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, but excluding newlines
  47 define( 'EXT_LINK_TEXT_CLASS', '[^\]\\x0a\\x0d]' );
  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',  '/\[(\b(' . wfUrlProtocols() . ')'.
  51         EXT_LINK_URL_CLASS.'+) *('.EXT_LINK_TEXT_CLASS.'*?)\]/S' );
  52 define( 'EXT_IMAGE_REGEX',
  53         '/^('.HTTP_PROTOCOLS.')'.  # Protocol
  54         '('.EXT_LINK_URL_CLASS.'+)\\/'.  # Hostname and path
  55         '('.EXT_IMAGE_FNAME_CLASS.'+)\\.((?i)'.EXT_IMAGE_EXTENSIONS.')$/S' # Filename
  56 );
  57
  58 // State constants for the definition list colon extraction
  59 define( 'MW_COLON_STATE_TEXT', 0 );
  60 define( 'MW_COLON_STATE_TAG', 1 );
  61 define( 'MW_COLON_STATE_TAGSTART', 2 );
  62 define( 'MW_COLON_STATE_CLOSETAG', 3 );
  63 define( 'MW_COLON_STATE_TAGSLASH', 4 );
  64 define( 'MW_COLON_STATE_COMMENT', 5 );
  65 define( 'MW_COLON_STATE_COMMENTDASH', 6 );
  66 define( 'MW_COLON_STATE_COMMENTDASHDASH', 7 );
  67
  68 /**
  69  * PHP Parser
  70  *
  71  * Processes wiki markup
  72  *
  73  * <pre>
  74  * There are three main entry points into the Parser class:
  75  * parse()
  76  *   produces HTML output
  77  * preSaveTransform().
  78  *   produces altered wiki markup.
  79  * transformMsg()
  80  *   performs brace substitution on MediaWiki messages
  81  *
  82  * Globals used:
  83  *    objects:   $wgLang, $wgContLang
  84  *
  85  * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
  86  *
  87  * settings:
  88  *  $wgUseTex*, $wgUseDynamicDates*, $wgInterwikiMagic*,
  89  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  90  *  $wgLocaltimezone, $wgAllowSpecialInclusion*
  91  *
  92  *  * only within ParserOptions
  93  * </pre>
  94  *
  95  * @package MediaWiki
  96  */
  97 class Parser
  98 {
  99         /**#@+
 100          * @private
 101          */
 102         # Persistent:
 103         var $mTagHooks, $mFunctionHooks;
 104
 105         # Cleared with clearState():
 106         var $mOutput, $mAutonumber, $mDTopen, $mStripState = array();
 107         var $mVariables, $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 108         var $mInterwikiLinkHolders, $mLinkHolders, $mUniqPrefix;
 109         var $mTemplates,        // cache of already loaded templates, avoids
 110                                 // multiple SQL queries for the same string
 111             $mTemplatePath;     // stores an unsorted hash of all the templates already loaded
 112                                 // in this path. Used for loop detection.
 113
 114         # Temporary
 115         # These are variables reset at least once per parse regardless of $clearState
 116         var $mOptions,      // ParserOptions object
 117                 $mTitle,        // Title context, used for self-link rendering and similar things
 118                 $mOutputType,   // Output type, one of the OT_xxx constants
 119                 $mRevisionId;   // ID to display in {{REVISIONID}} tags
 120
 121         /**#@-*/
 122
 123         /**
 124          * Constructor
 125          *
 126          * @public
 127          */
 128         function Parser() {
 129                 $this->mTagHooks = array();
 130                 $this->mFunctionHooks = array();
 131                 $this->mFunctionSynonymsValid = true;
 132                 $this->mFunctionSynonyms = array();
 133                 $this->clearState();
 134                 $this->setHook( 'pre', array( $this, 'renderPreTag' ) );
 135         }
 136
 137         /**
 138          * Clear Parser state
 139          *
 140          * @private
 141          */
 142         function clearState() {
 143                 $this->mOutput = new ParserOutput;
 144                 $this->mAutonumber = 0;
 145                 $this->mLastSection = '';
 146                 $this->mDTopen = false;
 147                 $this->mVariables = false;
 148                 $this->mIncludeCount = array();
 149                 $this->mStripState = array();
 150                 $this->mArgStack = array();
 151                 $this->mInPre = false;
 152                 $this->mInterwikiLinkHolders = array(
 153                         'texts' => array(),
 154                         'titles' => array()
 155                 );
 156                 $this->mLinkHolders = array(
 157                         'namespaces' => array(),
 158                         'dbkeys' => array(),
 159                         'queries' => array(),
 160                         'texts' => array(),
 161                         'titles' => array()
 162                 );
 163                 $this->mRevisionId = null;
 164
 165                 /**
 166                  * Prefix for temporary replacement strings for the multipass parser.
 167                  * \x07 should never appear in input as it's disallowed in XML.
 168                  * Using it at the front also gives us a little extra robustness
 169                  * since it shouldn't match when butted up against identifier-like
 170                  * string constructs.
 171                  */
 172                 $this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 173
 174                 # Clear these on every parse, bug 4549
 175                 $this->mTemplates = array();
 176                 $this->mTemplatePath = array();
 177
 178                 $this->mShowToc = true;
 179                 $this->mForceTocPosition = false;
 180
 181                 # This doesn't actually clear anything, but it seemed like a nice place to put it
 182                 # Fill the function cache, if it needs filling
 183                 $this->refreshFunctionCache();
 184
 185                 wfRunHooks( 'ParserClearState', array( &$this ) );
 186         }
 187
 188         /**
 189          * Accessor for mUniqPrefix.
 190          *
 191          * @public
 192          */
 193         function UniqPrefix() {
 194                 return $this->mUniqPrefix;
 195         }
 196
 197         /**
 198          * Convert wikitext to HTML
 199          * Do not call this function recursively.
 200          *
 201          * @private
 202          * @param string $text Text we want to parse
 203          * @param Title &$title A title object
 204          * @param array $options
 205          * @param boolean $linestart
 206          * @param boolean $clearState
 207          * @param int $revid number to pass in {{REVISIONID}}
 208          * @return ParserOutput a ParserOutput
 209          */
 210         function parse( $text, &$title, $options, $linestart = true, $clearState = true, $revid = null ) {
 211                 /**
 212                  * First pass--just handle <nowiki> sections, pass the rest off
 213                  * to internalParse() which does all the real work.
 214                  */
 215
 216                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
 217                 $fname = 'Parser::parse';
 218                 wfProfileIn( $fname );
 219
 220                 if ( $clearState ) {
 221                         $this->clearState();
 222                 }
 223
 224                 $this->mOptions = $options;
 225                 $this->mTitle =& $title;
 226                 $this->mRevisionId = $revid;
 227                 $this->mOutputType = OT_HTML;
 228
 229                 //$text = $this->strip( $text, $this->mStripState );
 230                 // VOODOO MAGIC FIX! Sometimes the above segfaults in PHP5.
 231                 $x =& $this->mStripState;
 232
 233                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$x ) );
 234                 $text = $this->strip( $text, $x );
 235                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$x ) );
 236
 237                 # Hook to suspend the parser in this state
 238                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$x ) ) ) {
 239                         wfProfileOut( $fname );
 240                         return $text ;
 241                 }
 242
 243                 $text = $this->internalParse( $text );
 244
 245                 $text = $this->unstrip( $text, $this->mStripState );
 246
 247                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 248                 $fixtags = array(
 249                         # french spaces, last one Guillemet-left
 250                         # only if there is something before the space
 251                         '/(.) (?=\\?|:|;|!|\\302\\273)/' => '\\1&nbsp;\\2',
 252                         # french spaces, Guillemet-right
 253                         '/(\\302\\253) /' => '\\1&nbsp;',
 254                 );
 255                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 256
 257                 # only once and last
 258                 $text = $this->doBlockLevels( $text, $linestart );
 259
 260                 $this->replaceLinkHolders( $text );
 261
 262                 # the position of the parserConvert() call should not be changed. it
 263                 # assumes that the links are all replaced and the only thing left
 264                 # is the <nowiki> mark.
 265                 # Side-effects: this calls $this->mOutput->setTitleText()
 266                 $text = $wgContLang->parserConvert( $text, $this );
 267
 268                 $text = $this->unstripNoWiki( $text, $this->mStripState );
 269
 270                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 271
 272                 $text = Sanitizer::normalizeCharReferences( $text );
 273
 274                 if (($wgUseTidy and $this->mOptions->mTidy) or $wgAlwaysUseTidy) {
 275                         $text = Parser::tidy($text);
 276                 } else {
 277                         # attempt to sanitize at least some nesting problems
 278                         # (bug #2702 and quite a few others)
 279                         $tidyregs = array(
 280                                 # ''Something [http://www.cool.com cool''] -->
 281                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 282                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 283                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 284                                 # fix up an anchor inside another anchor, only
 285                                 # at least for a single single nested link (bug 3695)
 286                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 287                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 288                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 289                                 # contains a div, so fix it up here; replace
 290                                 # div with escaped text
 291                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 292                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 293                                 # remove empty italic or bold tag pairs, some
 294                                 # introduced by rules above
 295                                 '/<([bi])><\/\\1>/' => ''
 296                         );
 297
 298                         $text = preg_replace(
 299                                 array_keys( $tidyregs ),
 300                                 array_values( $tidyregs ),
 301                                 $text );
 302                 }
 303
 304                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 305
 306                 $this->mOutput->setText( $text );
 307                 wfProfileOut( $fname );
 308
 309                 return $this->mOutput;
 310         }
 311
 312         /**
 313          * Get a random string
 314          *
 315          * @private
 316          * @static
 317          */
 318         function getRandomString() {
 319                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 320         }
 321
 322         function &getTitle() { return $this->mTitle; }
 323         function getOptions() { return $this->mOptions; }
 324
 325         /**
 326          * Replaces all occurrences of HTML-style comments and the given tags
 327          * in the text with a random marker and returns teh next text. The output
 328          * parameter $matches will be an associative array filled with data in
 329          * the form:
 330          *   'UNIQ-xxxxx' => array(
 331          *     'element',
 332          *     'tag content',
 333          *     array( 'param' => 'x' ),
 334          *     '<element param="x">tag content</element>' ) )
 335          *
 336          * @param $elements list of element names. Comments are always extracted.
 337          * @param $text Source text string.
 338          * @param $uniq_prefix
 339          *
 340          * @private
 341          * @static
 342          */
 343         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 344                 $rand = Parser::getRandomString();
 345                 $n = 1;
 346                 $stripped = '';
 347                 $matches = array();
 348
 349                 $taglist = implode( '|', $elements );
 350                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?>)|<(!--)/i";
 351
 352                 while ( '' != $text ) {
 353                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 354                         $stripped .= $p[0];
 355                         if( count( $p ) < 5 ) {
 356                                 break;
 357                         }
 358                         if( count( $p ) > 5 ) {
 359                                 // comment
 360                                 $element    = $p[4];
 361                                 $attributes = '';
 362                                 $close      = '';
 363                                 $inside     = $p[5];
 364                         } else {
 365                                 // tag
 366                                 $element    = $p[1];
 367                                 $attributes = $p[2];
 368                                 $close      = $p[3];
 369                                 $inside     = $p[4];
 370                         }
 371
 372                         $marker = "$uniq_prefix-$element-$rand" . sprintf('%08X', $n++) . '-QINU';
 373                         $stripped .= $marker;
 374
 375                         if ( $close === '/>' ) {
 376                                 // Empty element tag, <tag />
 377                                 $content = null;
 378                                 $text = $inside;
 379                                 $tail = null;
 380                         } else {
 381                                 if( $element == '!--' ) {
 382                                         $end = '/(-->)/';
 383                                 } else {
 384                                         $end = "/(<\\/$element\\s*>)/i";
 385                                 }
 386                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 387                                 $content = $q[0];
 388                                 if( count( $q ) < 3 ) {
 389                                         # No end tag -- let it run out to the end of the text.
