          * @param string &$before set to everything before the ':'
1971          * @param string &$after set to everything after the ':'
1972          * return string the position of the ':', or false if none found
1973          */
1974         function findColonNoLinks($str, &$before, &$after) {
1975                 # I wonder if we should make this count all tags, not just <a>
1976                 # and <span>. That would prevent us from matching a ':' that
1977                 # comes in the middle of italics other such formatting....
1978                 # -- Wil
1979                 $fname = 'Parser::findColonNoLinks';
1980                 wfProfileIn( $fname );
1981                 $pos = 0;
1982                 do {
1983                         $colon = strpos($str, ':', $pos);
1984
1985                         if ($colon !== false) {
1986                                 $before = substr($str, 0, $colon);
1987                                 $after = substr($str, $colon + 1);
1988
1989                                 # Skip any ':' within <a> or <span> pairs
1990                                 $a = substr_count($before, '<a');
1991                                 $s = substr_count($before, '<span');
1992                                 $ca = substr_count($before, '</a>');
1993                                 $cs = substr_count($before, '</span>');
1994
1995                                 if ($a <= $ca and $s <= $cs) {
1996                                         # Tags are balanced before ':'; ok
1997                                         break;
1998                                 }
1999                                 $pos = $colon + 1;
2000                         }
2001                 } while ($colon !== false);
2002                 wfProfileOut( $fname );
2003                 return $colon;
2004         }
2005
2006         /**
2007          * Return value of a magic variable (like PAGENAME)
2008          *
2009          * @private
2010          */
2011         function getVariableValue( $index ) {
2012                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2013
2014                 /**
2015                  * Some of these require message or data lookups and can be
2016                  * expensive to check many times.
2017                  */
2018                 static $varCache = array();
2019                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$varCache ) ) )
2020                         if ( isset( $varCache[$index] ) )
2021                                 return $varCache[$index];
2022
2023                 $ts = time();
2024                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2025
2026                 switch ( $index ) {
2027                         case MAG_CURRENTMONTH:
2028                                 return $varCache[$index] = $wgContLang->formatNum( date( 'm', $ts ) );
2029                         case MAG_CURRENTMONTHNAME:
2030                                 return $varCache[$index] = $wgContLang->getMonthName( date( 'n', $ts ) );
2031                         case MAG_CURRENTMONTHNAMEGEN:
2032                                 return $varCache[$index] = $wgContLang->getMonthNameGen( date( 'n', $ts ) );
2033                         case MAG_CURRENTMONTHABBREV:
2034                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( date( 'n', $ts ) );
2035                         case MAG_CURRENTDAY:
2036                                 return $varCache[$index] = $wgContLang->formatNum( date( 'j', $ts ) );
2037                         case MAG_CURRENTDAY2:
2038                                 return $varCache[$index] = $wgContLang->formatNum( date( 'd', $ts ) );
2039                         case MAG_PAGENAME:
2040                                 return $this->mTitle->getText();
2041                         case MAG_PAGENAMEE:
2042                                 return $this->mTitle->getPartialURL();
2043                         case MAG_FULLPAGENAME:
2044                                 return $this->mTitle->getPrefixedText();
2045                         case MAG_FULLPAGENAMEE:
2046                                 return $this->mTitle->getPrefixedURL();
2047                         case MAG_SUBPAGENAME:
2048                                 return $this->mTitle->getSubpageText();
2049                         case MAG_SUBPAGENAMEE:
2050                                 return $this->mTitle->getSubpageUrlForm();
2051                         case MAG_BASEPAGENAME:
2052                                 return $this->mTitle->getBaseText();
2053                         case MAG_BASEPAGENAMEE:
2054                                 return wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2055                         case MAG_TALKPAGENAME:
2056                                 if( $this->mTitle->canTalk() ) {
2057                                         $talkPage = $this->mTitle->getTalkPage();
2058                                         return $talkPage->getPrefixedText();
2059                                 } else {
2060                                         return '';
2061                                 }
2062                         case MAG_TALKPAGENAMEE:
2063                                 if( $this->mTitle->canTalk() ) {
2064                                         $talkPage = $this->mTitle->getTalkPage();
2065                                         return $talkPage->getPrefixedUrl();
2066                                 } else {
2067                                         return '';
2068                                 }
2069                         case MAG_SUBJECTPAGENAME:
2070                                 $subjPage = $this->mTitle->getSubjectPage();
2071                                 return $subjPage->getPrefixedText();
2072                         case MAG_SUBJECTPAGENAMEE:
2073                                 $subjPage = $this->mTitle->getSubjectPage();
2074                                 return $subjPage->getPrefixedUrl();
2075                         case MAG_REVISIONID:
2076                                 return $this->mRevisionId;
2077                         case MAG_NAMESPACE:
2078                                 return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2079                         case MAG_NAMESPACEE:
2080                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2081                         case MAG_TALKSPACE:
2082                                 return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2083                         case MAG_TALKSPACEE:
2084                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2085                         case MAG_SUBJECTSPACE:
2086                                 return $this->mTitle->getSubjectNsText();
2087                         case MAG_SUBJECTSPACEE:
2088                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2089                         case MAG_CURRENTDAYNAME:
2090                                 return $varCache[$index] = $wgContLang->getWeekdayName( date( 'w', $ts ) + 1 );
2091                         case MAG_CURRENTYEAR:
2092                                 return $varCache[$index] = $wgContLang->formatNum( date( 'Y', $ts ), true );
2093                         case MAG_CURRENTTIME:
2094                                 return $varCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2095                         case MAG_CURRENTWEEK:
2096                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2097                                 // int to remove the padding
2098                                 return $varCache[$index] = $wgContLang->formatNum( (int)date( 'W', $ts ) );
2099                         case MAG_CURRENTDOW:
2100                                 return $varCache[$index] = $wgContLang->formatNum( date( 'w', $ts ) );
2101                         case MAG_NUMBEROFARTICLES:
2102                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfArticles() );
2103                         case MAG_NUMBEROFFILES:
2104                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfFiles() );
2105                         case MAG_NUMBEROFUSERS:
2106                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfUsers() );
2107                         case MAG_NUMBEROFPAGES:
2108                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfPages() );
2109                         case MAG_CURRENTTIMESTAMP:
2110                                 return $varCache[$index] = wfTimestampNow();
2111                         case MAG_CURRENTVERSION:
2112                                 global $wgVersion;
2113                                 return $wgVersion;
2114                         case MAG_SITENAME:
2115                                 return $wgSitename;
2116                         case MAG_SERVER:
2117                                 return $wgServer;
2118                         case MAG_SERVERNAME:
2119                                 return $wgServerName;
2120                         case MAG_SCRIPTPATH:
2121                                 return $wgScriptPath;
2122                         case MAG_DIRECTIONMARK:
2123                                 return $wgContLang->getDirMark();
2124                         default:
2125                                 $ret = null;
2126                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$varCache, &$index, &$ret ) ) )
2127                                         return $ret;
2128                                 else
2129                                         return null;
2130                 }
2131         }
2132
2133         /**
2134          * initialise the magic variables (like CURRENTMONTHNAME)
2135          *
2136          * @private
2137          */
2138         function initialiseVariables() {
2139                 $fname = 'Parser::initialiseVariables';
2140                 wfProfileIn( $fname );
2141                 global $wgVariableIDs;
2142                 $this->mVariables = array();
2143                 foreach ( $wgVariableIDs as $id ) {
2144                         $mw =& MagicWord::get( $id );
2145                         $mw->addToArray( $this->mVariables, $id );
2146                 }
2147                 wfProfileOut( $fname );
2148         }
2149
2150         /**
2151          * parse any parentheses in format ((title|part|part))
2152          * and call callbacks to get a replacement text for any found piece
2153          *
2154          * @param string $text The text to parse
2155          * @param array $callbacks rules in form:
2156          *     '{' => array(                            # opening parentheses
2157          *                                      'end' => '}',   # closing parentheses
2158          *                                      'cb' => array(2 => callback,    # replacement callback to call if {{..}} is found
2159          *                                                                4 => callback         # replacement callback to call if {{{{..}}}} is found
2160          *                                                                )
2161          *                                      )
2162          * @private
2163          */
2164         function replace_callback ($text, $callbacks) {
2165                 $openingBraceStack = array();   # this array will hold a stack of parentheses which are not closed yet
2166                 $lastOpeningBrace = -1;                 # last not closed parentheses
2167
2168                 for ($i = 0; $i < strlen($text); $i++) {
2169                         # check for any opening brace
2170                         $rule = null;
2171                         $nextPos = -1;
2172                         foreach ($callbacks as $key => $value) {
2173                                 $pos = strpos ($text, $key, $i);
2174                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)) {
2175                                         $rule = $value;
2176                                         $nextPos = $pos;
2177                                 }
2178                         }
2179
2180                         if ($lastOpeningBrace >= 0) {
2181                                 $pos = strpos ($text, $openingBraceStack[$lastOpeningBrace]['braceEnd'], $i);
2182
2183                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)){
2184                                         $rule = null;
2185                                         $nextPos = $pos;
2186                                 }
2187
2188                                 $pos = strpos ($text, '|', $i);
2189
2190                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)){
2191                                         $rule = null;
2192                                         $nextPos = $pos;
2193                                 }
2194                         }
2195
2196                         if ($nextPos == -1)
2197                                 break;
2198
2199                         $i = $nextPos;
2200
2201                         # found openning brace, lets add it to parentheses stack
2202                         if (null != $rule) {
2203                                 $piece = array('brace' => $text[$i],
2204                                                            'braceEnd' => $rule['end'],
2205                                                            'count' => 1,
2206                                                            'title' => '',
2207                                                            'parts' => null);
2208
2209                                 # count openning brace characters
2210                                 while ($i+1 < strlen($text) && $text[$i+1] == $piece['brace']) {
2211                                         $piece['count']++;
2212                                         $i++;
2213                                 }
2214
2215                                 $piece['startAt'] = $i+1;
2216                                 $piece['partStart'] = $i+1;
2217
2218                                 # we need to add to stack only if openning brace count is enough for any given rule
2219                                 foreach ($rule['cb'] as $cnt => $fn) {
2220                                         if ($piece['count'] >= $cnt) {
2221                                                 $lastOpeningBrace ++;
2222                                                 $openingBraceStack[$lastOpeningBrace] = $piece;
2223                                                 break;
2224                                         }
2225                                 }
2226
2227                                 continue;
2228                         }
2229                         else if ($lastOpeningBrace >= 0) {
2230                                 # first check if it is a closing brace
2231                                 if ($openingBraceStack[$lastOpeningBrace]['braceEnd'] == $text[$i]) {
2232                                         # lets check if it is enough characters for closing brace
2233                                         $count = 1;
2234                                         while ($i+$count < strlen($text) && $text[$i+$count] == $text[$i])
2235                                                 $count++;
2236
2237                                         # if there are more closing parentheses than opening ones, we parse less
2238                                         if ($openingBraceStack[$lastOpeningBrace]['count'] < $count)
2239                                                 $count = $openingBraceStack[$lastOpeningBrace]['count'];
2240
2241                                         # check for maximum matching characters (if there are 5 closing characters, we will probably need only 3 - depending on the rules)
2242                                         $matchingCount = 0;
2243                                         $matchingCallback = null;
2244                                         foreach ($callbacks[$openingBraceStack[$lastOpeningBrace]['brace']]['cb'] as $cnt => $fn) {
2245                                                 if ($count >= $cnt && $matchingCount < $cnt) {
2246                                                         $matchingCount = $cnt;
2247                                                         $matchingCallback = $fn;
2248                                                 }
2249                                         }
2250
2251                                         if ($matchingCount == 0) {
2252                                                 $i += $count - 1;
2253                                                 continue;
2254                                         }
2255
2256                                         # lets set a title or last part (if '|' was found)
2257                                         if (null === $openingBraceStack[$lastOpeningBrace]['parts'])
2258                                                 $openingBraceStack[$lastOpeningBrace]['title'] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2259                                         else
2260                                                 $openingBraceStack[$lastOpeningBrace]['parts'][] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2261
2262                                         $pieceStart = $openingBraceStack[$lastOpeningBrace]['startAt'] - $matchingCount;
2263                                         $pieceEnd = $i + $matchingCount;
2264
2265                                         if( is_callable( $matchingCallback ) ) {
2266                                                 $cbArgs = array (
2267                                                                                  'text' => substr($text, $pieceStart, $pieceEnd - $pieceStart),
2268                                                                                  'title' => trim($openingBraceStack[$lastOpeningBrace]['title']),
2269                                                                                  'parts' => $openingBraceStack[$lastOpeningBrace]['parts'],
2270                                                                                  'lineStart' => (($pieceStart > 0) && ($text[$pieceStart-1] == '\n')),
2271                                                                                  );
2272                                                 # finally we can call a user callback and replace piece of text
2273                                                 $replaceWith = call_user_func( $matchingCallback, $cbArgs );
2274                                                 $text = substr($text, 0, $pieceStart) . $replaceWith . substr($text, $pieceEnd);
2275                                                 $i = $pieceStart + strlen($replaceWith) - 1;
2276                                         }
2277                                         else {
2278                                                 # null value for callback means that parentheses should be parsed, but not replaced
2279                                                 $i += $matchingCount - 1;
2280                                         }
2281
2282                                         # reset last openning parentheses, but keep it in case there are unused characters
2283                                         $piece = array('brace' => $openingBraceStack[$lastOpeningBrace]['brace'],
2284                                                                    'braceEnd' => $openingBraceStack[$lastOpeningBrace]['braceEnd'],
2285                                                                    'count' => $openingBraceStack[$lastOpeningBrace]['count'],
2286                                                                    'title' => '',
2287                                                                    'parts' => null,
2288                                                                    'startAt' => $openingBraceStack[$lastOpeningBrace]['startAt']);
2289                                         $openingBraceStack[$lastOpeningBrace--] = null;
2290
2291                                         if ($matchingCount < $piece['count']) {
2292                                                 $piece['count'] -= $matchingCount;
2293                                                 $piece['startAt'] -= $matchingCount;
2294                                                 $piece['partStart'] = $piece['startAt'];
2295                                                 # do we still qualify for any callback with remaining count?