 390                                         $tail = '';
 391                                         $text = '';
 392                                 } else {
 393                                         $tail = $q[1];
 394                                         $text = $q[2];
 395                                 }
 396                         }
 397
 398                         $matches[$marker] = array( $element,
 399                                 $content,
 400                                 Sanitizer::decodeTagAttributes( $attributes ),
 401                                 "<$element$attributes$close$content$tail" );
 402                 }
 403                 return $stripped;
 404         }
 405
 406         /**
 407          * Strips and renders nowiki, pre, math, hiero
 408          * If $render is set, performs necessary rendering operations on plugins
 409          * Returns the text, and fills an array with data needed in unstrip()
 410          * If the $state is already a valid strip state, it adds to the state
 411          *
 412          * @param bool $stripcomments when set, HTML comments <!-- like this -->
 413          *  will be stripped in addition to other tags. This is important
 414          *  for section editing, where these comments cause confusion when
 415          *  counting the sections in the wikisource
 416          *
 417          * @param array dontstrip contains tags which should not be stripped;
 418          *  used to prevent stipping of <gallery> when saving (fixes bug 2700)
 419          *
 420          * @private
 421          */
 422         function strip( $text, &$state, $stripcomments = false , $dontstrip = array () ) {
 423                 $render = ($this->mOutputType == OT_HTML);
 424
 425                 # Replace any instances of the placeholders
 426                 $uniq_prefix = $this->mUniqPrefix;
 427                 #$text = str_replace( $uniq_prefix, wfHtmlEscapeFirst( $uniq_prefix ), $text );
 428                 $commentState = array();
 429
 430                 $elements = array_merge(
 431                         array( 'nowiki', 'gallery' ),
 432                         array_keys( $this->mTagHooks ) );
 433                 global $wgRawHtml;
 434                 if( $wgRawHtml ) {
 435                         $elements[] = 'html';
 436                 }
 437                 if( $this->mOptions->getUseTeX() ) {
 438                         $elements[] = 'math';
 439                 }
 440
 441                 # Removing $dontstrip tags from $elements list (currently only 'gallery', fixing bug 2700)
 442                 foreach ( $elements AS $k => $v ) {
 443                         if ( !in_array ( $v , $dontstrip ) ) continue;
 444                         unset ( $elements[$k] );
 445                 }
 446
 447                 $matches = array();
 448                 $text = Parser::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 449
 450                 foreach( $matches as $marker => $data ) {
 451                         list( $element, $content, $params, $tag ) = $data;
 452                         if( $render ) {
 453                                 $tagName = strtolower( $element );
 454                                 switch( $tagName ) {
 455                                 case '!--':
 456                                         // Comment
 457                                         if( substr( $tag, -3 ) == '-->' ) {
 458                                                 $output = $tag;
 459                                         } else {
 460                                                 // Unclosed comment in input.
 461                                                 // Close it so later stripping can remove it
 462                                                 $output = "$tag-->";
 463                                         }
 464                                         break;
 465                                 case 'html':
 466                                         if( $wgRawHtml ) {
 467                                                 $output = $content;
 468                                                 break;
 469                                         }
 470                                         // Shouldn't happen otherwise. :)
 471                                 case 'nowiki':
 472                                         $output = wfEscapeHTMLTagsOnly( $content );
 473                                         break;
 474                                 case 'math':
 475                                         $output = MathRenderer::renderMath( $content );
 476                                         break;
 477                                 case 'gallery':
 478                                         $output = $this->renderImageGallery( $content, $params );
 479                                         break;
 480                                 default:
 481                                         if( isset( $this->mTagHooks[$tagName] ) ) {
 482                                                 $output = call_user_func_array( $this->mTagHooks[$tagName],
 483                                                         array( $content, $params, $this ) );
 484                                         } else {
 485                                                 throw new MWException( "Invalid call hook $element" );
 486                                         }
 487                                 }
 488                         } else {
 489                                 // Just stripping tags; keep the source
 490                                 $output = $tag;
 491                         }
 492                         if( !$stripcomments && $element == '!--' ) {
 493                                 $commentState[$marker] = $output;
 494                         } else {
 495                                 $state[$element][$marker] = $output;
 496                         }
 497                 }
 498
 499                 # Unstrip comments unless explicitly told otherwise.
 500                 # (The comments are always stripped prior to this point, so as to
 501                 # not invoke any extension tags / parser hooks contained within
 502                 # a comment.)
 503                 if ( !$stripcomments ) {
 504                         // Put them all back and forget them
 505                         $text = strtr( $text, $commentState );
 506                 }
 507
 508                 return $text;
 509         }
 510
 511         /**
 512          * Restores pre, math, and other extensions removed by strip()
 513          *
 514          * always call unstripNoWiki() after this one
 515          * @private
 516          */
 517         function unstrip( $text, &$state ) {
 518                 if ( !is_array( $state ) ) {
 519                         return $text;
 520                 }
 521
 522                 $replacements = array();
 523                 foreach( $state as $tag => $contentDict ) {
 524                         if( $tag != 'nowiki' && $tag != 'html' ) {
 525                                 foreach( $contentDict as $uniq => $content ) {
 526                                         $replacements[$uniq] = $content;
 527                                 }
 528                         }
 529                 }
 530                 $text = strtr( $text, $replacements );
 531
 532                 return $text;
 533         }
 534
 535         /**
 536          * Always call this after unstrip() to preserve the order
 537          *
 538          * @private
 539          */
 540         function unstripNoWiki( $text, &$state ) {
 541                 if ( !is_array( $state ) ) {
 542                         return $text;
 543                 }
 544
 545                 $replacements = array();
 546                 foreach( $state as $tag => $contentDict ) {
 547                         if( $tag == 'nowiki' || $tag == 'html' ) {
 548                                 foreach( $contentDict as $uniq => $content ) {
 549                                         $replacements[$uniq] = $content;
 550                                 }
 551                         }
 552                 }
 553                 $text = strtr( $text, $replacements );
 554
 555                 return $text;
 556         }
 557
 558         /**
 559          * Add an item to the strip state
 560          * Returns the unique tag which must be inserted into the stripped text
 561          * The tag will be replaced with the original text in unstrip()
 562          *
 563          * @private
 564          */
 565         function insertStripItem( $text, &$state ) {
 566                 $rnd = $this->mUniqPrefix . '-item' . Parser::getRandomString();
 567                 if ( !$state ) {
 568                         $state = array();
 569                 }
 570                 $state['item'][$rnd] = $text;
 571                 return $rnd;
 572         }
 573
 574         /**
 575          * Interface with html tidy, used if $wgUseTidy = true.
 576          * If tidy isn't able to correct the markup, the original will be
 577          * returned in all its glory with a warning comment appended.
 578          *
 579          * Either the external tidy program or the in-process tidy extension
 580          * will be used depending on availability. Override the default
 581          * $wgTidyInternal setting to disable the internal if it's not working.
 582          *
 583          * @param string $text Hideous HTML input
 584          * @return string Corrected HTML output
 585          * @public
 586          * @static
 587          */
 588         function tidy( $text ) {
 589                 global $wgTidyInternal;
 590                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 591 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 592 '<head><title>test</title></head><body>'.$text.'</body></html>';
 593                 if( $wgTidyInternal ) {
 594                         $correctedtext = Parser::internalTidy( $wrappedtext );
 595                 } else {
 596                         $correctedtext = Parser::externalTidy( $wrappedtext );
 597                 }
 598                 if( is_null( $correctedtext ) ) {
 599                         wfDebug( "Tidy error detected!\n" );
 600                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 601                 }
 602                 return $correctedtext;
 603         }
 604
 605         /**
 606          * Spawn an external HTML tidy process and get corrected markup back from it.
 607          *
 608          * @private
 609          * @static
 610          */
 611         function externalTidy( $text ) {
 612                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 613                 $fname = 'Parser::externalTidy';
 614                 wfProfileIn( $fname );
 615
 616                 $cleansource = '';
 617                 $opts = ' -utf8';
 618
 619                 $descriptorspec = array(
 620                         0 => array('pipe', 'r'),
 621                         1 => array('pipe', 'w'),
 622                         2 => array('file', '/dev/null', 'a')
 623                 );
 624                 $pipes = array();
 625                 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 626                 if (is_resource($process)) {
 627                         // Theoretically, this style of communication could cause a deadlock
 628                         // here. If the stdout buffer fills up, then writes to stdin could
 629                         // block. This doesn't appear to happen with tidy, because tidy only
 630                         // writes to stdout after it's finished reading from stdin. Search
 631                         // for tidyParseStdin and tidySaveStdout in console/tidy.c
 632                         fwrite($pipes[0], $text);
 633                         fclose($pipes[0]);
 634                         while (!feof($pipes[1])) {
 635                                 $cleansource .= fgets($pipes[1], 1024);
 636                         }
 637                         fclose($pipes[1]);
 638                         proc_close($process);
 639                 }
 640
 641                 wfProfileOut( $fname );
 642
 643                 if( $cleansource == '' && $text != '') {
 644                         // Some kind of error happened, so we couldn't get the corrected text.
 645                         // Just give up; we'll use the source text and append a warning.
 646                         return null;
 647                 } else {
 648                         return $cleansource;
 649                 }
 650         }
 651
 652         /**
 653          * Use the HTML tidy PECL extension to use the tidy library in-process,
 654          * saving the overhead of spawning a new process. Currently written to
 655          * the PHP 4.3.x version of the extension, may not work on PHP 5.
 656          *
 657          * 'pear install tidy' should be able to compile the extension module.
 658          *
 659          * @private
 660          * @static
 661          */
 662         function internalTidy( $text ) {
 663                 global $wgTidyConf;
 664                 $fname = 'Parser::internalTidy';
 665                 wfProfileIn( $fname );
 666
 667                 tidy_load_config( $wgTidyConf );
 668                 tidy_set_encoding( 'utf8' );
 669                 tidy_parse_string( $text );
 670                 tidy_clean_repair();
 671                 if( tidy_get_status() == 2 ) {
 672                         // 2 is magic number for fatal error
 673                         // http://www.php.net/manual/en/function.tidy-get-status.php
 674                         $cleansource = null;
 675                 } else {
 676                         $cleansource = tidy_get_output();
 677                 }
 678                 wfProfileOut( $fname );
 679                 return $cleansource;
 680         }
 681
 682         /**
 683          * parse the wiki syntax used to render tables
 684          *
 685          * @private
 686          */
 687         function doTableStuff ( $t ) {
 688                 $fname = 'Parser::doTableStuff';
 689                 wfProfileIn( $fname );
 690
 691                 $t = explode ( "\n" , $t ) ;
 692                 $td = array () ; # Is currently a td tag open?
 693                 $ltd = array () ; # Was it TD or TH?
 694                 $tr = array () ; # Is currently a tr tag open?
 695                 $ltr = array () ; # tr attributes
 696                 $has_opened_tr = array(); # Did this table open a <tr> element?
 697                 $indent_level = 0; # indent level of the table
 698                 foreach ( $t AS $k => $x )
 699                 {
 700                         $x = trim ( $x ) ;
 701                         $fc = substr ( $x , 0 , 1 ) ;
 702                         if ( preg_match( '/^(:*)\{\|(.*)$/', $x, $matches ) ) {
 703                                 $indent_level = strlen( $matches[1] );
 704
 705                                 $attributes = $this->unstripForHTML( $matches[2] );
 706
 707                                 $t[$k] = str_repeat( '<dl><dd>', $indent_level ) .
 708                                         '<table' . Sanitizer::fixTagAttributes ( $attributes, 'table' ) . '>' ;
 709                                 array_push ( $td , false ) ;
 710                                 array_push ( $ltd , '' ) ;
 711                                 array_push ( $tr , false ) ;
 712                                 array_push ( $ltr , '' ) ;
 713                                 array_push ( $has_opened_tr, false );
 714                         }
 715                         else if ( count ( $td ) == 0 ) { } # Don't do any of the following
 716                         else if ( '|}' == substr ( $x , 0 , 2 ) ) {
 717                                 $z = "</table>" . substr ( $x , 2);
 718                                 $l = array_pop ( $ltd ) ;
 719                                 if ( !array_pop ( $has_opened_tr ) ) $z = "<tr><td></td></tr>" . $z ;
 720                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 721                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 722                                 array_pop ( $ltr ) ;
 723                                 $t[$k] = $z . str_repeat( '</dd></dl>', $indent_level );
 724                         }
 725                         else if ( '|-' == substr ( $x , 0 , 2 ) ) { # Allows for |---------------
 726                                 $x = substr ( $x , 1 ) ;
 727                                 while ( $x != '' && substr ( $x , 0 , 1 ) == '-' ) $x = substr ( $x , 1 ) ;
 728                                 $z = '' ;
 729                                 $l = array_pop ( $ltd ) ;
 730                                 array_pop ( $has_opened_tr );
 731                                 array_push ( $has_opened_tr , true ) ;
 732                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 733                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 734                                 array_pop ( $ltr ) ;
 735                                 $t[$k] = $z ;
 736                                 array_push ( $tr , false ) ;
 737                                 array_push ( $td , false ) ;
 738                                 array_push ( $ltd , '' ) ;
 739                                 $attributes = $this->unstripForHTML( $x );
 740                                 array_push ( $ltr , Sanitizer::fixTagAttributes ( $attributes, 'tr' ) ) ;
 741                         }
 742                         else if ( '|' == $fc || '!' == $fc || '|+' == substr ( $x , 0 , 2 ) ) { # Caption
 743                                 # $x is a table row
 744                                 if ( '|+' == substr ( $x , 0 , 2 ) ) {
 745                                         $fc = '+' ;
 746                                         $x = substr ( $x , 1 ) ;
 747                                 }
 748                                 $after = substr ( $x , 1 ) ;
 749                                 if ( $fc == '!' ) $after = str_replace ( '!!' , '||' , $after ) ;
 750
 751                                 // Split up multiple cells on the same line.
 752                                 // FIXME: This can result in improper nesting of tags processed
 753                                 // by earlier parser steps, but should avoid splitting up eg
 754                                 // attribute values containing literal "||".