2296                                                 foreach ($callbacks[$piece['brace']]['cb'] as $cnt => $fn) {
2297                                                         if ($piece['count'] >= $cnt) {
2298                                                                 $lastOpeningBrace ++;
2299                                                                 $openingBraceStack[$lastOpeningBrace] = $piece;
2300                                                                 break;
2301                                                         }
2302                                                 }
2303                                         }
2304                                         continue;
2305                                 }
2306
2307                                 # lets set a title if it is a first separator, or next part otherwise
2308                                 if ($text[$i] == '|') {
2309                                         if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2310                                                 $openingBraceStack[$lastOpeningBrace]['title'] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2311                                                 $openingBraceStack[$lastOpeningBrace]['parts'] = array();
2312                                         }
2313                                         else
2314                                                 $openingBraceStack[$lastOpeningBrace]['parts'][] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2315
2316                                         $openingBraceStack[$lastOpeningBrace]['partStart'] = $i + 1;
2317                                 }
2318                         }
2319                 }
2320
2321                 return $text;
2322         }
2323
2324         /**
2325          * Replace magic variables, templates, and template arguments
2326          * with the appropriate text. Templates are substituted recursively,
2327          * taking care to avoid infinite loops.
2328          *
2329          * Note that the substitution depends on value of $mOutputType:
2330          *  OT_WIKI: only {{subst:}} templates
2331          *  OT_MSG: only magic variables
2332          *  OT_HTML: all templates and magic variables
2333          *
2334          * @param string $tex The text to transform
2335          * @param array $args Key-value pairs representing template parameters to substitute
2336          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2337          * @private
2338          */
2339         function replaceVariables( $text, $args = array(), $argsOnly = false ) {
2340                 # Prevent too big inclusions
2341                 if( strlen( $text ) > MAX_INCLUDE_SIZE ) {
2342                         return $text;
2343                 }
2344
2345                 $fname = 'Parser::replaceVariables';
2346                 wfProfileIn( $fname );
2347
2348                 # This function is called recursively. To keep track of arguments we need a stack:
2349                 array_push( $this->mArgStack, $args );
2350
2351                 $braceCallbacks = array();
2352                 if ( !$argsOnly ) {
2353                         $braceCallbacks[2] = array( &$this, 'braceSubstitution' );
2354                 }
2355                 if ( $this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI ) {
2356                         $braceCallbacks[3] = array( &$this, 'argSubstitution' );
2357                 }
2358                 $callbacks = array();
2359                 $callbacks['{'] = array('end' => '}', 'cb' => $braceCallbacks);
2360                 $callbacks['['] = array('end' => ']', 'cb' => array(2=>null));
2361                 $text = $this->replace_callback ($text, $callbacks);
2362
2363                 array_pop( $this->mArgStack );
2364
2365                 wfProfileOut( $fname );
2366                 return $text;
2367         }
2368
2369         /**
2370          * Replace magic variables
2371          * @private
2372          */
2373         function variableSubstitution( $matches ) {
2374                 $fname = 'Parser::variableSubstitution';
2375                 $varname = $matches[1];
2376                 wfProfileIn( $fname );
2377                 if ( !$this->mVariables ) {
2378                         $this->initialiseVariables();
2379                 }
2380                 $skip = false;
2381                 if ( $this->mOutputType == OT_WIKI ) {
2382                         # Do only magic variables prefixed by SUBST
2383                         $mwSubst =& MagicWord::get( MAG_SUBST );
2384                         if (!$mwSubst->matchStartAndRemove( $varname ))
2385                                 $skip = true;
2386                         # Note that if we don't substitute the variable below,
2387                         # we don't remove the {{subst:}} magic word, in case
2388                         # it is a template rather than a magic variable.
2389                 }
2390                 if ( !$skip && array_key_exists( $varname, $this->mVariables ) ) {
2391                         $id = $this->mVariables[$varname];
2392                         $text = $this->getVariableValue( $id );
2393                         $this->mOutput->mContainsOldMagic = true;
2394                 } else {
2395                         $text = $matches[0];
2396                 }
2397                 wfProfileOut( $fname );
2398                 return $text;
2399         }
2400
2401         # Split template arguments
2402         function getTemplateArgs( $argsString ) {
2403                 if ( $argsString === '' ) {
2404                         return array();
2405                 }
2406
2407                 $args = explode( '|', substr( $argsString, 1 ) );
2408
2409                 # If any of the arguments contains a '[[' but no ']]', it needs to be
2410                 # merged with the next arg because the '|' character between belongs
2411                 # to the link syntax and not the template parameter syntax.
2412                 $argc = count($args);
2413
2414                 for ( $i = 0; $i < $argc-1; $i++ ) {
2415                         if ( substr_count ( $args[$i], '[[' ) != substr_count ( $args[$i], ']]' ) ) {
2416                                 $args[$i] .= '|'.$args[$i+1];
2417                                 array_splice($args, $i+1, 1);
2418                                 $i--;
2419                                 $argc--;
2420                         }
2421                 }
2422
2423                 return $args;
2424         }
2425
2426         /**
2427          * Return the text of a template, after recursively
2428          * replacing any variables or templates within the template.
2429          *
2430          * @param array $piece The parts of the template
2431          *  $piece['text']: matched text
2432          *  $piece['title']: the title, i.e. the part before the |
2433          *  $piece['parts']: the parameter array
2434          * @return string the text of the template
2435          * @private
2436          */
2437         function braceSubstitution( $piece ) {
2438                 global $wgContLang, $wgLang, $wgAllowDisplayTitle, $action;
2439                 $fname = 'Parser::braceSubstitution';
2440                 wfProfileIn( $fname );
2441
2442                 # Flags
2443                 $found = false;             # $text has been filled
2444                 $nowiki = false;            # wiki markup in $text should be escaped
2445                 $noparse = false;           # Unsafe HTML tags should not be stripped, etc.
2446                 $noargs = false;            # Don't replace triple-brace arguments in $text
2447                 $replaceHeadings = false;   # Make the edit section links go to the template not the article
2448                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2449                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2450
2451                 # Title object, where $text came from
2452                 $title = NULL;
2453
2454                 $linestart = '';
2455
2456                 # $part1 is the bit before the first |, and must contain only title characters
2457                 # $args is a list of arguments, starting from index 0, not including $part1
2458
2459                 $part1 = $piece['title'];
2460                 # If the third subpattern matched anything, it will start with |
2461
2462                 if (null == $piece['parts']) {
2463                         $replaceWith = $this->variableSubstitution (array ($piece['text'], $piece['title']));
2464                         if ($replaceWith != $piece['text']) {
2465                                 $text = $replaceWith;
2466                                 $found = true;
2467                                 $noparse = true;
2468                                 $noargs = true;
2469                         }
2470                 }
2471
2472                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2473                 $argc = count( $args );
2474
2475                 # SUBST
2476                 if ( !$found ) {
2477                         $mwSubst =& MagicWord::get( MAG_SUBST );
2478                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor ($this->mOutputType == OT_WIKI) ) {
2479                                 # One of two possibilities is true:
2480                                 # 1) Found SUBST but not in the PST phase
2481                                 # 2) Didn't find SUBST and in the PST phase
2482                                 # In either case, return without further processing
2483                                 $text = $piece['text'];
2484                                 $found = true;
2485                                 $noparse = true;
2486                                 $noargs = true;
2487                         }
2488                 }
2489
2490                 # MSG, MSGNW, INT and RAW
2491                 if ( !$found ) {
2492                         # Check for MSGNW:
2493                         $mwMsgnw =& MagicWord::get( MAG_MSGNW );
2494                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2495                                 $nowiki = true;
2496                         } else {
2497                                 # Remove obsolete MSG:
2498                                 $mwMsg =& MagicWord::get( MAG_MSG );
2499                                 $mwMsg->matchStartAndRemove( $part1 );
2500                         }
2501
2502                         # Check for RAW:
2503                         $mwRaw =& MagicWord::get( MAG_RAW );
2504                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2505                                 $forceRawInterwiki = true;
2506                         }
2507
2508                         # Check if it is an internal message
2509                         $mwInt =& MagicWord::get( MAG_INT );
2510                         if ( $mwInt->matchStartAndRemove( $part1 ) ) {
2511                                 if ( $this->incrementIncludeCount( 'int:'.$part1 ) ) {
2512                                         $text = $linestart . wfMsgReal( $part1, $args, true );
2513                                         $found = true;
2514                                 }
2515                         }
2516                 }
2517
2518                 # NS
2519                 if ( !$found ) {
2520                         # Check for NS: (namespace expansion)
2521                         $mwNs = MagicWord::get( MAG_NS );
2522                         if ( $mwNs->matchStartAndRemove( $part1 ) ) {
2523                                 if ( intval( $part1 ) || $part1 == "0" ) {
2524                                         $text = $linestart . $wgContLang->getNsText( intval( $part1 ) );
2525                                         $found = true;
2526                                 } else {
2527                                         $index = Namespace::getCanonicalIndex( strtolower( $part1 ) );
2528                                         if ( !is_null( $index ) ) {
2529                                                 $text = $linestart . $wgContLang->getNsText( $index );
2530                                                 $found = true;
2531                                         }
2532                                 }
2533                         }
2534                 }
2535
2536                 # URLENCODE
2537                 if( !$found ) {
2538                         $urlencode =& MagicWord::get( MAG_URLENCODE );
2539                         if( $urlencode->matchStartAndRemove( $part1 ) ) {
2540                                 $text = $linestart . urlencode( $part1 );
2541                                 $found = true;
2542                         }
2543                 }
2544
2545                 # LCFIRST, UCFIRST, LC and UC
2546                 if ( !$found ) {
2547                         $lcfirst =& MagicWord::get( MAG_LCFIRST );
2548                         $ucfirst =& MagicWord::get( MAG_UCFIRST );
2549                         $lc =& MagicWord::get( MAG_LC );
2550                         $uc =& MagicWord::get( MAG_UC );
2551                         if ( $lcfirst->matchStartAndRemove( $part1 ) ) {
2552                                 $text = $linestart . $wgContLang->lcfirst( $part1 );
2553                                 $found = true;
2554                         } else if ( $ucfirst->matchStartAndRemove( $part1 ) ) {
2555                                 $text = $linestart . $wgContLang->ucfirst( $part1 );
2556                                 $found = true;
2557                         } else if ( $lc->matchStartAndRemove( $part1 ) ) {
2558                                 $text = $linestart . $wgContLang->lc( $part1 );
2559                                 $found = true;
2560                         } else if ( $uc->matchStartAndRemove( $part1 ) ) {
2561                                  $text = $linestart . $wgContLang->uc( $part1 );
2562                                  $found = true;
2563                         }
2564                 }
2565
2566                 # LOCALURL and FULLURL
2567                 if ( !$found ) {
2568                         $mwLocal =& MagicWord::get( MAG_LOCALURL );
2569                         $mwLocalE =& MagicWord::get( MAG_LOCALURLE );
2570                         $mwFull =& MagicWord::get( MAG_FULLURL );
2571                         $mwFullE =& MagicWord::get( MAG_FULLURLE );
2572
2573
2574                         if ( $mwLocal->matchStartAndRemove( $part1 ) ) {
2575                                 $func = 'getLocalURL';
2576                         } elseif ( $mwLocalE->matchStartAndRemove( $part1 ) ) {
2577                                 $func = 'escapeLocalURL';
2578                         } elseif ( $mwFull->matchStartAndRemove( $part1 ) ) {
2579                                 $func = 'getFullURL';
2580                         } elseif ( $mwFullE->matchStartAndRemove( $part1 ) ) {
2581                                 $func = 'escapeFullURL';
2582                         } else {
2583                                 $func = false;
2584                         }
2585
2586                         if ( $func !== false ) {
2587                                 $title = Title::newFromText( $part1 );
2588                                 # Due to order of execution of a lot of bits, the values might be encoded
2589                                 # before arriving here; if that's true, then the title can't be created
2590                                 # and the variable will fail. If we can't get a decent title from the first
2591                                 # attempt, url-decode and try for a second.