 755                                 $after = wfExplodeMarkup( '||', $after );
 756
 757                                 $t[$k] = '' ;
 758
 759                                 # Loop through each table cell
 760                                 foreach ( $after AS $theline )
 761                                 {
 762                                         $z = '' ;
 763                                         if ( $fc != '+' )
 764                                         {
 765                                                 $tra = array_pop ( $ltr ) ;
 766                                                 if ( !array_pop ( $tr ) ) $z = '<tr'.$tra.">\n" ;
 767                                                 array_push ( $tr , true ) ;
 768                                                 array_push ( $ltr , '' ) ;
 769                                                 array_pop ( $has_opened_tr );
 770                                                 array_push ( $has_opened_tr , true ) ;
 771                                         }
 772
 773                                         $l = array_pop ( $ltd ) ;
 774                                         if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 775                                         if ( $fc == '|' ) $l = 'td' ;
 776                                         else if ( $fc == '!' ) $l = 'th' ;
 777                                         else if ( $fc == '+' ) $l = 'caption' ;
 778                                         else $l = '' ;
 779                                         array_push ( $ltd , $l ) ;
 780
 781                                         # Cell parameters
 782                                         $y = explode ( '|' , $theline , 2 ) ;
 783                                         # Note that a '|' inside an invalid link should not
 784                                         # be mistaken as delimiting cell parameters
 785                                         if ( strpos( $y[0], '[[' ) !== false ) {
 786                                                 $y = array ($theline);
 787                                         }
 788                                         if ( count ( $y ) == 1 )
 789                                                 $y = "{$z}<{$l}>{$y[0]}" ;
 790                                         else {
 791                                                 $attributes = $this->unstripForHTML( $y[0] );
 792                                                 $y = "{$z}<{$l}".Sanitizer::fixTagAttributes($attributes, $l).">{$y[1]}" ;
 793                                         }
 794                                         $t[$k] .= $y ;
 795                                         array_push ( $td , true ) ;
 796                                 }
 797                         }
 798                 }
 799
 800                 # Closing open td, tr && table
 801                 while ( count ( $td ) > 0 )
 802                 {
 803                         $l = array_pop ( $ltd ) ;
 804                         if ( array_pop ( $td ) ) $t[] = '</td>' ;
 805                         if ( array_pop ( $tr ) ) $t[] = '</tr>' ;
 806                         if ( !array_pop ( $has_opened_tr ) ) $t[] = "<tr><td></td></tr>" ;
 807                         $t[] = '</table>' ;
 808                 }
 809
 810                 $t = implode ( "\n" , $t ) ;
 811                 # special case: don't return empty table
 812                 if($t == "<table>\n<tr><td></td></tr>\n</table>")
 813                         $t = '';
 814                 wfProfileOut( $fname );
 815                 return $t ;
 816         }
 817
 818         /**
 819          * Helper function for parse() that transforms wiki markup into
 820          * HTML. Only called for $mOutputType == OT_HTML.
 821          *
 822          * @private
 823          */
 824         function internalParse( $text ) {
 825                 $args = array();
 826                 $isMain = true;
 827                 $fname = 'Parser::internalParse';
 828                 wfProfileIn( $fname );
 829
 830                 # Remove <noinclude> tags and <includeonly> sections
 831                 $text = strtr( $text, array( '<onlyinclude>' => '' , '</onlyinclude>' => '' ) );
 832                 $text = strtr( $text, array( '<noinclude>' => '', '</noinclude>' => '') );
 833                 $text = preg_replace( '/<includeonly>.*?<\/includeonly>/s', '', $text );
 834
 835                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ) );
 836
 837                 $text = $this->replaceVariables( $text, $args );
 838
 839                 // Tables need to come after variable replacement for things to work
 840                 // properly; putting them before other transformations should keep
 841                 // exciting things like link expansions from showing up in surprising
 842                 // places.
 843                 $text = $this->doTableStuff( $text );
 844
 845                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 846
 847                 $text = $this->stripToc( $text );
 848                 $this->stripNoGallery( $text );
 849                 $text = $this->doHeadings( $text );
 850                 if($this->mOptions->getUseDynamicDates()) {
 851                         $df =& DateFormatter::getInstance();
 852                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 853                 }
 854                 $text = $this->doAllQuotes( $text );
 855                 $text = $this->replaceInternalLinks( $text );
 856                 $text = $this->replaceExternalLinks( $text );
 857
 858                 # replaceInternalLinks may sometimes leave behind
 859                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 860                 $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
 861
 862                 $text = $this->doMagicLinks( $text );
 863                 $text = $this->formatHeadings( $text, $isMain );
 864
 865                 wfProfileOut( $fname );
 866                 return $text;
 867         }
 868
 869         /**
 870          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 871          * magic external links.
 872          *
 873          * @private
 874          */
 875         function &doMagicLinks( &$text ) {
 876                 $text = $this->magicISBN( $text );
 877                 $text = $this->magicRFC( $text, 'RFC ', 'rfcurl' );
 878                 $text = $this->magicRFC( $text, 'PMID ', 'pubmedurl' );
 879                 return $text;
 880         }
 881
 882         /**
 883          * Parse headers and return html
 884          *
 885          * @private
 886          */
 887         function doHeadings( $text ) {
 888                 $fname = 'Parser::doHeadings';
 889                 wfProfileIn( $fname );
 890                 for ( $i = 6; $i >= 1; --$i ) {
 891                         $h = str_repeat( '=', $i );
 892                         $text = preg_replace( "/^{$h}(.+){$h}\\s*$/m",
 893                           "<h{$i}>\\1</h{$i}>\\2", $text );
 894                 }
 895                 wfProfileOut( $fname );
 896                 return $text;
 897         }
 898
 899         /**
 900          * Replace single quotes with HTML markup
 901          * @private
 902          * @return string the altered text
 903          */
 904         function doAllQuotes( $text ) {
 905                 $fname = 'Parser::doAllQuotes';
 906                 wfProfileIn( $fname );
 907                 $outtext = '';
 908                 $lines = explode( "\n", $text );
 909                 foreach ( $lines as $line ) {
 910                         $outtext .= $this->doQuotes ( $line ) . "\n";
 911                 }
 912                 $outtext = substr($outtext, 0,-1);
 913                 wfProfileOut( $fname );
 914                 return $outtext;
 915         }
 916
 917         /**
 918          * Helper function for doAllQuotes()
 919          * @private
 920          */
 921         function doQuotes( $text ) {
 922                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
 923                 if ( count( $arr ) == 1 )
 924                         return $text;
 925                 else
 926                 {
 927                         # First, do some preliminary work. This may shift some apostrophes from
 928                         # being mark-up to being text. It also counts the number of occurrences
 929                         # of bold and italics mark-ups.
 930                         $i = 0;
 931                         $numbold = 0;
 932                         $numitalics = 0;
 933                         foreach ( $arr as $r )
 934                         {
 935                                 if ( ( $i % 2 ) == 1 )
 936                                 {
 937                                         # If there are ever four apostrophes, assume the first is supposed to
 938                                         # be text, and the remaining three constitute mark-up for bold text.
 939                                         if ( strlen( $arr[$i] ) == 4 )
 940                                         {
 941                                                 $arr[$i-1] .= "'";
 942                                                 $arr[$i] = "'''";
 943                                         }
 944                                         # If there are more than 5 apostrophes in a row, assume they're all
 945                                         # text except for the last 5.
 946                                         else if ( strlen( $arr[$i] ) > 5 )
 947                                         {
 948                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
 949                                                 $arr[$i] = "'''''";
 950                                         }
 951                                         # Count the number of occurrences of bold and italics mark-ups.
 952                                         # We are not counting sequences of five apostrophes.
 953                                         if ( strlen( $arr[$i] ) == 2 ) $numitalics++;  else
 954                                         if ( strlen( $arr[$i] ) == 3 ) $numbold++;     else
 955                                         if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
 956                                 }
 957                                 $i++;
 958                         }
 959
 960                         # If there is an odd number of both bold and italics, it is likely
 961                         # that one of the bold ones was meant to be an apostrophe followed
 962                         # by italics. Which one we cannot know for certain, but it is more
 963                         # likely to be one that has a single-letter word before it.
 964                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
 965                         {
 966                                 $i = 0;
 967                                 $firstsingleletterword = -1;
 968                                 $firstmultiletterword = -1;
 969                                 $firstspace = -1;
 970                                 foreach ( $arr as $r )
 971                                 {
 972                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
 973                                         {
 974                                                 $x1 = substr ($arr[$i-1], -1);
 975                                                 $x2 = substr ($arr[$i-1], -2, 1);
 976                                                 if ($x1 == ' ') {
 977                                                         if ($firstspace == -1) $firstspace = $i;
 978                                                 } else if ($x2 == ' ') {
 979                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
 980                                                 } else {
 981                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
 982                                                 }
 983                                         }
 984                                         $i++;
 985                                 }
 986
 987                                 # If there is a single-letter word, use it!
 988                                 if ($firstsingleletterword > -1)
 989                                 {
 990                                         $arr [ $firstsingleletterword ] = "''";
 991                                         $arr [ $firstsingleletterword-1 ] .= "'";
 992                                 }
 993                                 # If not, but there's a multi-letter word, use that one.
 994                                 else if ($firstmultiletterword > -1)
 995                                 {
 996                                         $arr [ $firstmultiletterword ] = "''";
 997                                         $arr [ $firstmultiletterword-1 ] .= "'";
 998                                 }
 999                                 # ... otherwise use the first one that has neither.
1000                                 # (notice that it is possible for all three to be -1 if, for example,
1001                                 # there is only one pentuple-apostrophe in the line)
1002                                 else if ($firstspace > -1)
1003                                 {
1004                                         $arr [ $firstspace ] = "''";
1005                                         $arr [ $firstspace-1 ] .= "'";
1006                                 }
1007                         }
1008
1009                         # Now let's actually convert our apostrophic mush to HTML!
1010                         $output = '';
1011                         $buffer = '';
1012                         $state = '';
1013                         $i = 0;
1014                         foreach ($arr as $r)
1015                         {
1016                                 if (($i % 2) == 0)
1017                                 {
1018                                         if ($state == 'both')
1019                                                 $buffer .= $r;
1020                                         else
1021                                                 $output .= $r;
1022                                 }
1023                                 else
1024                                 {
1025                                         if (strlen ($r) == 2)
1026                                         {
1027                                                 if ($state == 'i')
1028                                                 { $output .= '</i>'; $state = ''; }
1029                                                 else if ($state == 'bi')
1030                                                 { $output .= '</i>'; $state = 'b'; }
1031                                                 else if ($state == 'ib')
1032                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1033                                                 else if ($state == 'both')
1034                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1035                                                 else # $state can be 'b' or ''
1036                                                 { $output .= '<i>'; $state .= 'i'; }
1037                                         }
1038                                         else if (strlen ($r) == 3)
1039                                         {
1040                                                 if ($state == 'b')
1041                                                 { $output .= '</b>'; $state = ''; }
1042                                                 else if ($state == 'bi')
1043                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1044                                                 else if ($state == 'ib')
1045                                                 { $output .= '</b>'; $state = 'i'; }
1046                                                 else if ($state == 'both')
1047                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1048                                                 else # $state can be 'i' or ''
1049                                                 { $output .= '<b>'; $state .= 'b'; }
1050                                         }
1051                                         else if (strlen ($r) == 5)
1052                                         {
1053                                                 if ($state == 'b')
1054                                                 { $output .= '</b><i>'; $state = 'i'; }
1055                                                 else if ($state == 'i')
1056                                                 { $output .= '</i><b>'; $state = 'b'; }
1057                                                 else if ($state == 'bi')
1058                                                 { $output .= '</i></b>'; $state = ''; }
1059                                                 else if ($state == 'ib')
1060                                                 { $output .= '</b></i>'; $state = ''; }
1061                                                 else if ($state == 'both')
1062                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1063                                                 else # ($state == '')
1064                                                 { $buffer = ''; $state = 'both'; }
1065                                         }
1066                                 }
1067                                 $i++;
1068                         }
1069                         # Now close all remaining tags.  Notice that the order is important.
1070                         if ($state == 'b' || $state == 'ib')
1071                                 $output .= '</b>';
1072                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
1073                                 $output .= '</i>';
1074                         if ($state == 'bi')
1075                                 $output .= '</b>';
1076                         if ($state == 'both')
1077                                 $output .= '<b><i>'.$buffer.'</i></b>';
1078                         return $output;
1079                 }
1080         }
1081
1082         /**
1083          * Replace external links
1084          *
1085          * Note: this is all very hackish and the order of execution matters a lot.
1086          * Make sure to run maintenance/parserTests.php if you change this code.
1087          *
1088          * @private
1089          */
1090         function replaceExternalLinks( $text ) {
1091                 global $wgContLang;
1092                 $fname = 'Parser::replaceExternalLinks';
1093                 wfProfileIn( $fname );
1094
1095                 $sk =& $this->mOptions->getSkin();
1096
1097                 $bits = preg_split( EXT_LINK_BRACKETED, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1098
1099                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
1100
1101                 $i = 0;
1102                 while ( $i<count( $bits ) ) {
1103                         $url = $bits[$i++];
1104                         $protocol = $bits[$i++];
1105                         $text = $bits[$i++];
1106                         $trail = $bits[$i++];
1107
1108                         # The characters '<' and '>' (which were escaped by
1109                         # removeHTMLtags()) should not be included in
1110                         # URLs, per RFC 2396.
1111                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1112                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1113                                 $url = substr($url, 0, $m2[0][1]);
1114                         }
1115
1116                         # If the link text is an image URL, replace it with an <img> tag
1117                         # This happened by accident in the original parser, but some people used it extensively
1118                         $img = $this->maybeMakeExternalImage( $text );
1119                         if ( $img !== false ) {
1120                                 $text = $img;
1121                         }
1122
1123                         $dtrail = '';
1124
1125                         # Set linktype for CSS - if URL==text, link is essentially free
1126                         $linktype = ($text == $url) ? 'free' : 'text';
1127
1128                         # No link text, e.g. [http://domain.tld/some.link]
1129                         if ( $text == '' ) {
1130                                 # Autonumber if allowed. See bug #5918
1131                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1132                                         $text = '[' . ++$this->mAutonumber . ']';
1133                                         $linktype = 'autonumber';
1134                                 } else {
1135                                         # Otherwise just use the URL
1136                                         $text = htmlspecialchars( $url );
1137                                         $linktype = 'free';
1138                                 }
1139                         } else {
1140                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1141                                 # Check for trail
1142                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1143                         }
1144
1145                         $text = $wgContLang->markNoConversion($text);
1146
1147                         # Normalize any HTML entities in input. They will be
1148                         # re-escaped by makeExternalLink().
1149                         $url = Sanitizer::decodeCharReferences( $url );
1150
1151                         # Escape any control characters introduced by the above step
1152                         $url = preg_replace( '/[\][<>"\\x00-\\x20\\x7F]/e', "urlencode('\\0')", $url );
1153
1154                         # Process the trail (i.e. everything after this link up until start of the next link),
1155                         # replacing any non-bracketed links
1156                         $trail = $this->replaceFreeExternalLinks( $trail );
1157
1158                         # Use the encoded URL
1159                         # This means that users can paste URLs directly into the text
1160                         # Funny characters like &ouml; aren't valid in URLs anyway
1161                         # This was changed in August 2004
1162                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype, $this->mTitle->getNamespace() ) . $dtrail . $trail;
1163
1164                         # Register link in the output object.
1165                         # Replace unnecessary URL escape codes with the referenced character
1166                         # This prevents spammers from hiding links from the filters
1167                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1168                         $this->mOutput->addExternalLink( $pasteurized );
1169                 }
1170
1171                 wfProfileOut( $fname );
1172                 return $s;
1173         }
1174
1175         /**
1176          * Replace anything that looks like a URL with a link
1177          * @private
1178          */
1179         function replaceFreeExternalLinks( $text ) {
1180                 global $wgContLang;
1181                 $fname = 'Parser::replaceFreeExternalLinks';
1182                 wfProfileIn( $fname );
1183
1184                 $bits = preg_split( '/(\b(?:' . wfUrlProtocols() . '))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1185                 $s = array_shift( $bits );
1186                 $i = 0;
1187
1188                 $sk =& $this->mOptions->getSkin();
1189
1190                 while ( $i < count( $bits ) ){
1191                         $protocol = $bits[$i++];
1192                         $remainder = $bits[$i++];
1193
1194                         if ( preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1195                                 # Found some characters after the protocol that look promising
1196                                 $url = $protocol . $m[1];
1197                                 $trail = $m[2];
1198
1199                                 # special case: handle urls as url args:
1200                                 # http://www.example.com/foo?=http://www.example.com/bar
1201                                 if(strlen($trail) == 0 &&
1202                                         isset($bits[$i]) &&
1203                                         preg_match('/^'. wfUrlProtocols() . '$/S', $bits[$i]) &&
1204                                         preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $bits[$i + 1], $m ))
1205                                 {
1206                                         # add protocol, arg
1207                                         $url .= $bits[$i] . $m[1]; # protocol, url as arg to previous link
1208                                         $i += 2;
1209                                         $trail = $m[2];
1210                                 }
1211
1212                                 # The characters '<' and '>' (which were escaped by
1213                                 # removeHTMLtags()) should not be included in
1214                                 # URLs, per RFC 2396.
1215                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1216                                         $trail = substr($url, $m2[0][1]) . $trail;
1217                                         $url = substr($url, 0, $m2[0][1]);
1218                                 }
1219
1220                                 # Move trailing punctuation to $trail
1221                                 $sep = ',;\.:!?';
1222                                 # If there is no left bracket, then consider right brackets fair game too
1223                                 if ( strpos( $url, '(' ) === false ) {
1224                                         $sep .= ')';
1225                                 }
1226
1227                                 $numSepChars = strspn( strrev( $url ), $sep );
1228                                 if ( $numSepChars ) {
1229                                         $trail = substr( $url, -$numSepChars ) . $trail;
1230                                         $url = substr( $url, 0, -$numSepChars );
1231                                 }
1232
1233                                 # Normalize any HTML entities in input. They will be
1234                                 # re-escaped by makeExternalLink() or maybeMakeExternalImage()
1235                                 $url = Sanitizer::decodeCharReferences( $url );
1236
1237                                 # Escape any control characters introduced by the above step
1238                                 $url = preg_replace( '/[\][<>"\\x00-\\x20\\x7F]/e', "urlencode('\\0')", $url );
1239
1240                                 # Is this an external image?
1241                                 $text = $this->maybeMakeExternalImage( $url );
1242                                 if ( $text === false ) {
1243                                         # Not an image, make a link
1244                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free', $this->mTitle->getNamespace() );
1245                                         # Register it in the output object...
1246                                         # Replace unnecessary URL escape codes with their equivalent characters
1247                                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1248                                         $this->mOutput->addExternalLink( $pasteurized );
1249                                 }
1250                                 $s .= $text . $trail;
1251                         } else {
1252                                 $s .= $protocol . $remainder;
1253                         }
1254                 }
1255                 wfProfileOut( $fname );
1256                 return $s;
1257         }
1258
1259         /**
1260          * Replace unusual URL escape codes with their equivalent characters
1261          * @param string
1262          * @return string
1263          * @static
1264          * @fixme This can merge genuinely required bits in the path or query string,
1265          *        breaking legit URLs. A proper fix would treat the various parts of
1266          *        the URL differently; as a workaround, just use the output for
1267          *        statistical records, not for actual linking/output.
1268          */
1269         function replaceUnusualEscapes( $url ) {
1270                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1271                         array( 'Parser', 'replaceUnusualEscapesCallback' ), $url );
1272         }
1273
1274         /**
1275          * Callback function used in replaceUnusualEscapes().
1276          * Replaces unusual URL escape codes with their equivalent character
1277          * @static
1278          * @private
1279          */
1280         function replaceUnusualEscapesCallback( $matches ) {
1281                 $char = urldecode( $matches[0] );
1282                 $ord = ord( $char );
1283                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1284                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1285                         // No, shouldn't be escaped
1286                         return $char;
1287                 } else {
1288                         // Yes, leave it escaped
1289                         return $matches[0];
1290                 }
1291         }
1292
1293         /**
1294          * make an image if it's allowed, either through the global
1295          * option or through the exception
1296          * @private
1297          */
1298         function maybeMakeExternalImage( $url ) {
1299                 $sk =& $this->mOptions->getSkin();
1300                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1301                 $imagesexception = !empty($imagesfrom);
1302                 $text = false;
1303                 if ( $this->mOptions->getAllowExternalImages()
1304                      || ( $imagesexception && strpos( $url, $imagesfrom ) === 0 ) ) {
1305                         if ( preg_match( EXT_IMAGE_REGEX, $url ) ) {
1306                                 # Image found
1307                                 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
1308                         }
1309                 }
1310                 return $text;
1311         }
1312
1313         /**
1314          * Process [[ ]] wikilinks
1315          *
1316          * @private
1317          */
1318         function replaceInternalLinks( $s ) {
1319                 global $wgContLang;
1320                 static $fname = 'Parser::replaceInternalLinks' ;
1321
1322                 wfProfileIn( $fname );
1323
1324                 wfProfileIn( $fname.'-setup' );
1325                 static $tc = FALSE;
1326                 # the % is needed to support urlencoded titles as well
1327                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1328
1329                 $sk =& $this->mOptions->getSkin();
1330
1331                 #split the entire text string on occurences of [[
1332                 $a = explode( '[[', ' ' . $s );
1333                 #get the first element (all text up to first [[), and remove the space we added
1334                 $s = array_shift( $a );
1335                 $s = substr( $s, 1 );
1336
1337                 # Match a link having the form [[namespace:link|alternate]]trail
1338                 static $e1 = FALSE;
1339                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1340                 # Match cases where there is no "]]", which might still be images
1341                 static $e1_img = FALSE;
1342                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1343                 # Match the end of a line for a word that's not followed by whitespace,
1344                 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1345                 $e2 = wfMsgForContent( 'linkprefix' );
1346
1347                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1348
1349                 if( is_null( $this->mTitle ) ) {
1350                         throw new MWException( 'nooo' );
1351                 }
1352                 $nottalk = !$this->mTitle->isTalkPage();
1353
1354                 if ( $useLinkPrefixExtension ) {
1355                         if ( preg_match( $e2, $s, $m ) ) {
1356                                 $first_prefix = $m[2];
1357                         } else {
1358                                 $first_prefix = false;
1359                         }
1360                 } else {
1361                         $prefix = '';
1362                 }
1363
1364                 $selflink = $this->mTitle->getPrefixedText();
1365                 wfProfileOut( $fname.'-setup' );
1366
1367                 $checkVariantLink = sizeof($wgContLang->getVariants())>1;
1368                 $useSubpages = $this->areSubpagesAllowed();
1369
1370                 # Loop for each link
1371                 for ($k = 0; isset( $a[$k] ); $k++) {
1372                         $line = $a[$k];
1373                         if ( $useLinkPrefixExtension ) {
1374                                 wfProfileIn( $fname.'-prefixhandling' );
1375                                 if ( preg_match( $e2, $s, $m ) ) {
1376                                         $prefix = $m[2];
1377                                         $s = $m[1];
1378                                 } else {
1379                                         $prefix='';
1380                                 }
1381                                 # first link
1382                                 if($first_prefix) {
1383                                         $prefix = $first_prefix;
1384                                         $first_prefix = false;
1385                                 }
1386                                 wfProfileOut( $fname.'-prefixhandling' );
1387                         }
1388
1389                         $might_be_img = false;
1390
1391                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1392                                 $text = $m[2];
1393                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1394                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1395                                 # the real problem is with the $e1 regex
1396                                 # See bug 1300.
1397                                 #
1398                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1399                                 # and no image is in sight. See bug 2095.
1400                                 #
1401                                 if( $text !== '' &&
1402                                         preg_match( "/^\](.*)/s", $m[3], $n ) &&
1403                                         strpos($text, '[') !== false
1404                                 )
1405                                 {
1406                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1407                                         $m[3] = $n[1];
1408                                 }
1409                                 # fix up urlencoded title texts
1410                                 if(preg_match('/%/', $m[1] ))
1411                                         # Should anchors '#' also be rejected?
1412                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1413                                 $trail = $m[3];
1414                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1415                                 $might_be_img = true;
1416                                 $text = $m[2];
1417                                 if(preg_match('/%/', $m[1] )) $m[1] = urldecode($m[1]);
1418                                 $trail = "";
1419                         } else { # Invalid form; output directly
1420                                 $s .= $prefix . '[[' . $line ;
1421                                 continue;
1422                         }
1423
1424                         # Don't allow internal links to pages containing
1425                         # PROTO: where PROTO is a valid URL protocol; these
1426                         # should be external links.