2592                                 if( is_null( $title ) )
2593                                         $title = Title::newFromUrl( urldecode( $part1 ) );
2594                                 if ( !is_null( $title ) ) {
2595                                         if ( $argc > 0 ) {
2596                                                 $text = $linestart . $title->$func( $args[0] );
2597                                         } else {
2598                                                 $text = $linestart . $title->$func();
2599                                         }
2600                                         $found = true;
2601                                 }
2602                         }
2603                 }
2604
2605                 $lang = $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
2606                 # GRAMMAR
2607                 if ( !$found && $argc == 1 ) {
2608                         $mwGrammar =& MagicWord::get( MAG_GRAMMAR );
2609                         if ( $mwGrammar->matchStartAndRemove( $part1 ) ) {
2610                                 $text = $linestart . $lang->convertGrammar( $args[0], $part1 );
2611                                 $found = true;
2612                         }
2613                 }
2614
2615                 # PLURAL
2616                 if ( !$found && $argc >= 2 ) {
2617                         $mwPluralForm =& MagicWord::get( MAG_PLURAL );
2618                         if ( $mwPluralForm->matchStartAndRemove( $part1 ) ) {
2619                                 if ($argc==2) {$args[2]=$args[1];}
2620                                 $text = $linestart . $lang->convertPlural( $part1, $args[0], $args[1], $args[2]);
2621                                 $found = true;
2622                         }
2623                 }
2624
2625                 # DISPLAYTITLE
2626                 if ( !$found && $argc == 1 && $wgAllowDisplayTitle ) {
2627                         $mwDT =& MagicWord::get( MAG_DISPLAYTITLE );
2628                         if ( $mwDT->matchStartAndRemove( $part1 ) ) {
2629
2630                                 # Set title in parser output object
2631                                 $param = $args[0];
2632                                 $parserOptions = new ParserOptions;
2633                                 $local_parser = new Parser ();
2634                                 $t2 = $local_parser->parse ( $param, $this->mTitle, $parserOptions, false );
2635                                 $this->mOutput->mHTMLtitle = $t2->GetText();
2636
2637                                 # Add subtitle
2638                                 $t = $this->mTitle->getPrefixedText();
2639                                 $this->mOutput->mSubtitle .= wfMsg('displaytitle', $t);
2640                                 $text = "" ;
2641                                 $found = true ;
2642                         }
2643                 }
2644
2645                 # NUMBEROFPAGES, NUMBEROFUSERS, NUMBEROFARTICLES, and NUMBEROFFILES
2646                 if( !$found ) {
2647                         $mwWordsToCheck = array( MAG_NUMBEROFPAGES => 'wfNumberOfPages',
2648                                                                          MAG_NUMBEROFUSERS => 'wfNumberOfUsers',
2649                                                                          MAG_NUMBEROFARTICLES => 'wfNumberOfArticles',
2650                                                                          MAG_NUMBEROFFILES => 'wfNumberOfFiles' );
2651                         foreach( $mwWordsToCheck as $word => $func ) {
2652                                 $mwCurrentWord =& MagicWord::get( $word );
2653                                 if( $mwCurrentWord->matchStartAndRemove( $part1 ) ) {
2654                                         $mwRawSuffix =& MagicWord::get( MAG_RAWSUFFIX );
2655                                         if( $mwRawSuffix->match( $args[0] ) ) {
2656                                                 # Raw and unformatted
2657                                                 $text = $linestart . call_user_func( $func );
2658                                         } else {
2659                                                 # Formatted according to the content default
2660                                                 $text = $linestart . $wgContLang->formatNum( call_user_func( $func ) );
2661                                         }
2662                                         $found = true;
2663                                 }
2664                         }
2665                 }
2666
2667                 # #LANGUAGE:
2668                 if( !$found ) {
2669                         $mwLanguage =& MagicWord::get( MAG_LANGUAGE );
2670                         if( $mwLanguage->matchStartAndRemove( $part1 ) ) {
2671                                 $lang = $wgContLang->getLanguageName( strtolower( $part1 ) );
2672                                 $text = $linestart . ( $lang != '' ? $lang : $part1 );
2673                                 $found = true;
2674                         }
2675                 }
2676
2677                 # Extensions
2678                 if ( !$found && substr( $part1, 0, 1 ) == '#' ) {
2679                         $colonPos = strpos( $part1, ':' );
2680                         if ( $colonPos !== false ) {
2681                                 $function = strtolower( substr( $part1, 1, $colonPos - 1 ) );
2682                                 if ( isset( $this->mFunctionHooks[$function] ) ) {
2683                                         $funcArgs = array_map( 'trim', $args );
2684                                         $funcArgs = array_merge( array( &$this, trim( substr( $part1, $colonPos + 1 ) ) ), $funcArgs );
2685                                         $result = call_user_func_array( $this->mFunctionHooks[$function], $funcArgs );
2686                                         $found = true;
2687
2688                                         // The text is usually already parsed, doesn't need triple-brace tags expanded, etc.
2689                                         //$noargs = true;
2690                                         //$noparse = true;
2691
2692                                         if ( is_array( $result ) ) {
2693                                                 $text = $linestart . $result[0];
2694                                                 unset( $result[0] );
2695
2696                                                 // Extract flags into the local scope
2697                                                 // This allows callers to set flags such as nowiki, noparse, found, etc.
2698                                                 extract( $result );
2699                                         } else {
2700                                                 $text = $linestart . $result;
2701                                         }
2702                                 }
2703                         }
2704                 }
2705
2706                 # Template table test
2707
2708                 # Did we encounter this template already? If yes, it is in the cache
2709                 # and we need to check for loops.
2710                 if ( !$found && isset( $this->mTemplates[$piece['title']] ) ) {
2711                         $found = true;
2712
2713                         # Infinite loop test
2714                         if ( isset( $this->mTemplatePath[$part1] ) ) {
2715                                 $noparse = true;
2716                                 $noargs = true;
2717                                 $found = true;
2718                                 $text = $linestart .
2719                                         '{{' . $part1 . '}}' .
2720                                         '<!-- WARNING: template loop detected -->';
2721                                 wfDebug( "$fname: template loop broken at '$part1'\n" );
2722                         } else {
2723                                 # set $text to cached message.
2724                                 $text = $linestart . $this->mTemplates[$piece['title']];
2725                         }
2726                 }
2727
2728                 # Load from database
2729                 $lastPathLevel = $this->mTemplatePath;
2730                 if ( !$found ) {
2731                         $ns = NS_TEMPLATE;
2732                         # declaring $subpage directly in the function call
2733                         # does not work correctly with references and breaks
2734                         # {{/subpage}}-style inclusions
2735                         $subpage = '';
2736                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
2737                         if ($subpage !== '') {
2738                                 $ns = $this->mTitle->getNamespace();
2739                         }
2740                         $title = Title::newFromText( $part1, $ns );
2741
2742                         if ( !is_null( $title ) ) {
2743                                 if ( !$title->isExternal() ) {
2744                                         # Check for excessive inclusion
2745                                         $dbk = $title->getPrefixedDBkey();
2746                                         if ( $this->incrementIncludeCount( $dbk ) ) {
2747                                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->mOutputType != OT_WIKI ) {
2748                                                         $text = SpecialPage::capturePath( $title );
2749                                                         if ( is_string( $text ) ) {
2750                                                                 $found = true;
2751                                                                 $noparse = true;
2752                                                                 $noargs = true;
2753                                                                 $isHTML = true;
2754                                                                 $this->disableCache();
2755                                                         }
2756                                                 } else {
2757                                                         $articleContent = $this->fetchTemplate( $title );
2758                                                         if ( $articleContent !== false ) {
2759                                                                 $found = true;
2760                                                                 $text = $articleContent;
2761                                                                 $replaceHeadings = true;
2762                                                         }
2763                                                 }
2764                                         }
2765
2766                                         # If the title is valid but undisplayable, make a link to it
2767                                         if ( $this->mOutputType == OT_HTML && !$found ) {
2768                                                 $text = '[['.$title->getPrefixedText().']]';
2769                                                 $found = true;
2770                                         }
2771                                 } elseif ( $title->isTrans() ) {
2772                                         // Interwiki transclusion
2773                                         if ( $this->mOutputType == OT_HTML && !$forceRawInterwiki ) {
2774                                                 $text = $this->interwikiTransclude( $title, 'render' );
2775                                                 $isHTML = true;
2776                                                 $noparse = true;
2777                                         } else {
2778                                                 $text = $this->interwikiTransclude( $title, 'raw' );
2779                                                 $replaceHeadings = true;
2780                                         }
2781                                         $found = true;
2782                                 }
2783
2784                                 # Template cache array insertion
2785                                 # Use the original $piece['title'] not the mangled $part1, so that
2786                                 # modifiers such as RAW: produce separate cache entries
2787                                 if( $found ) {
2788                                         $this->mTemplates[$piece['title']] = $text;
2789                                         $text = $linestart . $text;
2790                                 }
2791                         }
2792                 }
2793
2794                 # Recursive parsing, escaping and link table handling
2795                 # Only for HTML output
2796                 if ( $nowiki && $found && $this->mOutputType == OT_HTML ) {
2797                         $text = wfEscapeWikiText( $text );
2798                 } elseif ( ($this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI) && $found ) {
2799                         if ( $noargs ) {
2800                                 $assocArgs = array();
2801                         } else {
2802                                 # Clean up argument array
2803                                 $assocArgs = array();
2804                                 $index = 1;
2805                                 foreach( $args as $arg ) {
2806                                         $eqpos = strpos( $arg, '=' );
2807                                         if ( $eqpos === false ) {
2808                                                 $assocArgs[$index++] = $arg;
2809                                         } else {
2810                                                 $name = trim( substr( $arg, 0, $eqpos ) );
2811                                                 $value = trim( substr( $arg, $eqpos+1 ) );
2812                                                 if ( $value === false ) {
2813                                                         $value = '';
2814                                                 }
2815                                                 if ( $name !== false ) {
2816                                                         $assocArgs[$name] = $value;
2817                                                 }
2818                                         }
2819                                 }
2820
2821                                 # Add a new element to the templace recursion path
2822                                 $this->mTemplatePath[$part1] = 1;
2823                         }
2824
2825                         if ( !$noparse ) {
2826                                 # If there are any <onlyinclude> tags, only include them
2827                                 if ( in_string( '<onlyinclude>', $text ) && in_string( '</onlyinclude>', $text ) ) {
2828                                         preg_match_all( '/<onlyinclude>(.*?)\n?<\/onlyinclude>/s', $text, $m );
2829                                         $text = '';
2830                                         foreach ($m[1] as $piece)
2831                                                 $text .= $piece;
2832                                 }
2833                                 # Remove <noinclude> sections and <includeonly> tags
2834                                 $text = preg_replace( '/<noinclude>.*?<\/noinclude>/s', '', $text );
2835                                 $text = strtr( $text, array( '<includeonly>' => '' , '</includeonly>' => '' ) );
2836
2837                                 if( $this->mOutputType == OT_HTML ) {
2838                                         # Strip <nowiki>, <pre>, etc.