1427                         if (preg_match('/^(\b(?:' . wfUrlProtocols() . '))/', $m[1])) {
1428                                 $s .= $prefix . '[[' . $line ;
1429                                 continue;
1430                         }
1431
1432                         # Make subpage if necessary
1433                         if( $useSubpages ) {
1434                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1435                         } else {
1436                                 $link = $m[1];
1437                         }
1438
1439                         $noforce = (substr($m[1], 0, 1) != ':');
1440                         if (!$noforce) {
1441                                 # Strip off leading ':'
1442                                 $link = substr($link, 1);
1443                         }
1444
1445                         $nt = Title::newFromText( $this->unstripNoWiki($link, $this->mStripState) );
1446                         if( !$nt ) {
1447                                 $s .= $prefix . '[[' . $line;
1448                                 continue;
1449                         }
1450
1451                         #check other language variants of the link
1452                         #if the article does not exist
1453                         if( $checkVariantLink
1454                             && $nt->getArticleID() == 0 ) {
1455                                 $wgContLang->findVariantLink($link, $nt);
1456                         }
1457
1458                         $ns = $nt->getNamespace();
1459                         $iw = $nt->getInterWiki();
1460
1461                         if ($might_be_img) { # if this is actually an invalid link
1462                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1463                                         $found = false;
1464                                         while (isset ($a[$k+1]) ) {
1465                                                 #look at the next 'line' to see if we can close it there
1466                                                 $spliced = array_splice( $a, $k + 1, 1 );
1467                                                 $next_line = array_shift( $spliced );
1468                                                 if( preg_match("/^(.*?]].*?)]](.*)$/sD", $next_line, $m) ) {
1469                                                 # the first ]] closes the inner link, the second the image
1470                                                         $found = true;
1471                                                         $text .= '[[' . $m[1];
1472                                                         $trail = $m[2];
1473                                                         break;
1474                                                 } elseif( preg_match("/^.*?]].*$/sD", $next_line, $m) ) {
1475                                                         #if there's exactly one ]] that's fine, we'll keep looking
1476                                                         $text .= '[[' . $m[0];
1477                                                 } else {
1478                                                         #if $next_line is invalid too, we need look no further
1479                                                         $text .= '[[' . $next_line;
1480                                                         break;
1481                                                 }
1482                                         }
1483                                         if ( !$found ) {
1484                                                 # we couldn't find the end of this imageLink, so output it raw
1485                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1486                                                 $text = $this->replaceInternalLinks($text);
1487                                                 $s .= $prefix . '[[' . $link . '|' . $text;
1488                                                 # note: no $trail, because without an end, there *is* no trail
1489                                                 continue;
1490                                         }
1491                                 } else { #it's not an image, so output it raw
1492                                         $s .= $prefix . '[[' . $link . '|' . $text;
1493                                         # note: no $trail, because without an end, there *is* no trail
1494                                         continue;
1495                                 }
1496                         }
1497
1498                         $wasblank = ( '' == $text );
1499                         if( $wasblank ) $text = $link;
1500
1501
1502                         # Link not escaped by : , create the various objects
1503                         if( $noforce ) {
1504
1505                                 # Interwikis
1506                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1507                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1508                                         $s = rtrim($s . "\n");
1509                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1510                                         continue;
1511                                 }
1512
1513                                 if ( $ns == NS_IMAGE ) {
1514                                         wfProfileIn( "$fname-image" );
1515                                         if ( !wfIsBadImage( $nt->getDBkey() ) ) {
1516                                                 # recursively parse links inside the image caption
1517                                                 # actually, this will parse them in any other parameters, too,
1518                                                 # but it might be hard to fix that, and it doesn't matter ATM
1519                                                 $text = $this->replaceExternalLinks($text);
1520                                                 $text = $this->replaceInternalLinks($text);
1521
1522                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1523                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text ) ) . $trail;
1524                                                 $this->mOutput->addImage( $nt->getDBkey() );
1525
1526                                                 wfProfileOut( "$fname-image" );
1527                                                 continue;
1528                                         }
1529                                         wfProfileOut( "$fname-image" );
1530
1531                                 }
1532
1533                                 if ( $ns == NS_CATEGORY ) {
1534                                         wfProfileIn( "$fname-category" );
1535                                         $s = rtrim($s . "\n"); # bug 87
1536
1537                                         if ( $wasblank ) {
1538                                                 if ( $this->mTitle->getNamespace() == NS_CATEGORY ) {
1539                                                         $sortkey = $this->mTitle->getText();
1540                                                 } else {
1541                                                         $sortkey = $this->mTitle->getPrefixedText();
1542                                                 }
1543                                         } else {
1544                                                 $sortkey = $text;
1545                                         }
1546                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1547                                         $sortkey = str_replace( "\n", '', $sortkey );
1548                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1549                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1550
1551                                         /**
1552                                          * Strip the whitespace Category links produce, see bug 87
1553                                          * @todo We might want to use trim($tmp, "\n") here.
1554                                          */
1555                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1556
1557                                         wfProfileOut( "$fname-category" );
1558                                         continue;
1559                                 }
1560                         }
1561
1562                         if( ( $nt->getPrefixedText() === $selflink ) &&
1563                             ( $nt->getFragment() === '' ) ) {
1564                                 # Self-links are handled specially; generally de-link and change to bold.
1565                                 $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1566                                 continue;
1567                         }
1568
1569                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1570                         if( $ns == NS_MEDIA ) {
1571                                 $link = $sk->makeMediaLinkObj( $nt, $text );
1572                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1573                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1574                                 $this->mOutput->addImage( $nt->getDBkey() );
1575                                 continue;
1576                         } elseif( $ns == NS_SPECIAL ) {
1577                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1578                                 continue;
1579                         } elseif( $ns == NS_IMAGE ) {
1580                                 $img = Image::newFromTitle( $nt );
1581                                 if( $img->exists() ) {
1582                                         // Force a blue link if the file exists; may be a remote
1583                                         // upload on the shared repository, and we want to see its
1584                                         // auto-generated page.
1585                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1586                                         continue;
1587                                 }
1588                         }
1589                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1590                 }
1591                 wfProfileOut( $fname );
1592                 return $s;
1593         }
1594
1595         /**
1596          * Make a link placeholder. The text returned can be later resolved to a real link with
1597          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1598          * parsing of interwiki links, and secondly to allow all extistence checks and
1599          * article length checks (for stub links) to be bundled into a single query.
1600          *
1601          */
1602         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1603                 if ( ! is_object($nt) ) {
1604                         # Fail gracefully
1605                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1606                 } else {
1607                         # Separate the link trail from the rest of the link
1608                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1609
1610                         if ( $nt->isExternal() ) {
1611                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1612                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1613                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1614                         } else {
1615                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1616                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1617                                 $this->mLinkHolders['queries'][] = $query;
1618                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1619                                 $this->mLinkHolders['titles'][] = $nt;
1620
1621                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1622                         }
1623                 }
1624                 return $retVal;
1625         }
1626
1627         /**
1628          * Render a forced-blue link inline; protect against double expansion of
1629          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1630          * Since this little disaster has to split off the trail text to avoid
1631          * breaking URLs in the following text without breaking trails on the
1632          * wiki links, it's been made into a horrible function.
1633          *
1634          * @param Title $nt
1635          * @param string $text
1636          * @param string $query
1637          * @param string $trail
1638          * @param string $prefix
1639          * @return string HTML-wikitext mix oh yuck
1640          */
1641         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1642                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1643                 $sk =& $this->mOptions->getSkin();
1644                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1645                 return $this->armorLinks( $link ) . $trail;
1646         }
1647
1648         /**
1649          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1650          * going to go through further parsing steps before inline URL expansion.
1651          *
1652          * In particular this is important when using action=render, which causes
1653          * full URLs to be included.
1654          *
1655          * Oh man I hate our multi-layer parser!
1656          *
1657          * @param string more-or-less HTML
1658          * @return string less-or-more HTML with NOPARSE bits
1659          */
1660         function armorLinks( $text ) {
1661                 return preg_replace( "/\b(" . wfUrlProtocols() . ')/',
1662                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1663         }
1664
1665         /**
1666          * Return true if subpage links should be expanded on this page.
1667          * @return bool
1668          */
1669         function areSubpagesAllowed() {
1670                 # Some namespaces don't allow subpages
1671                 global $wgNamespacesWithSubpages;
1672                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1673         }
1674
1675         /**
1676          * Handle link to subpage if necessary
1677          * @param string $target the source of the link
1678          * @param string &$text the link text, modified as necessary
1679          * @return string the full name of the link
1680          * @private
1681          */
1682         function maybeDoSubpageLink($target, &$text) {
1683                 # Valid link forms:
1684                 # Foobar -- normal
1685                 # :Foobar -- override special treatment of prefix (images, language links)
1686                 # /Foobar -- convert to CurrentPage/Foobar
1687                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1688                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1689                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1690
1691                 $fname = 'Parser::maybeDoSubpageLink';
1692                 wfProfileIn( $fname );
1693                 $ret = $target; # default return value is no change
1694
1695                 # Some namespaces don't allow subpages,
1696                 # so only perform processing if subpages are allowed
1697                 if( $this->areSubpagesAllowed() ) {
1698                         # Look at the first character
1699                         if( $target != '' && $target{0} == '/' ) {
1700                                 # / at end means we don't want the slash to be shown
1701                                 if( substr( $target, -1, 1 ) == '/' ) {
1702                                         $target = substr( $target, 1, -1 );
1703                                         $noslash = $target;
1704                                 } else {
1705                                         $noslash = substr( $target, 1 );
1706                                 }
1707
1708                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash);
1709                                 if( '' === $text ) {
1710                                         $text = $target;
1711                                 } # this might be changed for ugliness reasons
1712                         } else {
1713                                 # check for .. subpage backlinks
1714                                 $dotdotcount = 0;
1715                                 $nodotdot = $target;
1716                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1717                                         ++$dotdotcount;
1718                                         $nodotdot = substr( $nodotdot, 3 );
1719                                 }
1720                                 if($dotdotcount > 0) {
1721                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1722                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1723                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1724                                                 # / at the end means don't show full path
1725                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1726                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1727                                                         if( '' === $text ) {
1728                                                                 $text = $nodotdot;
1729                                                         }
1730                                                 }
1731                                                 $nodotdot = trim( $nodotdot );
1732                                                 if( $nodotdot != '' ) {
1733                                                         $ret .= '/' . $nodotdot;
1734                                                 }
1735                                         }
1736                                 }
1737                         }
1738                 }
1739
1740                 wfProfileOut( $fname );
1741                 return $ret;
1742         }
1743
1744         /**#@+
1745          * Used by doBlockLevels()
1746          * @private
1747          */
1748         /* private */ function closeParagraph() {
1749                 $result = '';
1750                 if ( '' != $this->mLastSection ) {
1751                         $result = '</' . $this->mLastSection  . ">\n";
1752                 }
1753                 $this->mInPre = false;
1754                 $this->mLastSection = '';
1755                 return $result;
1756         }
1757         # getCommon() returns the length of the longest common substring
1758         # of both arguments, starting at the beginning of both.
1759         #
1760         /* private */ function getCommon( $st1, $st2 ) {
1761                 $fl = strlen( $st1 );
1762                 $shorter = strlen( $st2 );
1763                 if ( $fl < $shorter ) { $shorter = $fl; }
1764
1765                 for ( $i = 0; $i < $shorter; ++$i ) {
1766                         if ( $st1{$i} != $st2{$i} ) { break; }
1767                 }
1768                 return $i;
1769         }
1770         # These next three functions open, continue, and close the list
1771         # element appropriate to the prefix character passed into them.
1772         #
1773         /* private */ function openList( $char ) {
1774                 $result = $this->closeParagraph();
1775
1776                 if ( '*' == $char ) { $result .= '<ul><li>'; }
1777                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
1778                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
1779                 else if ( ';' == $char ) {
1780                         $result .= '<dl><dt>';
1781                         $this->mDTopen = true;
1782                 }
1783                 else { $result = '<!-- ERR 1 -->'; }
1784
1785                 return $result;
1786         }
1787
1788         /* private */ function nextItem( $char ) {
1789                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
1790                 else if ( ':' == $char || ';' == $char ) {
1791                         $close = '</dd>';
1792                         if ( $this->mDTopen ) { $close = '</dt>'; }
1793                         if ( ';' == $char ) {
1794                                 $this->mDTopen = true;
1795                                 return $close . '<dt>';
1796                         } else {
1797                                 $this->mDTopen = false;
1798                                 return $close . '<dd>';
1799                         }
1800                 }
1801                 return '<!-- ERR 2 -->';
1802         }
1803
1804         /* private */ function closeList( $char ) {
1805                 if ( '*' == $char ) { $text = '</li></ul>'; }
1806                 else if ( '#' == $char ) { $text = '</li></ol>'; }
1807                 else if ( ':' == $char ) {
1808                         if ( $this->mDTopen ) {
1809                                 $this->mDTopen = false;
1810                                 $text = '</dt></dl>';
1811                         } else {
1812                                 $text = '</dd></dl>';
1813                         }
1814                 }
1815                 else {  return '<!-- ERR 3 -->'; }
1816                 return $text."\n";
1817         }
1818         /**#@-*/
1819
1820         /**
1821          * Make lists from lines starting with ':', '*', '#', etc.
1822          *
1823          * @private
1824          * @return string the lists rendered as HTML
1825          */
1826         function doBlockLevels( $text, $linestart ) {
1827                 $fname = 'Parser::doBlockLevels';
1828                 wfProfileIn( $fname );
1829
1830                 # Parsing through the text line by line.  The main thing
1831                 # happening here is handling of block-level elements p, pre,
1832                 # and making lists from lines starting with * # : etc.
1833                 #
1834                 $textLines = explode( "\n", $text );
1835
1836                 $lastPrefix = $output = '';
1837                 $this->mDTopen = $inBlockElem = false;
1838                 $prefixLength = 0;
1839                 $paragraphStack = false;
1840
1841                 if ( !$linestart ) {
1842                         $output .= array_shift( $textLines );
1843                 }
1844                 foreach ( $textLines as $oLine ) {
1845                         $lastPrefixLength = strlen( $lastPrefix );
1846                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
1847                         $preOpenMatch = preg_match('/<pre/i', $oLine );
1848                         if ( !$this->mInPre ) {
1849                                 # Multiple prefixes may abut each other for nested lists.
1850                                 $prefixLength = strspn( $oLine, '*#:;' );
1851                                 $pref = substr( $oLine, 0, $prefixLength );
1852
1853                                 # eh?
1854                                 $pref2 = str_replace( ';', ':', $pref );
1855                                 $t = substr( $oLine, $prefixLength );
1856                                 $this->mInPre = !empty($preOpenMatch);
1857                         } else {
1858                                 # Don't interpret any other prefixes in preformatted text
1859                                 $prefixLength = 0;
1860                                 $pref = $pref2 = '';
1861                                 $t = $oLine;
1862                         }
1863
1864                         # List generation
1865                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
1866                                 # Same as the last item, so no need to deal with nesting or opening stuff
1867                                 $output .= $this->nextItem( substr( $pref, -1 ) );
1868                                 $paragraphStack = false;
1869
1870                                 if ( substr( $pref, -1 ) == ';') {
1871                                         # The one nasty exception: definition lists work like this:
1872                                         # ; title : definition text
1873                                         # So we check for : in the remainder text to split up the
1874                                         # title and definition, without b0rking links.
1875                                         $term = $t2 = '';
1876                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
1877                                                 $t = $t2;
1878                                                 $output .= $term . $this->nextItem( ':' );
1879                                         }
1880                                 }
1881                         } elseif( $prefixLength || $lastPrefixLength ) {
1882                                 # Either open or close a level...
1883                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
1884                                 $paragraphStack = false;
1885
1886                                 while( $commonPrefixLength < $lastPrefixLength ) {
1887                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
1888                                         --$lastPrefixLength;
1889                                 }
1890                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
1891                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
1892                                 }
1893                                 while ( $prefixLength > $commonPrefixLength ) {
1894                                         $char = substr( $pref, $commonPrefixLength, 1 );
1895                                         $output .= $this->openList( $char );
1896
1897                                         if ( ';' == $char ) {
1898                                                 # FIXME: This is dupe of code above
1899                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
1900                                                         $t = $t2;
1901                                                         $output .= $term . $this->nextItem( ':' );
1902                                                 }
1903                                         }
1904                                         ++$commonPrefixLength;
1905                                 }
1906                                 $lastPrefix = $pref2;
1907                         }
1908                         if( 0 == $prefixLength ) {
1909                                 wfProfileIn( "$fname-paragraph" );
1910                                 # No prefix (not in list)--go to paragraph mode
1911                                 // XXX: use a stack for nestable elements like span, table and div
1912                                 $openmatch = preg_match('/(<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/center|<\\/tr|<\\/td|<\\/th)/iS', $t );
1913                                 $closematch = preg_match(
1914                                         '/(<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
1915                                         '<td|<th|<div|<\\/div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<center)/iS', $t );
1916                                 if ( $openmatch or $closematch ) {
1917                                         $paragraphStack = false;
1918                                         # TODO bug 5718: paragraph closed
1919                                         $output .= $this->closeParagraph();
1920                                         if ( $preOpenMatch and !$preCloseMatch ) {
1921                                                 $this->mInPre = true;
1922                                         }
1923                                         if ( $closematch ) {
1924                                                 $inBlockElem = false;
1925                                         } else {
1926                                                 $inBlockElem = true;
1927                                         }
1928                                 } else if ( !$inBlockElem && !$this->mInPre ) {
1929                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
1930                                                 // pre
1931                                                 if ($this->mLastSection != 'pre') {
1932                                                         $paragraphStack = false;
1933                                                         $output .= $this->closeParagraph().'<pre>';
1934                                                         $this->mLastSection = 'pre';
1935                                                 }
1936                                                 $t = substr( $t, 1 );
1937                                         } else {
1938                                                 // paragraph
1939                                                 if ( '' == trim($t) ) {
1940                                                         if ( $paragraphStack ) {
1941                                                                 $output .= $paragraphStack.'<br />';
1942                                                                 $paragraphStack = false;
1943                                                                 $this->mLastSection = 'p';
1944                                                         } else {
1945                                                                 if ($this->mLastSection != 'p' ) {
1946                                                                         $output .= $this->closeParagraph();
1947                                                                         $this->mLastSection = '';
1948                                                                         $paragraphStack = '<p>';
1949                                                                 } else {
1950                                                                         $paragraphStack = '</p><p>';
1951                                                                 }
1952                                                         }
1953                                                 } else {
1954                                                         if ( $paragraphStack ) {
1955                                                                 $output .= $paragraphStack;
1956                                                                 $paragraphStack = false;
1957                                                                 $this->mLastSection = 'p';
1958                                                         } else if ($this->mLastSection != 'p') {
1959                                                                 $output .= $this->closeParagraph().'<p>';
1960                                                                 $this->mLastSection = 'p';
1961                                                         }
1962                                                 }
1963                                         }
1964                                 }
1965                                 wfProfileOut( "$fname-paragraph" );
1966                         }
1967                         // somewhere above we forget to get out of pre block (bug 785)
1968                         if($preCloseMatch && $this->mInPre) {
1969                                 $this->mInPre = false;
1970                         }
1971                         if ($paragraphStack === false) {
1972                                 $output .= $t."\n";
1973                         }
1974                 }
1975                 while ( $prefixLength ) {
1976                         $output .= $this->closeList( $pref2{$prefixLength-1} );
1977                         --$prefixLength;
1978                 }
1979                 if ( '' != $this->mLastSection ) {
1980                         $output .= '</' . $this->mLastSection . '>';
1981                         $this->mLastSection = '';
1982                 }
1983
1984                 wfProfileOut( $fname );
1985                 return $output;
1986         }
1987
1988         /**
1989          * Split up a string on ':', ignoring any occurences inside tags
1990          * to prevent illegal overlapping.
1991          * @param string $str the string to split
1992          * @param string &$before set to everything before the ':'
1993          * @param string &$after set to everything after the ':'
1994          * return string the position of the ':', or false if none found
1995          */
1996         function findColonNoLinks($str, &$before, &$after) {
1997                 $fname = 'Parser::findColonNoLinks';
1998                 wfProfileIn( $fname );
1999
2000                 $pos = strpos( $str, ':' );
2001                 if( $pos === false ) {
2002                         // Nothing to find!
2003                         wfProfileOut( $fname );
2004                         return false;
2005                 }
2006
2007                 $lt = strpos( $str, '<' );
2008                 if( $lt === false || $lt > $pos ) {
2009                         // Easy; no tag nesting to worry about
2010                         $before = substr( $str, 0, $pos );
2011                         $after = substr( $str, $pos+1 );
2012                         wfProfileOut( $fname );
2013                         return $pos;
2014                 }
2015
2016                 // Ugly state machine to walk through avoiding tags.
2017                 $state = MW_COLON_STATE_TEXT;
2018                 $stack = 0;
2019                 $len = strlen( $str );
2020                 for( $i = 0; $i < $len; $i++ ) {
2021                         $c = $str{$i};
2022
2023                         switch( $state ) {
2024                         // (Using the number is a performance hack for common cases)
2025                         case 0: // MW_COLON_STATE_TEXT:
2026                                 switch( $c ) {
2027                                 case "<":
2028                                         // Could be either a <start> tag or an </end> tag
2029                                         $state = MW_COLON_STATE_TAGSTART;
2030                                         break;
2031                                 case ":":
2032                                         if( $stack == 0 ) {
2033                                                 // We found it!
2034                                                 $before = substr( $str, 0, $i );
2035                                                 $after = substr( $str, $i + 1 );
2036                                                 wfProfileOut( $fname );
2037                                                 return $i;
2038                                         }
2039                                         // Embedded in a tag; don't break it.
2040                                         break;
2041                                 default:
2042                                         // Skip ahead looking for something interesting
2043                                         $colon = strpos( $str, ':', $i );
2044                                         if( $colon === false ) {
2045                                                 // Nothing else interesting
2046                                                 wfProfileOut( $fname );
2047                                                 return false;
2048                                         }
2049                                         $lt = strpos( $str, '<', $i );
2050                                         if( $stack === 0 ) {
2051                                                 if( $lt === false || $colon < $lt ) {
2052                                                         // We found it!
2053                                                         $before = substr( $str, 0, $colon );
2054                                                         $after = substr( $str, $colon + 1 );
2055                                                         wfProfileOut( $fname );
2056                                                         return $i;
2057                                                 }
2058                                         }
2059                                         if( $lt === false ) {
2060                                                 // Nothing else interesting to find; abort!
2061                                                 // We're nested, but there's no close tags left. Abort!
2062                                                 break 2;
2063                                         }
2064                                         // Skip ahead to next tag start
2065                                         $i = $lt;
2066                                         $state = MW_COLON_STATE_TAGSTART;
2067                                 }
2068                                 break;
2069                         case 1: // MW_COLON_STATE_TAG:
2070                                 // In a <tag>
2071                                 switch( $c ) {
2072                                 case ">":
2073                                         $stack++;
2074                                         $state = MW_COLON_STATE_TEXT;
2075                                         break;
2076                                 case "/":
2077                                         // Slash may be followed by >?
2078                                         $state = MW_COLON_STATE_TAGSLASH;
2079                                         break;
2080                                 default:
2081                                         // ignore
2082                                 }
2083                                 break;
2084                         case 2: // MW_COLON_STATE_TAGSTART:
2085                                 switch( $c ) {
2086                                 case "/":
2087                                         $state = MW_COLON_STATE_CLOSETAG;
2088                                         break;
2089                                 case "!":
2090                                         $state = MW_COLON_STATE_COMMENT;
2091                                         break;
2092                                 case ">":
2093                                         // Illegal early close? This shouldn't happen D:
2094                                         $state = MW_COLON_STATE_TEXT;
2095                                         break;
2096                                 default:
2097                                         $state = MW_COLON_STATE_TAG;
2098                                 }
2099                                 break;
2100                         case 3: // MW_COLON_STATE_CLOSETAG:
2101                                 // In a </tag>
2102                                 if( $c == ">" ) {
2103                                         $stack--;
2104                                         if( $stack < 0 ) {
2105                                                 wfDebug( "Invalid input in $fname; too many close tags\n" );
2106                                                 wfProfileOut( $fname );
2107                                                 return false;
2108                                         }
2109                                         $state = MW_COLON_STATE_TEXT;
2110                                 }
2111                                 break;
2112                         case MW_COLON_STATE_TAGSLASH:
2113                                 if( $c == ">" ) {
2114                                         // Yes, a self-closed tag <blah/>
2115                                         $state = MW_COLON_STATE_TEXT;
2116                                 } else {
2117                                         // Probably we're jumping the gun, and this is an attribute
2118                                         $state = MW_COLON_STATE_TAG;
2119                                 }
2120                                 break;
2121                         case 5: // MW_COLON_STATE_COMMENT:
2122                                 if( $c == "-" ) {
2123                                         $state = MW_COLON_STATE_COMMENTDASH;
2124                                 }
2125                                 break;
2126                         case MW_COLON_STATE_COMMENTDASH:
2127                                 if( $c == "-" ) {
2128                                         $state = MW_COLON_STATE_COMMENTDASHDASH;
2129                                 } else {
2130                                         $state = MW_COLON_STATE_COMMENT;
2131                                 }
2132                                 break;
2133                         case MW_COLON_STATE_COMMENTDASHDASH:
2134                                 if( $c == ">" ) {
2135                                         $state = MW_COLON_STATE_TEXT;
2136                                 } else {
2137                                         $state = MW_COLON_STATE_COMMENT;
2138                                 }
2139                                 break;
2140                         default:
2141                                 throw new MWException( "State machine error in $fname" );
2142                         }
2143                 }
2144                 if( $stack > 0 ) {
2145                         wfDebug( "Invalid input in $fname; not enough close tags (stack $stack, state $state)\n" );
2146                         return false;
2147                 }
2148                 wfProfileOut( $fname );
2149                 return false;
2150         }
2151
2152         /**
2153          * Return value of a magic variable (like PAGENAME)
2154          *
2155          * @private
2156          */
2157         function getVariableValue( $index ) {
2158                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2159
2160                 /**
2161                  * Some of these require message or data lookups and can be
2162                  * expensive to check many times.