2839                                         $text = $this->strip( $text, $this->mStripState );
2840                                         $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'replaceVariables' ), $assocArgs );
2841                                 }
2842                                 $text = $this->replaceVariables( $text, $assocArgs );
2843
2844                                 # If the template begins with a table or block-level
2845                                 # element, it should be treated as beginning a new line.
2846                                 if (!$piece['lineStart'] && preg_match('/^({\\||:|;|#|\*)/', $text)) {
2847                                         $text = "\n" . $text;
2848                                 }
2849                         } elseif ( !$noargs ) {
2850                                 # $noparse and !$noargs
2851                                 # Just replace the arguments, not any double-brace items
2852                                 # This is used for rendered interwiki transclusion
2853                                 $text = $this->replaceVariables( $text, $assocArgs, true );
2854                         }
2855                 }
2856                 # Prune lower levels off the recursion check path
2857                 $this->mTemplatePath = $lastPathLevel;
2858
2859                 if ( !$found ) {
2860                         wfProfileOut( $fname );
2861                         return $piece['text'];
2862                 } else {
2863                         if ( $isHTML ) {
2864                                 # Replace raw HTML by a placeholder
2865                                 # Add a blank line preceding, to prevent it from mucking up
2866                                 # immediately preceding headings
2867                                 $text = "\n\n" . $this->insertStripItem( $text, $this->mStripState );
2868                         } else {
2869                                 # replace ==section headers==
2870                                 # XXX this needs to go away once we have a better parser.
2871                                 if ( $this->mOutputType != OT_WIKI && $replaceHeadings ) {
2872                                         if( !is_null( $title ) )
2873                                                 $encodedname = base64_encode($title->getPrefixedDBkey());
2874                                         else
2875                                                 $encodedname = base64_encode("");
2876                                         $m = preg_split('/(^={1,6}.*?={1,6}\s*?$)/m', $text, -1,
2877                                                 PREG_SPLIT_DELIM_CAPTURE);
2878                                         $text = '';
2879                                         $nsec = 0;
2880                                         for( $i = 0; $i < count($m); $i += 2 ) {
2881                                                 $text .= $m[$i];
2882                                                 if (!isset($m[$i + 1]) || $m[$i + 1] == "") continue;
2883                                                 $hl = $m[$i + 1];
2884                                                 if( strstr($hl, "<!--MWTEMPLATESECTION") ) {
2885                                                         $text .= $hl;
2886                                                         continue;
2887                                                 }
2888                                                 preg_match('/^(={1,6})(.*?)(={1,6})\s*?$/m', $hl, $m2);
2889                                                 $text .= $m2[1] . $m2[2] . "<!--MWTEMPLATESECTION="
2890                                                         . $encodedname . "&" . base64_encode("$nsec") . "-->" . $m2[3];
2891
2892                                                 $nsec++;
2893                                         }
2894                                 }
2895                         }
2896                 }
2897
2898                 # Prune lower levels off the recursion check path
2899                 $this->mTemplatePath = $lastPathLevel;
2900
2901                 if ( !$found ) {
2902                         wfProfileOut( $fname );
2903                         return $piece['text'];
2904                 } else {
2905                         wfProfileOut( $fname );
2906                         return $text;
2907                 }
2908         }
2909
2910         /**
2911          * Fetch the unparsed text of a template and register a reference to it.
2912          */
2913         function fetchTemplate( $title ) {
2914                 $text = false;
2915                 // Loop to fetch the article, with up to 1 redirect
2916                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
2917                         $rev = Revision::newFromTitle( $title );
2918                         $this->mOutput->addTemplate( $title, $title->getArticleID() );
2919                         if ( !$rev ) {
2920                                 break;
2921                         }
2922                         $text = $rev->getText();
2923                         if ( $text === false ) {
2924                                 break;
2925                         }
2926                         // Redirect?
2927                         $title = Title::newFromRedirect( $text );
2928                 }
2929                 return $text;
2930         }
2931
2932         /**
2933          * Transclude an interwiki link.
2934          */
2935         function interwikiTransclude( $title, $action ) {
2936                 global $wgEnableScaryTranscluding, $wgCanonicalNamespaceNames;
2937
2938                 if (!$wgEnableScaryTranscluding)
2939                         return wfMsg('scarytranscludedisabled');
2940
2941                 // The namespace will actually only be 0 or 10, depending on whether there was a leading :
2942                 // But we'll handle it generally anyway
2943                 if ( $title->getNamespace() ) {
2944                         // Use the canonical namespace, which should work anywhere
2945                         $articleName = $wgCanonicalNamespaceNames[$title->getNamespace()] . ':' . $title->getDBkey();
2946                 } else {
2947                         $articleName = $title->getDBkey();
2948                 }
2949
2950                 $url = str_replace('$1', urlencode($articleName), Title::getInterwikiLink($title->getInterwiki()));
2951                 $url .= "?action=$action";
2952                 if (strlen($url) > 255)
2953                         return wfMsg('scarytranscludetoolong');
2954                 return $this->fetchScaryTemplateMaybeFromCache($url);
2955         }
2956
2957         function fetchScaryTemplateMaybeFromCache($url) {
2958                 global $wgTranscludeCacheExpiry;
2959                 $dbr =& wfGetDB(DB_SLAVE);
2960                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
2961                                 array('tc_url' => $url));
2962                 if ($obj) {
2963                         $time = $obj->tc_time;
2964                         $text = $obj->tc_contents;
2965                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
2966                                 return $text;
2967                         }
2968                 }
2969
2970                 $text = HttpFunctions::getHTTP($url);
2971                 if (!$text)
2972                         return wfMsg('scarytranscludefailed', $url);
2973
2974                 $dbw =& wfGetDB(DB_MASTER);
2975                 $dbw->replace('transcache', array('tc_url'), array(
2976                         'tc_url' => $url,
2977                         'tc_time' => time(),
2978                         'tc_contents' => $text));
2979                 return $text;
2980         }
2981
2982
2983         /**
2984          * Triple brace replacement -- used for template arguments
2985          * @private
2986          */
2987         function argSubstitution( $matches ) {
2988                 $arg = trim( $matches['title'] );
2989                 $text = $matches['text'];
2990                 $inputArgs = end( $this->mArgStack );
2991
2992                 if ( array_key_exists( $arg, $inputArgs ) ) {
2993                         $text = $inputArgs[$arg];
2994                 } else if ($this->mOutputType == OT_HTML && null != $matches['parts'] && count($matches['parts']) > 0) {
2995                         $text = $matches['parts'][0];
2996                 }
2997
2998                 return $text;
2999         }
3000
3001         /**
3002          * Returns true if the function is allowed to include this entity
3003          * @private
3004          */
3005         function incrementIncludeCount( $dbk ) {
3006                 if ( !array_key_exists( $dbk, $this->mIncludeCount ) ) {
3007                         $this->mIncludeCount[$dbk] = 0;
3008                 }
3009                 if ( ++$this->mIncludeCount[$dbk] <= MAX_INCLUDE_REPEAT ) {
3010                         return true;
3011                 } else {
3012                         return false;
3013                 }
3014         }
3015
3016         /**
3017          * Detect __TOC__ magic word and set a placeholder
3018          */
3019         function stripToc( $text ) {
3020                 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
3021                 # do not add TOC
3022                 $mw = MagicWord::get( MAG_NOTOC );
3023                 if( $mw->matchAndRemove( $text ) ) {
3024                         $this->mShowToc = false;
3025                 }
3026
3027                 $mw = MagicWord::get( MAG_TOC );
3028                 if( $mw->match( $text ) ) {
3029                         $this->mShowToc = true;
3030                         $this->mForceTocPosition = true;
3031
3032                         // Set a placeholder. At the end we'll fill it in with the TOC.
3033                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3034
3035                         // Only keep the first one.
3036                         $text = $mw->replace( '', $text );
3037                 }
3038                 return $text;
3039         }
3040
3041         /**
3042          * This function accomplishes several tasks:
3043          * 1) Auto-number headings if that option is enabled
3044          * 2) Add an [edit] link to sections for logged in users who have enabled the option
3045          * 3) Add a Table of contents on the top for users who have enabled the option
3046          * 4) Auto-anchor headings
3047          *
3048          * It loops through all headlines, collects the necessary data, then splits up the
3049          * string and re-inserts the newly formatted headlines.
3050          *
3051          * @param string $text
3052          * @param boolean $isMain
3053          * @private
3054          */
3055         function formatHeadings( $text, $isMain=true ) {
3056                 global $wgMaxTocLevel, $wgContLang;
3057
3058                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3059                 if( !$this->mTitle->userCanEdit() ) {
3060                         $showEditLink = 0;
3061                 } else {
3062                         $showEditLink = $this->mOptions->getEditSection();
3063                 }
3064
3065                 # Inhibit editsection links if requested in the page
3066                 $esw =& MagicWord::get( MAG_NOEDITSECTION );
3067                 if( $esw->matchAndRemove( $text ) ) {
3068                         $showEditLink = 0;
3069                 }
3070
3071                 # Get all headlines for numbering them and adding funky stuff like [edit]
3072                 # links - this is for later, but we need the number of headlines right now
3073                 $numMatches = preg_match_all( '/<H([1-6])(.*?'.'>)(.*?)<\/H[1-6] *>/i', $text, $matches );
3074
3075                 # if there are fewer than 4 headlines in the article, do not show TOC
3076                 # unless it's been explicitly enabled.
3077                 $enoughToc = $this->mShowToc &&
3078                         (($numMatches >= 4) || $this->mForceTocPosition);
3079
3080                 # Allow user to stipulate that a page should have a "new section"
3081                 # link added via __NEWSECTIONLINK__
3082                 $mw =& MagicWord::get( MAG_NEWSECTIONLINK );
3083                 if( $mw->matchAndRemove( $text ) )
3084                         $this->mOutput->setNewSection( true );
3085
3086                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3087                 # override above conditions and always show TOC above first header
3088                 $mw =& MagicWord::get( MAG_FORCETOC );
3089                 if ($mw->matchAndRemove( $text ) ) {
3090                         $this->mShowToc = true;
3091                         $enoughToc = true;
3092                 }
3093
3094                 # Never ever show TOC if no headers
3095                 if( $numMatches < 1 ) {
3096                         $enoughToc = false;
3097                 }
3098
3099                 # We need this to perform operations on the HTML
3100                 $sk =& $this->mOptions->getSkin();
3101
3102                 # headline counter
3103                 $headlineCount = 0;
3104                 $sectionCount = 0; # headlineCount excluding template sections
3105
3106                 # Ugh .. the TOC should have neat indentation levels which can be
3107                 # passed to the skin functions. These are determined here
3108                 $toc = '';
3109                 $full = '';
3110                 $head = array();
3111                 $sublevelCount = array();
3112                 $levelCount = array();
3113                 $toclevel = 0;
3114                 $level = 0;
3115                 $prevlevel = 0;
3116                 $toclevel = 0;
3117                 $prevtoclevel = 0;
3118
3119                 foreach( $matches[3] as $headline ) {
3120                         $istemplate = 0;
3121                         $templatetitle = '';
3122                         $templatesection = 0;
3123                         $numbering = '';
3124
3125                         if (preg_match("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", $headline, $mat)) {
3126                                 $istemplate = 1;
3127                                 $templatetitle = base64_decode($mat[1]);
3128                                 $templatesection = 1 + (int)base64_decode($mat[2]);
3129                                 $headline = preg_replace("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", "", $headline);
3130                         }
3131
3132                         if( $toclevel ) {
3133                                 $prevlevel = $level;
3134                                 $prevtoclevel = $toclevel;
3135                         }
3136                         $level = $matches[1][$headlineCount];
3137
3138                         if( $doNumberHeadings || $enoughToc ) {
3139
3140                                 if ( $level > $prevlevel ) {
3141                                         # Increase TOC level
3142                                         $toclevel++;
3143                                         $sublevelCount[$toclevel] = 0;
3144                                         if( $toclevel<$wgMaxTocLevel ) {
3145                                                 $toc .= $sk->tocIndent();
3146                                         }
3147                                 }
3148                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3149                                         # Decrease TOC level, find level to jump to
3150
3151                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
3152                                                 # Can only go down to level 1
3153                                                 $toclevel = 1;
3154                                         } else {
3155                                                 for ($i = $toclevel; $i > 0; $i--) {
3156                                                         if ( $levelCount[$i] == $level ) {
3157                                                                 # Found last matching level
3158                                                                 $toclevel = $i;
3159                                                                 break;
3160                                                         }
3161                                                         elseif ( $levelCount[$i] < $level ) {
3162                                                                 # Found first matching level below current level
3163                                                                 $toclevel = $i + 1;
3164                                                                 break;
3165                                                         }
3166                                                 }
3167                                         }
3168                                         if( $toclevel<$wgMaxTocLevel ) {
3169                                                 $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3170                                         }
3171                                 }
3172                                 else {
3173                                         # No change in level, end TOC line
3174                                         if( $toclevel<$wgMaxTocLevel ) {
3175                                                 $toc .= $sk->tocLineEnd();
3176                                         }
3177                                 }
3178
3179                                 $levelCount[$toclevel] = $level;
3180
3181                                 # count number of headlines for each level
3182                                 @$sublevelCount[$toclevel]++;
3183                                 $dot = 0;
3184                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3185                                         if( !empty( $sublevelCount[$i] ) ) {
3186                                                 if( $dot ) {
3187                                                         $numbering .= '.';
3188                                                 }
3189                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3190                                                 $dot = 1;
3191                                         }
3192                                 }
3193                         }
3194
3195                         # The canonized header is a version of the header text safe to use for links
3196                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3197                         $canonized_headline = $this->unstrip( $headline, $this->mStripState );
3198                         $canonized_headline = $this->unstripNoWiki( $canonized_headline, $this->mStripState );
3199
3200                         # Remove link placeholders by the link text.
3201                         #     <!--LINK number-->
3202                         # turns into
3203                         #     link text with suffix
3204                         $canonized_headline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
3205                                                             "\$this->mLinkHolders['texts'][\$1]",
3206                                                             $canonized_headline );
3207                         $canonized_headline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
3208                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
3209                                                             $canonized_headline );
3210
3211                         # strip out HTML
3212                         $canonized_headline = preg_replace( '/<.*?' . '>/','',$canonized_headline );
3213                         $tocline = trim( $canonized_headline );
3214                         # Save headline for section edit hint before it's escaped
3215                         $headline_hint = trim( $canonized_headline );
3216                         $canonized_headline = Sanitizer::escapeId( $tocline );
3217                         $refers[$headlineCount] = $canonized_headline;
3218
3219                         # count how many in assoc. array so we can track dupes in anchors
3220                         @$refers[$canonized_headline]++;
3221                         $refcount[$headlineCount]=$refers[$canonized_headline];
3222
3223                         # Don't number the heading if it is the only one (looks silly)
3224                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3225                                 # the two are different if the line contains a link
3226                                 $headline=$numbering . ' ' . $headline;
3227                         }
3228
3229                         # Create the anchor for linking from the TOC to the section
3230                         $anchor = $canonized_headline;
3231                         if($refcount[$headlineCount] > 1 ) {
3232                                 $anchor .= '_' . $refcount[$headlineCount];
3233                         }
3234                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3235                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3236                         }
3237                         if( $showEditLink && ( !$istemplate || $templatetitle !== "" ) ) {
3238                                 if ( empty( $head[$headlineCount] ) ) {
3239                                         $head[$headlineCount] = '';
3240                                 }
3241                                 if( $istemplate )
3242                                         $head[$headlineCount] .= $sk->editSectionLinkForOther($templatetitle, $templatesection);
3243                                 else
3244                                         $head[$headlineCount] .= $sk->editSectionLink($this->mTitle, $sectionCount+1, $headline_hint);
3245                         }
3246
3247                         # give headline the correct <h#> tag
3248                         @$head[$headlineCount] .= "<a name=\"$anchor\"></a><h".$level.$matches[2][$headlineCount] .$headline.'</h'.$level.'>';
3249
3250                         $headlineCount++;
3251                         if( !$istemplate )
3252                                 $sectionCount++;
3253                 }
3254
3255                 if( $enoughToc ) {
3256                         if( $toclevel<$wgMaxTocLevel ) {
3257                                 $toc .= $sk->tocUnindent( $toclevel - 1 );
3258                         }
3259                         $toc = $sk->tocList( $toc );
3260                 }
3261
3262                 # split up and insert constructed headlines
3263
3264                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3265                 $i = 0;
3266
3267                 foreach( $blocks as $block ) {
3268                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
3269                                 # This is the [edit] link that appears for the top block of text when
3270                                 # section editing is enabled
3271
3272                                 # Disabled because it broke block formatting
3273                                 # For example, a bullet point in the top line
3274                                 # $full .= $sk->editSectionLink(0);
3275                         }
3276                         $full .= $block;
3277                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3278                                 # Top anchor now in skin
3279                                 $full = $full.$toc;
3280                         }
3281
3282                         if( !empty( $head[$i] ) ) {
3283                                 $full .= $head[$i];
3284                         }
3285                         $i++;
3286                 }
3287                 if( $this->mForceTocPosition ) {
3288                         return str_replace( '<!--MWTOC-->', $toc, $full );
3289                 } else {
3290                         return $full;
3291                 }
3292         }
3293
3294         /**
3295          * Return an HTML link for the "ISBN 123456" text
3296          * @private
3297          */
3298         function magicISBN( $text ) {
3299                 $fname = 'Parser::magicISBN';
3300                 wfProfileIn( $fname );
3301
3302                 $a = split( 'ISBN ', ' '.$text );
3303                 if ( count ( $a ) < 2 ) {
3304                         wfProfileOut( $fname );
3305                         return $text;
3306                 }
3307                 $text = substr( array_shift( $a ), 1);
3308                 $valid = '0123456789-Xx';
3309
3310                 foreach ( $a as $x ) {
3311                         # hack: don't replace inside thumbnail title/alt
3312                         # attributes
3313                         if(preg_match('/<[^>]+(alt|title)="[^">]*$/', $text)) {
3314                                 $text .= "ISBN $x";
3315                                 continue;
3316                         }
3317
3318                         $isbn = $blank = '' ;
3319                         while ( ' ' == $x{0} ) {
3320                                 $blank .= ' ';
3321                                 $x = substr( $x, 1 );
3322                         }
3323                         if ( $x == '' ) { # blank isbn
3324                                 $text .= "ISBN $blank";
3325                                 continue;
3326                         }
3327                         while ( strstr( $valid, $x{0} ) != false ) {
3328                                 $isbn .= $x{0};
3329                                 $x = substr( $x, 1 );
3330                         }
3331                         $num = str_replace( '-', '', $isbn );
3332                         $num = str_replace( ' ', '', $num );
3333                         $num = str_replace( 'x', 'X', $num );
3334
3335                         if ( '' == $num ) {
3336                                 $text .= "ISBN $blank$x";
3337                         } else {
3338                                 $titleObj = Title::makeTitle( NS_SPECIAL, 'Booksources' );
3339                                 $text .= '<a href="' .
3340                                         $titleObj->escapeLocalUrl( 'isbn='.$num ) .
3341                                         "\" class=\"internal\">ISBN $isbn</a>";
3342                                 $text .= $x;
3343                         }
3344                 }
3345                 wfProfileOut( $fname );
3346                 return $text;
3347         }
3348
3349         /**
3350          * Return an HTML link for the "RFC 1234" text
3351          *
3352          * @private
3353          * @param string $text     Text to be processed
3354          * @param string $keyword  Magic keyword to use (default RFC)
3355          * @param string $urlmsg   Interface message to use (default rfcurl)
3356          * @return string
3357          */
3358         function magicRFC( $text, $keyword='RFC ', $urlmsg='rfcurl'  ) {
3359
3360                 $valid = '0123456789';
3361                 $internal = false;
3362
3363                 $a = split( $keyword, ' '.$text );
3364                 if ( count ( $a ) < 2 ) {
3365                         return $text;
3366                 }
3367                 $text = substr( array_shift( $a ), 1);
3368
3369                 /* Check if keyword is preceed by [[.
3370                  * This test is made here cause of the array_shift above
3371                  * that prevent the test to be done in the foreach.
3372                  */
3373                 if ( substr( $text, -2 ) == '[[' ) {
3374                         $internal = true;
3375                 }
3376
3377                 foreach ( $a as $x ) {
3378                         /* token might be empty if we have RFC RFC 1234 */
3379                         if ( $x=='' ) {
3380                                 $text.=$keyword;
3381                                 continue;
3382                                 }
3383
3384                         # hack: don't replace inside thumbnail title/alt
3385                         # attributes
3386                         if(preg_match('/<[^>]+(alt|title)="[^">]*$/', $text)) {
3387                                 $text .= $keyword . $x;
3388                                 continue;
3389                         }
3390
3391                         $id = $blank = '' ;
3392
3393                         /** remove and save whitespaces in $blank */
3394                         while ( $x{0} == ' ' ) {
3395                                 $blank .= ' ';
3396                                 $x = substr( $x, 1 );
3397                         }
3398
3399                         /** remove and save the rfc number in $id */
3400                         while ( strstr( $valid, $x{0} ) != false ) {
3401                                 $id .= $x{0};
3402                                 $x = substr( $x, 1 );
3403                         }
3404
3405                         if ( $id == '' ) {
3406                                 /* call back stripped spaces*/
3407                                 $text .= $keyword.$blank.$x;
3408                         } elseif( $internal ) {
3409                                 /* normal link */
3410                                 $text .= $keyword.$id.$x;
3411                         } else {
3412                                 /* build the external link*/
3413                                 $url = wfMsg( $urlmsg, $id);
3414                                 $sk =& $this->mOptions->getSkin();
3415                                 $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
3416                                 $text .= "<a href='{$url}'{$la}>{$keyword}{$id}</a>{$x}";
3417                         }
3418
3419                         /* Check if the next RFC keyword is preceed by [[ */
3420                         $internal = ( substr($x,-2) == '[[' );
3421                 }
3422                 return $text;
3423         }
3424
3425         /**
3426          * Transform wiki markup when saving a page by doing \r\n -> \n
3427          * conversion, substitting signatures, {{subst:}} templates, etc.
3428          *
3429          * @param string $text the text to transform
3430          * @param Title &$title the Title object for the current article
3431          * @param User &$user the User object describing the current user
3432          * @param ParserOptions $options parsing options
3433          * @param bool $clearState whether to clear the parser state first
3434          * @return string the altered wiki markup
3435          * @public
3436          */
3437         function preSaveTransform( $text, &$title, &$user, $options, $clearState = true ) {
3438                 $this->mOptions = $options;
3439                 $this->mTitle =& $title;
3440                 $this->mOutputType = OT_WIKI;
3441
3442                 if ( $clearState ) {
3443                         $this->clearState();
3444                 }
3445
3446                 $stripState = false;
3447                 $pairs = array(
3448                         "\r\n" => "\n",
3449                 );
3450                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3451                 $text = $this->strip( $text, $stripState, true );
3452                 $text = $this->pstPass2( $text, $stripState, $user );
3453                 $text = $this->unstrip( $text, $stripState );
3454                 $text = $this->unstripNoWiki( $text, $stripState );
3455                 return $text;
3456         }
3457
3458         /**
3459          * Pre-save transform helper function
3460          * @private
3461          */
3462         function pstPass2( $text, &$stripState, &$user ) {
3463                 global $wgContLang, $wgLocaltimezone;
3464
3465                 /* Note: This is the timestamp saved as hardcoded wikitext to
3466                  * the database, we use $wgContLang here in order to give
3467                  * everyone the same signature and use the default one rather
3468                  * than the one selected in each user's preferences.
3469                  */
3470                 if ( isset( $wgLocaltimezone ) ) {
3471                         $oldtz = getenv( 'TZ' );
3472                         putenv( 'TZ='.$wgLocaltimezone );
3473                 }
3474                 $d = $wgContLang->timeanddate( date( 'YmdHis' ), false, false) .