2163                  */
2164                 static $varCache = array();
2165                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$varCache ) ) )
2166                         if ( isset( $varCache[$index] ) )
2167                                 return $varCache[$index];
2168
2169                 $ts = time();
2170                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2171
2172                 switch ( $index ) {
2173                         case MAG_CURRENTMONTH:
2174                                 return $varCache[$index] = $wgContLang->formatNum( date( 'm', $ts ) );
2175                         case MAG_CURRENTMONTHNAME:
2176                                 return $varCache[$index] = $wgContLang->getMonthName( date( 'n', $ts ) );
2177                         case MAG_CURRENTMONTHNAMEGEN:
2178                                 return $varCache[$index] = $wgContLang->getMonthNameGen( date( 'n', $ts ) );
2179                         case MAG_CURRENTMONTHABBREV:
2180                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( date( 'n', $ts ) );
2181                         case MAG_CURRENTDAY:
2182                                 return $varCache[$index] = $wgContLang->formatNum( date( 'j', $ts ) );
2183                         case MAG_CURRENTDAY2:
2184                                 return $varCache[$index] = $wgContLang->formatNum( date( 'd', $ts ) );
2185                         case MAG_PAGENAME:
2186                                 return $this->mTitle->getText();
2187                         case MAG_PAGENAMEE:
2188                                 return $this->mTitle->getPartialURL();
2189                         case MAG_FULLPAGENAME:
2190                                 return $this->mTitle->getPrefixedText();
2191                         case MAG_FULLPAGENAMEE:
2192                                 return $this->mTitle->getPrefixedURL();
2193                         case MAG_SUBPAGENAME:
2194                                 return $this->mTitle->getSubpageText();
2195                         case MAG_SUBPAGENAMEE:
2196                                 return $this->mTitle->getSubpageUrlForm();
2197                         case MAG_BASEPAGENAME:
2198                                 return $this->mTitle->getBaseText();
2199                         case MAG_BASEPAGENAMEE:
2200                                 return wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2201                         case MAG_TALKPAGENAME:
2202                                 if( $this->mTitle->canTalk() ) {
2203                                         $talkPage = $this->mTitle->getTalkPage();
2204                                         return $talkPage->getPrefixedText();
2205                                 } else {
2206                                         return '';
2207                                 }
2208                         case MAG_TALKPAGENAMEE:
2209                                 if( $this->mTitle->canTalk() ) {
2210                                         $talkPage = $this->mTitle->getTalkPage();
2211                                         return $talkPage->getPrefixedUrl();
2212                                 } else {
2213                                         return '';
2214                                 }
2215                         case MAG_SUBJECTPAGENAME:
2216                                 $subjPage = $this->mTitle->getSubjectPage();
2217                                 return $subjPage->getPrefixedText();
2218                         case MAG_SUBJECTPAGENAMEE:
2219                                 $subjPage = $this->mTitle->getSubjectPage();
2220                                 return $subjPage->getPrefixedUrl();
2221                         case MAG_REVISIONID:
2222                                 return $this->mRevisionId;
2223                         case MAG_NAMESPACE:
2224                                 return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2225                         case MAG_NAMESPACEE:
2226                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2227                         case MAG_TALKSPACE:
2228                                 return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2229                         case MAG_TALKSPACEE:
2230                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2231                         case MAG_SUBJECTSPACE:
2232                                 return $this->mTitle->getSubjectNsText();
2233                         case MAG_SUBJECTSPACEE:
2234                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2235                         case MAG_CURRENTDAYNAME:
2236                                 return $varCache[$index] = $wgContLang->getWeekdayName( date( 'w', $ts ) + 1 );
2237                         case MAG_CURRENTYEAR:
2238                                 return $varCache[$index] = $wgContLang->formatNum( date( 'Y', $ts ), true );
2239                         case MAG_CURRENTTIME:
2240                                 return $varCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2241                         case MAG_CURRENTWEEK:
2242                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2243                                 // int to remove the padding
2244                                 return $varCache[$index] = $wgContLang->formatNum( (int)date( 'W', $ts ) );
2245                         case MAG_CURRENTDOW:
2246                                 return $varCache[$index] = $wgContLang->formatNum( date( 'w', $ts ) );
2247                         case MAG_NUMBEROFARTICLES:
2248                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfArticles() );
2249                         case MAG_NUMBEROFFILES:
2250                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfFiles() );
2251                         case MAG_NUMBEROFUSERS:
2252                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfUsers() );
2253                         case MAG_NUMBEROFPAGES:
2254                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfPages() );
2255                         case MAG_NUMBEROFADMINS:
2256                                 return $varCache[$index]  = $wgContLang->formatNum( wfNumberOfAdmins() );
2257                         case MAG_CURRENTTIMESTAMP:
2258                                 return $varCache[$index] = wfTimestampNow();
2259                         case MAG_CURRENTVERSION:
2260                                 global $wgVersion;
2261                                 return $wgVersion;
2262                         case MAG_SITENAME:
2263                                 return $wgSitename;
2264                         case MAG_SERVER:
2265                                 return $wgServer;
2266                         case MAG_SERVERNAME:
2267                                 return $wgServerName;
2268                         case MAG_SCRIPTPATH:
2269                                 return $wgScriptPath;
2270                         case MAG_DIRECTIONMARK:
2271                                 return $wgContLang->getDirMark();
2272                         case MAG_CONTENTLANGUAGE:
2273                                 global $wgContLanguageCode;
2274                                 return $wgContLanguageCode;
2275                         default:
2276                                 $ret = null;
2277                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$varCache, &$index, &$ret ) ) )
2278                                         return $ret;
2279                                 else
2280                                         return null;
2281                 }
2282         }
2283
2284         /**
2285          * initialise the magic variables (like CURRENTMONTHNAME)
2286          *
2287          * @private
2288          */
2289         function initialiseVariables() {
2290                 $fname = 'Parser::initialiseVariables';
2291                 wfProfileIn( $fname );
2292                 global $wgVariableIDs;
2293                 $this->mVariables = array();
2294                 foreach ( $wgVariableIDs as $id ) {
2295                         $mw =& MagicWord::get( $id );
2296                         $mw->addToArray( $this->mVariables, $id );
2297                 }
2298                 wfProfileOut( $fname );
2299         }
2300
2301         /**
2302          * parse any parentheses in format ((title|part|part))
2303          * and call callbacks to get a replacement text for any found piece
2304          *
2305          * @param string $text The text to parse
2306          * @param array $callbacks rules in form:
2307          *     '{' => array(                            # opening parentheses
2308          *                                      'end' => '}',   # closing parentheses
2309          *                                      'cb' => array(2 => callback,    # replacement callback to call if {{..}} is found
2310          *                                                                4 => callback         # replacement callback to call if {{{{..}}}} is found
2311          *                                                                )
2312          *                                      )
2313          * @private
2314          */
2315         function replace_callback ($text, $callbacks) {
2316                 $openingBraceStack = array();   # this array will hold a stack of parentheses which are not closed yet
2317                 $lastOpeningBrace = -1;                 # last not closed parentheses
2318
2319                 for ($i = 0; $i < strlen($text); $i++) {
2320                         # check for any opening brace
2321                         $rule = null;
2322                         $nextPos = -1;
2323                         foreach ($callbacks as $key => $value) {
2324                                 $pos = strpos ($text, $key, $i);
2325                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)) {
2326                                         $rule = $value;
2327                                         $nextPos = $pos;
2328                                 }
2329                         }
2330
2331                         if ($lastOpeningBrace >= 0) {
2332                                 $pos = strpos ($text, $openingBraceStack[$lastOpeningBrace]['braceEnd'], $i);
2333
2334                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)){
2335                                         $rule = null;
2336                                         $nextPos = $pos;
2337                                 }
2338
2339                                 $pos = strpos ($text, '|', $i);
2340
2341                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)){
2342                                         $rule = null;
2343                                         $nextPos = $pos;
2344                                 }
2345                         }
2346
2347                         if ($nextPos == -1)
2348                                 break;
2349
2350                         $i = $nextPos;
2351
2352                         # found openning brace, lets add it to parentheses stack
2353                         if (null != $rule) {
2354                                 $piece = array('brace' => $text[$i],
2355                                                            'braceEnd' => $rule['end'],
2356                                                            'count' => 1,
2357                                                            'title' => '',
2358                                                            'parts' => null);
2359
2360                                 # count openning brace characters
2361                                 while ($i+1 < strlen($text) && $text[$i+1] == $piece['brace']) {
2362                                         $piece['count']++;
2363                                         $i++;
2364                                 }
2365
2366                                 $piece['startAt'] = $i+1;
2367                                 $piece['partStart'] = $i+1;
2368
2369                                 # we need to add to stack only if openning brace count is enough for any given rule
2370                                 foreach ($rule['cb'] as $cnt => $fn) {
2371                                         if ($piece['count'] >= $cnt) {
2372                                                 $lastOpeningBrace ++;
2373                                                 $openingBraceStack[$lastOpeningBrace] = $piece;
2374                                                 break;
2375                                         }
2376                                 }
2377
2378                                 continue;
2379                         }
2380                         else if ($lastOpeningBrace >= 0) {
2381                                 # first check if it is a closing brace
2382                                 if ($openingBraceStack[$lastOpeningBrace]['braceEnd'] == $text[$i]) {
2383                                         # lets check if it is enough characters for closing brace
2384                                         $count = 1;
2385                                         while ($i+$count < strlen($text) && $text[$i+$count] == $text[$i])
2386                                                 $count++;
2387
2388                                         # if there are more closing parentheses than opening ones, we parse less
2389                                         if ($openingBraceStack[$lastOpeningBrace]['count'] < $count)
2390                                                 $count = $openingBraceStack[$lastOpeningBrace]['count'];
2391
2392                                         # check for maximum matching characters (if there are 5 closing characters, we will probably need only 3 - depending on the rules)
2393                                         $matchingCount = 0;
2394                                         $matchingCallback = null;
2395                                         foreach ($callbacks[$openingBraceStack[$lastOpeningBrace]['brace']]['cb'] as $cnt => $fn) {
2396                                                 if ($count >= $cnt && $matchingCount < $cnt) {
2397                                                         $matchingCount = $cnt;
2398                                                         $matchingCallback = $fn;
2399                                                 }
2400                                         }
2401
2402                                         if ($matchingCount == 0) {
2403                                                 $i += $count - 1;
2404                                                 continue;
2405                                         }
2406
2407                                         # lets set a title or last part (if '|' was found)
2408                                         if (null === $openingBraceStack[$lastOpeningBrace]['parts'])
2409                                                 $openingBraceStack[$lastOpeningBrace]['title'] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2410                                         else
2411                                                 $openingBraceStack[$lastOpeningBrace]['parts'][] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2412
2413                                         $pieceStart = $openingBraceStack[$lastOpeningBrace]['startAt'] - $matchingCount;
2414                                         $pieceEnd = $i + $matchingCount;
2415
2416                                         if( is_callable( $matchingCallback ) ) {
2417                                                 $cbArgs = array (
2418                                                                                  'text' => substr($text, $pieceStart, $pieceEnd - $pieceStart),
2419                                                                                  'title' => trim($openingBraceStack[$lastOpeningBrace]['title']),
2420                                                                                  'parts' => $openingBraceStack[$lastOpeningBrace]['parts'],
2421                                                                                  'lineStart' => (($pieceStart > 0) && ($text[$pieceStart-1] == "\n")),
2422                                                                                  );
2423                                                 # finally we can call a user callback and replace piece of text
2424                                                 $replaceWith = call_user_func( $matchingCallback, $cbArgs );
2425                                                 $text = substr($text, 0, $pieceStart) . $replaceWith . substr($text, $pieceEnd);
2426                                                 $i = $pieceStart + strlen($replaceWith) - 1;
2427                                         }
2428                                         else {
2429                                                 # null value for callback means that parentheses should be parsed, but not replaced
2430                                                 $i += $matchingCount - 1;
2431                                         }
2432
2433                                         # reset last openning parentheses, but keep it in case there are unused characters
2434                                         $piece = array('brace' => $openingBraceStack[$lastOpeningBrace]['brace'],
2435                                                                    'braceEnd' => $openingBraceStack[$lastOpeningBrace]['braceEnd'],
2436                                                                    'count' => $openingBraceStack[$lastOpeningBrace]['count'],
2437                                                                    'title' => '',
2438                                                                    'parts' => null,
2439                                                                    'startAt' => $openingBraceStack[$lastOpeningBrace]['startAt']);
2440                                         $openingBraceStack[$lastOpeningBrace--] = null;
2441
2442                                         if ($matchingCount < $piece['count']) {
2443                                                 $piece['count'] -= $matchingCount;
2444                                                 $piece['startAt'] -= $matchingCount;
2445                                                 $piece['partStart'] = $piece['startAt'];
2446                                                 # do we still qualify for any callback with remaining count?
2447                                                 foreach ($callbacks[$piece['brace']]['cb'] as $cnt => $fn) {
2448                                                         if ($piece['count'] >= $cnt) {
2449                                                                 $lastOpeningBrace ++;
2450                                                                 $openingBraceStack[$lastOpeningBrace] = $piece;
2451                                                                 break;
2452                                                         }
2453                                                 }
2454                                         }
2455                                         continue;
2456                                 }
2457
2458                                 # lets set a title if it is a first separator, or next part otherwise
2459                                 if ($text[$i] == '|') {
2460                                         if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2461                                                 $openingBraceStack[$lastOpeningBrace]['title'] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2462                                                 $openingBraceStack[$lastOpeningBrace]['parts'] = array();
2463                                         }
2464                                         else
2465                                                 $openingBraceStack[$lastOpeningBrace]['parts'][] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2466
2467                                         $openingBraceStack[$lastOpeningBrace]['partStart'] = $i + 1;
2468                                 }
2469                         }
2470                 }
2471
2472                 return $text;
2473         }
2474
2475         /**
2476          * Replace magic variables, templates, and template arguments
2477          * with the appropriate text. Templates are substituted recursively,
2478          * taking care to avoid infinite loops.