3475                   ' (' . date( 'T' ) . ')';
3476                 if ( isset( $wgLocaltimezone ) ) {
3477                         putenv( 'TZ='.$oldtz );
3478                 }
3479
3480                 # Variable replacement
3481                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3482                 $text = $this->replaceVariables( $text );
3483
3484                 # Strip out <nowiki> etc. added via replaceVariables
3485                 $text = $this->strip( $text, $stripState );
3486
3487                 # Signatures
3488                 $sigText = $this->getUserSig( $user );
3489                 $text = strtr( $text, array(
3490                         '~~~~~' => $d,
3491                         '~~~~' => "$sigText $d",
3492                         '~~~' => $sigText
3493                 ) );
3494
3495                 # Context links: [[|name]] and [[name (context)|]]
3496                 #
3497                 global $wgLegalTitleChars;
3498                 $tc = "[$wgLegalTitleChars]";
3499                 $np = str_replace( array( '(', ')' ), array( '', '' ), $tc ); # No parens
3500
3501                 $namespacechar = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
3502                 $conpat = "/^({$np}+) \\(({$tc}+)\\)$/";
3503
3504                 $p1 = "/\[\[({$np}+) \\(({$np}+)\\)\\|]]/";             # [[page (context)|]]
3505                 $p2 = "/\[\[\\|({$tc}+)]]/";                                    # [[|page]]
3506                 $p3 = "/\[\[(:*$namespacechar+):({$np}+)\\|]]/";                # [[namespace:page|]] and [[:namespace:page|]]
3507                 $p4 = "/\[\[(:*$namespacechar+):({$np}+) \\(({$np}+)\\)\\|]]/"; # [[ns:page (cont)|]] and [[:ns:page (cont)|]]
3508                 $context = '';
3509                 $t = $this->mTitle->getText();
3510                 if ( preg_match( $conpat, $t, $m ) ) {
3511                         $context = $m[2];
3512                 }
3513                 $text = preg_replace( $p4, '[[\\1:\\2 (\\3)|\\2]]', $text );
3514                 $text = preg_replace( $p1, '[[\\1 (\\2)|\\1]]', $text );
3515                 $text = preg_replace( $p3, '[[\\1:\\2|\\2]]', $text );
3516
3517                 if ( '' == $context ) {
3518                         $text = preg_replace( $p2, '[[\\1]]', $text );
3519                 } else {
3520                         $text = preg_replace( $p2, "[[\\1 ({$context})|\\1]]", $text );
3521                 }
3522
3523                 # Trim trailing whitespace
3524                 # MAG_END (__END__) tag allows for trailing
3525                 # whitespace to be deliberately included
3526                 $text = rtrim( $text );
3527                 $mw =& MagicWord::get( MAG_END );
3528                 $mw->matchAndRemove( $text );
3529
3530                 return $text;
3531         }
3532
3533         /**
3534          * Fetch the user's signature text, if any, and normalize to
3535          * validated, ready-to-insert wikitext.
3536          *
3537          * @param User $user
3538          * @return string
3539          * @private
3540          */
3541         function getUserSig( &$user ) {
3542                 $username = $user->getName();
3543                 $nickname = $user->getOption( 'nickname' );
3544                 $nickname = $nickname === '' ? $username : $nickname;
3545
3546                 if( $user->getBoolOption( 'fancysig' ) !== false ) {
3547                         # Sig. might contain markup; validate this
3548                         if( $this->validateSig( $nickname ) !== false ) {
3549                                 # Validated; clean up (if needed) and return it
3550                                 return $this->cleanSig( $nickname, true );
3551                         } else {
3552                                 # Failed to validate; fall back to the default
3553                                 $nickname = $username;
3554                                 wfDebug( "Parser::getUserSig: $username has bad XML tags in signature.\n" );
3555                         }
3556                 }
3557
3558                 # If we're still here, make it a link to the user page
3559                 $userpage = $user->getUserPage();
3560                 return( '[[' . $userpage->getPrefixedText() . '|' . wfEscapeWikiText( $nickname ) . ']]' );
3561         }
3562
3563         /**
3564          * Check that the user's signature contains no bad XML
3565          *
3566          * @param string $text
3567          * @return mixed An expanded string, or false if invalid.
3568          */
3569         function validateSig( $text ) {
3570                 return( wfIsWellFormedXmlFragment( $text ) ? $text : false );
3571         }
3572
3573         /**
3574          * Clean up signature text
3575          *
3576          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures
3577          * 2) Substitute all transclusions
3578          *
3579          * @param string $text
3580          * @param $parsing Whether we're cleaning (preferences save) or parsing
3581          * @return string Signature text
3582          */
3583         function cleanSig( $text, $parsing = false ) {
3584                 global $wgTitle;
3585                 $this->startExternalParse( $wgTitle, new ParserOptions(), $parsing ? OT_WIKI : OT_MSG );
3586
3587                 $substWord = MagicWord::get( MAG_SUBST );
3588                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3589                 $substText = '{{' . $substWord->getSynonym( 0 );
3590
3591                 $text = preg_replace( $substRegex, $substText, $text );
3592                 $text = preg_replace( '/~{3,5}/', '', $text );
3593                 $text = $this->replaceVariables( $text );
3594
3595                 $this->clearState();
3596                 return $text;
3597         }
3598
3599         /**
3600          * Set up some variables which are usually set up in parse()
3601          * so that an external function can call some class members with confidence
3602          * @public
3603          */
3604         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
3605                 $this->mTitle =& $title;
3606                 $this->mOptions = $options;
3607                 $this->mOutputType = $outputType;
3608                 if ( $clearState ) {
3609                         $this->clearState();
3610                 }
3611         }
3612
3613         /**
3614          * Transform a MediaWiki message by replacing magic variables.
3615          *
3616          * @param string $text the text to transform
3617          * @param ParserOptions $options  options
3618          * @return string the text with variables substituted
3619          * @public
3620          */
3621         function transformMsg( $text, $options ) {
3622                 global $wgTitle;
3623                 static $executing = false;
3624
3625                 $fname = "Parser::transformMsg";
3626
3627                 # Guard against infinite recursion
3628                 if ( $executing ) {
3629                         return $text;
3630                 }
3631                 $executing = true;
3632
3633                 wfProfileIn($fname);
3634
3635                 $this->mTitle = $wgTitle;
3636                 $this->mOptions = $options;
3637                 $this->mOutputType = OT_MSG;
3638                 $this->clearState();
3639                 $text = $this->replaceVariables( $text );
3640
3641                 $executing = false;
3642                 wfProfileOut($fname);
3643                 return $text;
3644         }
3645
3646         /**
3647          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
3648          * The callback should have the following form:
3649          *    function myParserHook( $text, $params, &$parser ) { ... }
3650          *
3651          * Transform and return $text. Use $parser for any required context, e.g. use
3652          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
3653          *
3654          * @public
3655          *
3656          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
3657          * @param mixed $callback The callback function (and object) to use for the tag
3658          *
3659          * @return The old value of the mTagHooks array associated with the hook
3660          */
3661         function setHook( $tag, $callback ) {
3662                 $tag = strtolower( $tag );
3663                 $oldVal = @$this->mTagHooks[$tag];
3664                 $this->mTagHooks[$tag] = $callback;
3665
3666                 return $oldVal;
3667         }
3668
3669         /**
3670          * Create a function, e.g. {{sum:1|2|3}}
3671          * The callback function should have the form:
3672          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
3673          *
3674          * The callback may either return the text result of the function, or an array with the text
3675          * in element 0, and a number of flags in the other elements. The names of the flags are
3676          * specified in the keys. Valid flags are:
3677          *   found                     The text returned is valid, stop processing the template. This
3678          *                             is on by default.
3679          *   nowiki                    Wiki markup in the return value should be escaped
3680          *   noparse                   Unsafe HTML tags should not be stripped, etc.
3681          *   noargs                    Don't replace triple-brace arguments in the return value
3682          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
3683          *
3684          * @public
3685          *
3686          * @param string $name The function name. Function names are case-insensitive.
3687          * @param mixed $callback The callback function (and object) to use
3688          *
3689          * @return The old callback function for this name, if any
3690          */
3691         function setFunctionHook( $name, $callback ) {
3692                 $name = strtolower( $name );
3693                 $oldVal = @$this->mFunctionHooks[$name];
3694                 $this->mFunctionHooks[$name] = $callback;
3695                 return $oldVal;
3696         }
3697
3698         /**
3699          * Replace <!--LINK--> link placeholders with actual links, in the buffer
3700          * Placeholders created in Skin::makeLinkObj()
3701          * Returns an array of links found, indexed by PDBK:
3702          *  0 - broken
3703          *  1 - normal link
3704          *  2 - stub
3705          * $options is a bit field, RLH_FOR_UPDATE to select for update
3706          */
3707         function replaceLinkHolders( &$text, $options = 0 ) {
3708                 global $wgUser;
3709                 global $wgOutputReplace;
3710
3711                 $fname = 'Parser::replaceLinkHolders';
3712                 wfProfileIn( $fname );
3713
3714                 $pdbks = array();
3715                 $colours = array();
3716                 $sk =& $this->mOptions->getSkin();
3717                 $linkCache =& LinkCache::singleton();
3718
3719                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
3720                         wfProfileIn( $fname.'-check' );
3721                         $dbr =& wfGetDB( DB_SLAVE );
3722                         $page = $dbr->tableName( 'page' );
3723                         $threshold = $wgUser->getOption('stubthreshold');
3724
3725                         # Sort by namespace
3726                         asort( $this->mLinkHolders['namespaces'] );
3727
3728                         # Generate query
3729                         $query = false;
3730                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3731                                 # Make title object
3732                                 $title = $this->mLinkHolders['titles'][$key];
3733
3734                                 # Skip invalid entries.
3735                                 # Result will be ugly, but prevents crash.
3736                                 if ( is_null( $title ) ) {
3737                                         continue;
3738                                 }
3739                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
3740
3741                                 # Check if it's a static known link, e.g. interwiki
3742                                 if ( $title->isAlwaysKnown() ) {
3743                                         $colours[$pdbk] = 1;
3744                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
3745                                         $colours[$pdbk] = 1;
3746                                         $this->mOutput->addLink( $title, $id );
3747                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
3748                                         $colours[$pdbk] = 0;
3749                                 } else {
3750                                         # Not in the link cache, add it to the query
3751                                         if ( !isset( $current ) ) {
3752                                                 $current = $ns;
3753                                                 $query =  "SELECT page_id, page_namespace, page_title";
3754                                                 if ( $threshold > 0 ) {
3755                                                         $query .= ', page_len, page_is_redirect';
3756                                                 }
3757                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
3758                                         } elseif ( $current != $ns ) {
3759                                                 $current = $ns;
3760                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
3761                                         } else {
3762                                                 $query .= ', ';
3763                                         }
3764
3765                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
3766                                 }
3767                         }
3768                         if ( $query ) {
3769                                 $query .= '))';
3770                                 if ( $options & RLH_FOR_UPDATE ) {
3771                                         $query .= ' FOR UPDATE';
3772                                 }
3773
3774                                 $res = $dbr->query( $query, $fname );
3775
3776                                 # Fetch data and form into an associative array
3777                                 # non-existent = broken
3778                                 # 1 = known
3779                                 # 2 = stub
3780                                 while ( $s = $dbr->fetchObject($res) ) {
3781                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
3782                                         $pdbk = $title->getPrefixedDBkey();
3783                                         $linkCache->addGoodLinkObj( $s->page_id, $title );
3784                                         $this->mOutput->addLink( $title, $s->page_id );
3785
3786                                         if ( $threshold >  0 ) {
3787                                                 $size = $s->page_len;
3788                                                 if ( $s->page_is_redirect || $s->page_namespace != 0 || $size >= $threshold ) {
3789                                                         $colours[$pdbk] = 1;
3790                                                 } else {
3791                                                         $colours[$pdbk] = 2;
3792                                                 }
3793                                         } else {
3794                                                 $colours[$pdbk] = 1;
3795                                         }
3796                                 }
3797                         }
3798                         wfProfileOut( $fname.'-check' );
3799
3800                         # Construct search and replace arrays
3801                         wfProfileIn( $fname.'-construct' );
3802                         $wgOutputReplace = array();
3803                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3804                                 $pdbk = $pdbks[$key];
3805                                 $searchkey = "<!--LINK $key-->";
3806                                 $title = $this->mLinkHolders['titles'][$key];
3807                                 if ( empty( $colours[$pdbk] ) ) {
3808                                         $linkCache->addBadLinkObj( $title );
3809                                         $colours[$pdbk] = 0;
3810                                         $this->mOutput->addLink( $title, 0 );
3811                                         $wgOutputReplace[$searchkey] = $sk->makeBrokenLinkObj( $title,
3812                                                                         $this->mLinkHolders['texts'][$key],
3813                                                                         $this->mLinkHolders['queries'][$key] );
3814                                 } elseif ( $colours[$pdbk] == 1 ) {
3815                                         $wgOutputReplace[$searchkey] = $sk->makeKnownLinkObj( $title,
3816                                                                         $this->mLinkHolders['texts'][$key],
3817                                                                         $this->mLinkHolders['queries'][$key] );
3818                                 } elseif ( $colours[$pdbk] == 2 ) {
3819                                         $wgOutputReplace[$searchkey] = $sk->makeStubLinkObj( $title,
3820                                                                         $this->mLinkHolders['texts'][$key],
3821                                                                         $this->mLinkHolders['queries'][$key] );
3822                                 }
3823                         }
3824                         wfProfileOut( $fname.'-construct' );
3825
3826                         # Do the thing
3827                         wfProfileIn( $fname.'-replace' );
3828
3829                         $text = preg_replace_callback(
3830                                 '/(<!--LINK .*?-->)/',
3831                                 "wfOutputReplaceMatches",
3832                                 $text);
3833
3834                         wfProfileOut( $fname.'-replace' );
3835                 }
3836
3837                 # Now process interwiki link holders
3838                 # This is quite a bit simpler than internal links
3839                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
3840                         wfProfileIn( $fname.'-interwiki' );
3841                         # Make interwiki link HTML
3842                         $wgOutputReplace = array();
3843                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
3844                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
3845                                 $wgOutputReplace[$key] = $sk->makeLinkObj( $title, $link );
3846                         }
3847
3848                         $text = preg_replace_callback(
3849                                 '/<!--IWLINK (.*?)-->/',
3850                                 "wfOutputReplaceMatches",
3851                                 $text );
3852                         wfProfileOut( $fname.'-interwiki' );
3853                 }
3854
3855                 wfProfileOut( $fname );
3856                 return $colours;
3857         }
3858
3859         /**
3860          * Replace <!--LINK--> link placeholders with plain text of links
3861          * (not HTML-formatted).
3862          * @param string $text
3863          * @return string
3864          */
3865         function replaceLinkHoldersText( $text ) {
3866                 $fname = 'Parser::replaceLinkHoldersText';
3867                 wfProfileIn( $fname );
3868
3869                 $text = preg_replace_callback(
3870                         '/<!--(LINK|IWLINK) (.*?)-->/',
3871                         array( &$this, 'replaceLinkHoldersTextCallback' ),
3872                         $text );
3873
3874                 wfProfileOut( $fname );
3875                 return $text;
3876         }
3877
3878         /**
3879          * @param array $matches
3880          * @return string
3881          * @private
3882          */
3883         function replaceLinkHoldersTextCallback( $matches ) {
3884                 $type = $matches[1];
3885                 $key  = $matches[2];
3886                 if( $type == 'LINK' ) {
3887                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
3888                                 return $this->mLinkHolders['texts'][$key];
3889                         }
3890                 } elseif( $type == 'IWLINK' ) {
3891                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
3892                                 return $this->mInterwikiLinkHolders['texts'][$key];
3893                         }
3894                 }
3895                 return $matches[0];
3896         }
3897
3898         /**
3899          * Renders an image gallery from a text with one line per image.
3900          * text labels may be given by using |-style alternative text. E.g.
3901          *   Image:one.jpg|The number "1"
3902          *   Image:tree.jpg|A tree
3903          * given as text will return the HTML of a gallery with two images,
3904          * labeled 'The number "1"' and
3905          * 'A tree'.
3906          */
3907         function renderImageGallery( $text ) {
3908                 # Setup the parser
3909                 $parserOptions = new ParserOptions;
3910                 $localParser = new Parser();
3911
3912                 $ig = new ImageGallery();
3913                 $ig->setShowBytes( false );
3914                 $ig->setShowFilename( false );
3915                 $ig->setParsing();
3916                 $lines = explode( "\n", $text );
3917
3918                 foreach ( $lines as $line ) {
3919                         # match lines like these:
3920                         # Image:someimage.jpg|This is some image
3921                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
3922                         # Skip empty lines
3923                         if ( count( $matches ) == 0 ) {
3924                                 continue;
3925                         }
3926                         $nt =& Title::newFromText( $matches[1] );
3927                         if( is_null( $nt ) ) {
3928                                 # Bogus title. Ignore these so we don't bomb out later.
3929                                 continue;
3930                         }
3931                         if ( isset( $matches[3] ) ) {
3932                                 $label = $matches[3];
3933                         } else {
3934                                 $label = '';
3935                         }
3936
3937                         $pout = $localParser->parse( $label , $this->mTitle, $parserOptions );
3938                         $html = $pout->getText();
3939
3940                         $ig->add( new Image( $nt ), $html );
3941
3942                         # Only add real images (bug #5586)
3943                         if ( $nt->getNamespace() == NS_IMAGE ) {
3944                                 $this->mOutput->addImage( $nt->getDBkey() );
3945                         }
3946
3947                         # Register links with the parent parser
3948                         foreach( $pout->getLinks() as $ns => $keys ) {
3949                                 foreach( $keys as $dbk => $id )
3950                                         $this->mOutput->addLink( Title::makeTitle( $ns, $dbk ), $id );
3951                         }
3952
3953                 }
3954                 return $ig->toHTML();
3955         }
3956
3957         /**
3958          * Parse image options text and use it to make an image
3959          */
3960         function makeImage( &$nt, $options ) {
3961                 global $wgUseImageResize;
3962
3963                 $align = '';
3964
3965                 # Check if the options text is of the form "options|alt text"
3966                 # Options are:
3967                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
3968                 #  * left               no resizing, just left align. label is used for alt= only
3969                 #  * right              same, but right aligned
3970                 #  * none               same, but not aligned
3971                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
3972                 #  * center             center the image
3973                 #  * framed             Keep original image size, no magnify-button.
3974
3975                 $part = explode( '|', $options);
3976
3977                 $mwThumb  =& MagicWord::get( MAG_IMG_THUMBNAIL );
3978                 $mwManualThumb =& MagicWord::get( MAG_IMG_MANUALTHUMB );
3979                 $mwLeft   =& MagicWord::get( MAG_IMG_LEFT );
3980                 $mwRight  =& MagicWord::get( MAG_IMG_RIGHT );
3981                 $mwNone   =& MagicWord::get( MAG_IMG_NONE );
3982                 $mwWidth  =& MagicWord::get( MAG_IMG_WIDTH );
3983                 $mwCenter =& MagicWord::get( MAG_IMG_CENTER );
3984                 $mwFramed =& MagicWord::get( MAG_IMG_FRAMED );
3985                 $caption = '';
3986
3987                 $width = $height = $framed = $thumb = false;
3988                 $manual_thumb = '' ;
3989
3990                 foreach( $part as $key => $val ) {
3991                         if ( $wgUseImageResize && ! is_null( $mwThumb->matchVariableStartToEnd($val) ) ) {
3992                                 $thumb=true;
3993                         } elseif ( ! is_null( $match = $mwManualThumb->matchVariableStartToEnd($val) ) ) {
3994                                 # use manually specified thumbnail
3995                                 $thumb=true;
3996                                 $manual_thumb = $match;
3997                         } elseif ( ! is_null( $mwRight->matchVariableStartToEnd($val) ) ) {
3998                                 # remember to set an alignment, don't render immediately
3999                                 $align = 'right';
4000                         } elseif ( ! is_null( $mwLeft->matchVariableStartToEnd($val) ) ) {
4001                                 # remember to set an alignment, don't render immediately
4002                                 $align = 'left';
4003                         } elseif ( ! is_null( $mwCenter->matchVariableStartToEnd($val) ) ) {
4004                                 # remember to set an alignment, don't render immediately
4005                                 $align = 'center';
4006                         } elseif ( ! is_null( $mwNone->matchVariableStartToEnd($val) ) ) {
4007                                 # remember to set an alignment, don't render immediately
4008                                 $align = 'none';
4009                         } elseif ( $wgUseImageResize && ! is_null( $match = $mwWidth->matchVariableStartToEnd($val) ) ) {
4010                                 wfDebug( "MAG_IMG_WIDTH match: $match\n" );
4011                                 # $match is the image width in pixels
4012                                 if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $match, $m ) ) {
4013                                         $width = intval( $m[1] );
4014                                         $height = intval( $m[2] );
4015                                 } else {
4016                                         $width = intval($match);
4017                                 }
4018                         } elseif ( ! is_null( $mwFramed->matchVariableStartToEnd($val) ) ) {
4019                                 $framed=true;
4020                         } else {
4021                                 $caption = $val;
4022                         }
4023                 }
4024                 # Strip bad stuff out of the alt text
4025                 $alt = $this->replaceLinkHoldersText( $caption );
4026
4027                 # make sure there are no placeholders in thumbnail attributes
4028                 # that are later expanded to html- so expand them now and
4029                 # remove the tags
4030                 $alt = $this->unstrip($alt, $this->mStripState);
4031                 $alt = Sanitizer::stripAllTags( $alt );
4032
4033                 # Linker does the rest
4034                 $sk =& $this->mOptions->getSkin();
4035                 return $sk->makeImageLinkObj( $nt, $caption, $alt, $align, $width, $height, $framed, $thumb, $manual_thumb );
4036         }
4037
4038         /**
4039          * Set a flag in the output object indicating that the content is dynamic and
4040          * shouldn't be cached.
4041          */
4042         function disableCache() {
4043                 wfDebug( "Parser output marked as uncacheable.\n" );
4044                 $this->mOutput->mCacheTime = -1;
4045         }
4046
4047         /**#@+
4048          * Callback from the Sanitizer for expanding items found in HTML attribute
4049          * values, so they can be safely tested and escaped.
4050          * @param string $text
4051          * @param array $args
4052          * @return string
4053          * @private
4054          */
4055         function attributeStripCallback( &$text, $args ) {
4056                 $text = $this->replaceVariables( $text, $args );
4057                 $text = $this->unstripForHTML( $text );
4058                 return $text;
4059         }
4060
4061         function unstripForHTML( $text ) {
4062                 $text = $this->unstrip( $text, $this->mStripState );
4063                 $text = $this->unstripNoWiki( $text, $this->mStripState );
4064                 return $text;
4065         }
4066         /**#@-*/
4067
4068         /**#@+
4069          * Accessor/mutator
4070          */
4071         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4072         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4073         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4074         /**#@-*/
4075
4076         /**#@+
4077          * Accessor
4078          */
4079         function getTags() { return array_keys( $this->mTagHooks ); }
4080         /**#@-*/
4081 }
4082
4083 /**
4084  * @todo document
4085  * @package MediaWiki
4086  */
4087 class ParserOutput
4088 {
4089         var $mText,             # The output text
4090                 $mLanguageLinks,    # List of the full text of language links, in the order they appear
4091                 $mCategories,       # Map of category names to sort keys
4092                 $mContainsOldMagic, # Boolean variable indicating if the input contained variables like {{CURRENTDAY}}
4093                 $mCacheTime,        # Time when this object was generated, or -1 for uncacheable. Used in ParserCache.
4094                 $mVersion,          # Compatibility check
4095                 $mTitleText,        # title text of the chosen language variant
4096                 $mLinks,            # 2-D map of NS/DBK to ID for the links in the document. ID=zero for broken.
4097                 $mTemplates,        # 2-D map of NS/DBK to ID for the template references. ID=zero for broken.
4098                 $mImages,           # DB keys of the images used, in the array key only
4099                 $mExternalLinks,    # External link URLs, in the key only
4100                 $mHTMLtitle,            # Display HTML title
4101                 $mSubtitle,                     # Additional subtitle
4102                 $mNewSection;           # Show a new section link?
4103
4104         function ParserOutput( $text = '', $languageLinks = array(), $categoryLinks = array(),
4105                 $containsOldMagic = false, $titletext = '' )
4106         {
4107                 $this->mText = $text;
4108                 $this->mLanguageLinks = $languageLinks;
4109                 $this->mCategories = $categoryLinks;
4110                 $this->mContainsOldMagic = $containsOldMagic;
4111                 $this->mCacheTime = '';
4112                 $this->mVersion = MW_PARSER_VERSION;
4113                 $this->mTitleText = $titletext;
4114                 $this->mLinks = array();
4115                 $this->mTemplates = array();
4116                 $this->mImages = array();
4117                 $this->mExternalLinks = array();
4118                 $this->mHTMLtitle = "" ;
4119                 $this->mSubtitle = "" ;
4120                 $this->mNewSection = false;
4121         }
4122
4123         function getText()                   { return $this->mText; }
4124         function &getLanguageLinks()          { return $this->mLanguageLinks; }
4125         function getCategoryLinks()          { return array_keys( $this->mCategories ); }
4126         function &getCategories()            { return $this->mCategories; }
4127         function getCacheTime()              { return $this->mCacheTime; }
4128         function getTitleText()              { return $this->mTitleText; }
4129         function &getLinks()                 { return $this->mLinks; }
4130         function &getTemplates()             { return $this->mTemplates; }
4131         function &getImages()                { return $this->mImages; }
4132         function &getExternalLinks()         { return $this->mExternalLinks; }
4133
4134         function containsOldMagic()          { return $this->mContainsOldMagic; }
4135         function setText( $text )            { return wfSetVar( $this->mText, $text ); }
4136         function setLanguageLinks( $ll )     { return wfSetVar( $this->mLanguageLinks, $ll ); }
4137         function setCategoryLinks( $cl )     { return wfSetVar( $this->mCategories, $cl ); }
4138         function setContainsOldMagic( $com ) { return wfSetVar( $this->mContainsOldMagic, $com ); }
4139         function setCacheTime( $t )          { return wfSetVar( $this->mCacheTime, $t ); }
4140         function setTitleText( $t )          { return wfSetVar ($this->mTitleText, $t); }
4141
4142         function addCategory( $c, $sort )    { $this->mCategories[$c] = $sort; }
4143         function addImage( $name )           { $this->mImages[$name] = 1; }
4144         function addLanguageLink( $t )       { $this->mLanguageLinks[] = $t; }
4145         function addExternalLink( $url )     { $this->mExternalLinks[$url] = 1; }
4146
4147         function setNewSection( $value ) {
4148                 $this->mNewSection = (bool)$value;
4149         }
4150         function getNewSection() {
4151                 return (bool)$this->mNewSection;
4152         }
4153
4154         function addLink( $title, $id ) {
4155                 $ns = $title->getNamespace();
4156                 $dbk = $title->getDBkey();
4157                 if ( !isset( $this->mLinks[$ns] ) ) {
4158                         $this->mLinks[$ns] = array();
4159                 }
4160                 $this->mLinks[$ns][$dbk] = $id;
4161         }
4162
4163         function addTemplate( $title, $id ) {
4164                 $ns = $title->getNamespace();
4165                 $dbk = $title->getDBkey();
4166                 if ( !isset( $this->mTemplates[$ns] ) ) {
4167                         $this->mTemplates[$ns] = array();
4168                 }
4169                 $this->mTemplates[$ns][$dbk] = $id;
4170         }
4171
4172         /**
4173          * Return true if this cached output object predates the global or
4174          * per-article cache invalidation timestamps, or if it comes from
4175          * an incompatible older version.
4176          *
4177          * @param string $touched the affected article's last touched timestamp
4178          * @return bool
4179          * @public
4180          */
4181         function expired( $touched ) {
4182                 global $wgCacheEpoch;
4183                 return $this->getCacheTime() == -1 || // parser says it's uncacheable
4184                        $this->getCacheTime() < $touched ||
4185                        $this->getCacheTime() <= $wgCacheEpoch ||
4186                        !isset( $this->mVersion ) ||
4187                        version_compare( $this->mVersion, MW_PARSER_VERSION, "lt" );
4188         }
4189 }
4190
4191 /**
4192  * Set options of the Parser
4193  * @todo document
4194  * @package MediaWiki
4195  */
4196 class ParserOptions
4197 {
4198         # All variables are private
4199         var $mUseTeX;                    # Use texvc to expand <math> tags
4200         var $mUseDynamicDates;           # Use DateFormatter to format dates
4201         var $mInterwikiMagic;            # Interlanguage links are removed and returned in an array
4202         var $mAllowExternalImages;       # Allow external images inline
4203         var $mAllowExternalImagesFrom;   # If not, any exception?
4204         var $mSkin;                      # Reference to the preferred skin
4205         var $mDateFormat;                # Date format index
4206         var $mEditSection;               # Create "edit section" links
4207         var $mNumberHeadings;            # Automatically number headings
4208         var $mAllowSpecialInclusion;     # Allow inclusion of special pages
4209         var $mTidy;                      # Ask for tidy cleanup
4210         var $mInterfaceMessage;          # Which lang to call for PLURAL and GRAMMAR
4211
4212         function getUseTeX()                        { return $this->mUseTeX; }
4213         function getUseDynamicDates()               { return $this->mUseDynamicDates; }
4214         function getInterwikiMagic()                { return $this->mInterwikiMagic; }
4215         function getAllowExternalImages()           { return $this->mAllowExternalImages; }
4216         function getAllowExternalImagesFrom()       { return $this->mAllowExternalImagesFrom; }
4217         function &getSkin()                         { return $this->mSkin; }
4218         function getDateFormat()                    { return $this->mDateFormat; }
4219         function getEditSection()                   { return $this->mEditSection; }
4220         function getNumberHeadings()                { return $this->mNumberHeadings; }
4221         function getAllowSpecialInclusion()         { return $this->mAllowSpecialInclusion; }
4222         function getTidy()                          { return $this->mTidy; }
4223         function getInterfaceMessage()              { return $this->mInterfaceMessage; }
4224
4225         function setUseTeX( $x )                    { return wfSetVar( $this->mUseTeX, $x ); }
4226         function setUseDynamicDates( $x )           { return wfSetVar( $this->mUseDynamicDates, $x ); }
4227         function setInterwikiMagic( $x )            { return wfSetVar( $this->mInterwikiMagic, $x ); }
4228         function setAllowExternalImages( $x )       { return wfSetVar( $this->mAllowExternalImages, $x ); }
4229         function setAllowExternalImagesFrom( $x )   { return wfSetVar( $this->mAllowExternalImagesFrom, $x ); }
4230         function setDateFormat( $x )                { return wfSetVar( $this->mDateFormat, $x ); }
4231         function setEditSection( $x )               { return wfSetVar( $this->mEditSection, $x ); }
4232         function setNumberHeadings( $x )            { return wfSetVar( $this->mNumberHeadings, $x ); }
4233         function setAllowSpecialInclusion( $x )     { return wfSetVar( $this->mAllowSpecialInclusion, $x ); }
4234         function setTidy( $x )                      { return wfSetVar( $this->mTidy, $x); }
4235         function setSkin( &$x ) { $this->mSkin =& $x; }
4236         function setInterfaceMessage( $x )          { return wfSetVar( $this->mInterfaceMessage, $x); }
4237
4238         function ParserOptions() {
4239                 global $wgUser;
4240                 $this->initialiseFromUser( $wgUser );
4241         }
4242
4243         /**
4244          * Get parser options
4245          * @static
4246          */
4247         function newFromUser( &$user ) {
4248                 $popts = new ParserOptions;
4249                 $popts->initialiseFromUser( $user );
4250                 return $popts;
4251         }
4252
4253         /** Get user options */
4254         function initialiseFromUser( &$userInput ) {
4255                 global $wgUseTeX, $wgUseDynamicDates, $wgInterwikiMagic, $wgAllowExternalImages;
4256                 global $wgAllowExternalImagesFrom, $wgAllowSpecialInclusion;
4257                 $fname = 'ParserOptions::initialiseFromUser';
4258                 wfProfileIn( $fname );
4259                 if ( !$userInput ) {
4260                         $user = new User;
4261                         $user->setLoaded( true );
4262                 } else {
4263                         $user =& $userInput;
4264                 }
4265
4266                 $this->mUseTeX = $wgUseTeX;
4267                 $this->mUseDynamicDates = $wgUseDynamicDates;
4268                 $this->mInterwikiMagic = $wgInterwikiMagic;
4269                 $this->mAllowExternalImages = $wgAllowExternalImages;
4270                 $this->mAllowExternalImagesFrom = $wgAllowExternalImagesFrom;
4271                 wfProfileIn( $fname.'-skin' );
4272                 $this->mSkin =& $user->getSkin();
4273                 wfProfileOut( $fname.'-skin' );
4274                 $this->mDateFormat = $user->getOption( 'date' );
4275                 $this->mEditSection = true;
4276                 $this->mNumberHeadings = $user->getOption( 'numberheadings' );
4277                 $this->mAllowSpecialInclusion = $wgAllowSpecialInclusion;
4278                 $this->mTidy = false;
4279                 $this->mInterfaceMessage = false;
4280                 wfProfileOut( $fname );
4281         }
4282 }
4283
4284 /**
4285  * Callback function used by Parser::replaceLinkHolders()
4286  * to substitute link placeholders.
4287  */
4288 function &wfOutputReplaceMatches( $matches ) {
4289         global $wgOutputReplace;
4290         return $wgOutputReplace[$matches[1]];
4291 }
4292
4293 /**
4294  * Return the total number of articles
4295  */
4296 function wfNumberOfArticles() {
4297         global $wgNumberOfArticles;
4298
4299         wfLoadSiteStats();
4300         return $wgNumberOfArticles;
4301 }
4302
4303 /**
4304  * Return the number of files
4305  */
4306 function wfNumberOfFiles() {
4307         $fname = 'wfNumberOfFiles';
4308
4309         wfProfileIn( $fname );
4310         $dbr =& wfGetDB( DB_SLAVE );
4311         $numImages = $dbr->selectField('site_stats', 'ss_images', array(), $fname );
4312         wfProfileOut( $fname );
4313
4314         return $numImages;
4315 }
4316
4317 /**
4318  * Return the number of user accounts
4319  * @return integer
4320  */
4321 function wfNumberOfUsers() {
4322         wfProfileIn( 'wfNumberOfUsers' );
4323         $dbr =& wfGetDB( DB_SLAVE );
4324         $count = $dbr->selectField( 'site_stats', 'ss_users', array(), 'wfNumberOfUsers' );
4325         wfProfileOut( 'wfNumberOfUsers' );
4326         return (int)$count;
4327 }
4328
4329 /**
4330  * Return the total number of pages
4331  * @return integer
4332  */
4333 function wfNumberOfPages() {
4334         wfProfileIn( 'wfNumberOfPages' );
4335         $dbr =& wfGetDB( DB_SLAVE );
4336         $count = $dbr->selectField( 'site_stats', 'ss_total_pages', array(), 'wfNumberOfPages' );
4337         wfProfileOut( 'wfNumberOfPages' );
4338         return (int)$count;
4339 }
4340
4341 /**
4342  * Get various statistics from the database
4343  * @private
4344  */
4345 function wfLoadSiteStats() {
4346         global $wgNumberOfArticles, $wgTotalViews, $wgTotalEdits;
4347         $fname = 'wfLoadSiteStats';
4348
4349         if ( -1 != $wgNumberOfArticles ) return;
4350         $dbr =& wfGetDB( DB_SLAVE );
4351         $s = $dbr->selectRow( 'site_stats',
4352                 array( 'ss_total_views', 'ss_total_edits', 'ss_good_articles' ),
4353                 array( 'ss_row_id' => 1 ), $fname
4354         );
4355
4356         if ( $s === false ) {
4357                 return;
4358         } else {
4359                 $wgTotalViews = $s->ss_total_views;
4360                 $wgTotalEdits = $s->ss_total_edits;
4361                 $wgNumberOfArticles = $s->ss_good_articles;
4362         }
4363 }
4364
4365 /**
4366  * Escape html tags
4367  * Basically replacing " > and < with HTML entities ( &quot;, &gt;, &lt;)
4368  *
4369  * @param $in String: text that might contain HTML tags.
4370  * @return string Escaped string
4371  */
4372 function wfEscapeHTMLTagsOnly( $in ) {
4373         return str_replace(
4374                 array( '"', '>', '<' ),
4375                 array( '&quot;', '&gt;', '&lt;' ),
4376                 $in );
4377 }
4378
4379 ?>