2479          *
2480          * Note that the substitution depends on value of $mOutputType:
2481          *  OT_WIKI: only {{subst:}} templates
2482          *  OT_MSG: only magic variables
2483          *  OT_HTML: all templates and magic variables
2484          *
2485          * @param string $tex The text to transform
2486          * @param array $args Key-value pairs representing template parameters to substitute
2487          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2488          * @private
2489          */
2490         function replaceVariables( $text, $args = array(), $argsOnly = false ) {
2491                 # Prevent too big inclusions
2492                 if( strlen( $text ) > MAX_INCLUDE_SIZE ) {
2493                         return $text;
2494                 }
2495
2496                 $fname = 'Parser::replaceVariables';
2497                 wfProfileIn( $fname );
2498
2499                 # This function is called recursively. To keep track of arguments we need a stack:
2500                 array_push( $this->mArgStack, $args );
2501
2502                 $braceCallbacks = array();
2503                 if ( !$argsOnly ) {
2504                         $braceCallbacks[2] = array( &$this, 'braceSubstitution' );
2505                 }
2506                 if ( $this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI ) {
2507                         $braceCallbacks[3] = array( &$this, 'argSubstitution' );
2508                 }
2509                 $callbacks = array();
2510                 $callbacks['{'] = array('end' => '}', 'cb' => $braceCallbacks);
2511                 $callbacks['['] = array('end' => ']', 'cb' => array(2=>null));
2512                 $text = $this->replace_callback ($text, $callbacks);
2513
2514                 array_pop( $this->mArgStack );
2515
2516                 wfProfileOut( $fname );
2517                 return $text;
2518         }
2519
2520         /**
2521          * Replace magic variables
2522          * @private
2523          */
2524         function variableSubstitution( $matches ) {
2525                 $fname = 'Parser::variableSubstitution';
2526                 $varname = $matches[1];
2527                 wfProfileIn( $fname );
2528                 if ( !$this->mVariables ) {
2529                         $this->initialiseVariables();
2530                 }
2531                 $skip = false;
2532                 if ( $this->mOutputType == OT_WIKI ) {
2533                         # Do only magic variables prefixed by SUBST
2534                         $mwSubst =& MagicWord::get( MAG_SUBST );
2535                         if (!$mwSubst->matchStartAndRemove( $varname ))
2536                                 $skip = true;
2537                         # Note that if we don't substitute the variable below,
2538                         # we don't remove the {{subst:}} magic word, in case
2539                         # it is a template rather than a magic variable.
2540                 }
2541                 if ( !$skip && array_key_exists( $varname, $this->mVariables ) ) {
2542                         $id = $this->mVariables[$varname];
2543                         $text = $this->getVariableValue( $id );
2544                         $this->mOutput->mContainsOldMagic = true;
2545                 } else {
2546                         $text = $matches[0];
2547                 }
2548                 wfProfileOut( $fname );
2549                 return $text;
2550         }
2551
2552         # Split template arguments
2553         function getTemplateArgs( $argsString ) {
2554                 if ( $argsString === '' ) {
2555                         return array();
2556                 }
2557
2558                 $args = explode( '|', substr( $argsString, 1 ) );
2559
2560                 # If any of the arguments contains a '[[' but no ']]', it needs to be
2561                 # merged with the next arg because the '|' character between belongs
2562                 # to the link syntax and not the template parameter syntax.
2563                 $argc = count($args);
2564
2565                 for ( $i = 0; $i < $argc-1; $i++ ) {
2566                         if ( substr_count ( $args[$i], '[[' ) != substr_count ( $args[$i], ']]' ) ) {
2567                                 $args[$i] .= '|'.$args[$i+1];
2568                                 array_splice($args, $i+1, 1);
2569                                 $i--;
2570                                 $argc--;
2571                         }
2572                 }
2573
2574                 return $args;
2575         }
2576
2577         /**
2578          * Return the text of a template, after recursively
2579          * replacing any variables or templates within the template.
2580          *
2581          * @param array $piece The parts of the template
2582          *  $piece['text']: matched text
2583          *  $piece['title']: the title, i.e. the part before the |
2584          *  $piece['parts']: the parameter array
2585          * @return string the text of the template
2586          * @private
2587          */
2588         function braceSubstitution( $piece ) {
2589                 global $wgContLang, $wgLang, $wgAllowDisplayTitle, $action;
2590                 $fname = 'Parser::braceSubstitution';
2591                 wfProfileIn( $fname );
2592
2593                 # Flags
2594                 $found = false;             # $text has been filled
2595                 $nowiki = false;            # wiki markup in $text should be escaped
2596                 $noparse = false;           # Unsafe HTML tags should not be stripped, etc.
2597                 $noargs = false;            # Don't replace triple-brace arguments in $text
2598                 $replaceHeadings = false;   # Make the edit section links go to the template not the article
2599                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2600                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2601
2602                 # Title object, where $text came from
2603                 $title = NULL;
2604
2605                 $linestart = '';
2606
2607                 # $part1 is the bit before the first |, and must contain only title characters
2608                 # $args is a list of arguments, starting from index 0, not including $part1
2609
2610                 $part1 = $piece['title'];
2611                 # If the third subpattern matched anything, it will start with |
2612
2613                 if (null == $piece['parts']) {
2614                         $replaceWith = $this->variableSubstitution (array ($piece['text'], $piece['title']));
2615                         if ($replaceWith != $piece['text']) {
2616                                 $text = $replaceWith;
2617                                 $found = true;
2618                                 $noparse = true;
2619                                 $noargs = true;
2620                         }
2621                 }
2622
2623                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2624                 $argc = count( $args );
2625
2626                 # SUBST
2627                 if ( !$found ) {
2628                         $mwSubst =& MagicWord::get( MAG_SUBST );
2629                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor ($this->mOutputType == OT_WIKI) ) {
2630                                 # One of two possibilities is true:
2631                                 # 1) Found SUBST but not in the PST phase
2632                                 # 2) Didn't find SUBST and in the PST phase
2633                                 # In either case, return without further processing
2634                                 $text = $piece['text'];
2635                                 $found = true;
2636                                 $noparse = true;
2637                                 $noargs = true;
2638                         }
2639                 }
2640
2641                 # MSG, MSGNW, INT and RAW
2642                 if ( !$found ) {
2643                         # Check for MSGNW:
2644                         $mwMsgnw =& MagicWord::get( MAG_MSGNW );
2645                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2646                                 $nowiki = true;
2647                         } else {
2648                                 # Remove obsolete MSG:
2649                                 $mwMsg =& MagicWord::get( MAG_MSG );
2650                                 $mwMsg->matchStartAndRemove( $part1 );
2651                         }
2652
2653                         # Check for RAW:
2654                         $mwRaw =& MagicWord::get( MAG_RAW );
2655                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2656                                 $forceRawInterwiki = true;
2657                         }
2658
2659                         # Check if it is an internal message
2660                         $mwInt =& MagicWord::get( MAG_INT );
2661                         if ( $mwInt->matchStartAndRemove( $part1 ) ) {
2662                                 if ( $this->incrementIncludeCount( 'int:'.$part1 ) ) {
2663                                         $text = $linestart . wfMsgReal( $part1, $args, true );
2664                                         $found = true;
2665                                 }
2666                         }
2667                 }
2668
2669                 # NS
2670                 if ( !$found ) {
2671                         # Check for NS: (namespace expansion)
2672                         $mwNs = MagicWord::get( MAG_NS );
2673                         if ( $mwNs->matchStartAndRemove( $part1 ) ) {
2674                                 if ( intval( $part1 ) || $part1 == "0" ) {
2675                                         $text = $linestart . $wgContLang->getNsText( intval( $part1 ) );
2676                                         $found = true;
2677                                 } else {
2678                                         $param = str_replace( ' ', '_', strtolower( $part1 ) );
2679                                         $index = Namespace::getCanonicalIndex( strtolower( $param ) );
2680                                         if ( !is_null( $index ) ) {
2681                                                 $text = $linestart . $wgContLang->getNsText( $index );
2682                                                 $found = true;
2683                                         }
2684                                 }
2685                         }
2686                 }
2687
2688                 # URLENCODE
2689                 if( !$found ) {
2690                         $urlencode =& MagicWord::get( MAG_URLENCODE );
2691                         if( $urlencode->matchStartAndRemove( $part1 ) ) {
2692                                 $text = $linestart . urlencode( $part1 );
2693                                 $found = true;
2694                         }
2695                 }
2696
2697                 # LCFIRST, UCFIRST, LC and UC
2698                 if ( !$found ) {
2699                         $lcfirst =& MagicWord::get( MAG_LCFIRST );
2700                         $ucfirst =& MagicWord::get( MAG_UCFIRST );
2701                         $lc =& MagicWord::get( MAG_LC );
2702                         $uc =& MagicWord::get( MAG_UC );
2703                         if ( $lcfirst->matchStartAndRemove( $part1 ) ) {
2704                                 $text = $linestart . $wgContLang->lcfirst( $part1 );
2705                                 $found = true;
2706                         } else if ( $ucfirst->matchStartAndRemove( $part1 ) ) {
2707                                 $text = $linestart . $wgContLang->ucfirst( $part1 );
2708                                 $found = true;
2709                         } else if ( $lc->matchStartAndRemove( $part1 ) ) {
2710                                 $text = $linestart . $wgContLang->lc( $part1 );
2711                                 $found = true;
2712                         } else if ( $uc->matchStartAndRemove( $part1 ) ) {
2713                                  $text = $linestart . $wgContLang->uc( $part1 );
2714                                  $found = true;
2715                         }
2716                 }
2717
2718                 # LOCALURL and FULLURL
2719                 if ( !$found ) {
2720                         $mwLocal =& MagicWord::get( MAG_LOCALURL );
2721                         $mwLocalE =& MagicWord::get( MAG_LOCALURLE );
2722                         $mwFull =& MagicWord::get( MAG_FULLURL );
2723                         $mwFullE =& MagicWord::get( MAG_FULLURLE );
2724
2725
2726                         if ( $mwLocal->matchStartAndRemove( $part1 ) ) {
2727                                 $func = 'getLocalURL';
2728                         } elseif ( $mwLocalE->matchStartAndRemove( $part1 ) ) {
2729                                 $func = 'escapeLocalURL';
2730                         } elseif ( $mwFull->matchStartAndRemove( $part1 ) ) {
2731                                 $func = 'getFullURL';
2732                         } elseif ( $mwFullE->matchStartAndRemove( $part1 ) ) {
2733                                 $func = 'escapeFullURL';
2734                         } else {
2735                                 $func = false;
2736                         }
2737
2738                         if ( $func !== false ) {
2739                                 $title = Title::newFromText( $part1 );
2740                                 # Due to order of execution of a lot of bits, the values might be encoded
2741                                 # before arriving here; if that's true, then the title can't be created
2742                                 # and the variable will fail. If we can't get a decent title from the first
2743                                 # attempt, url-decode and try for a second.
2744                                 if( is_null( $title ) )
2745                                         $title = Title::newFromUrl( urldecode( $part1 ) );
2746                                 if ( !is_null( $title ) ) {
2747                                         if ( $argc > 0 ) {
2748                                                 $text = $linestart . $title->$func( $args[0] );
2749                                         } else {
2750                                                 $text = $linestart . $title->$func();
2751                                         }
2752                                         $found = true;
2753                                 }
2754                         }
2755                 }
2756
2757                 $lang = $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
2758                 if ( !$found && $argc == 1 ) {
2759                         $mwGrammar =& MagicWord::get( MAG_FORMATNUM );
2760                         if ( $mwGrammar->matchStartAndRemove( $part1 ) ) {
2761                                 $text = $linestart . $lang->formatNum( $args[0] );
2762                                 $found = true;
2763                         }
2764                 }
2765
2766
2767                 # GRAMMAR
2768                 if ( !$found && $argc == 1 ) {
2769                         $mwGrammar =& MagicWord::get( MAG_GRAMMAR );
2770                         if ( $mwGrammar->matchStartAndRemove( $part1 ) ) {
2771                                 $text = $linestart . $lang->convertGrammar( $args[0], $part1 );
2772                                 $found = true;
2773                         }
2774                 }
2775
2776                 # PLURAL
2777                 if ( !$found && $argc >= 2 ) {
2778                         $mwPluralForm =& MagicWord::get( MAG_PLURAL );
2779                         if ( $mwPluralForm->matchStartAndRemove( $part1 ) ) {
2780                                 while ( count($args) < 5 ) { $args[] = $args[count($args)-1]; }
2781                                 $text = $linestart . $lang->convertPlural( $part1, $args[0], $args[1],
2782                                         $args[2], $args[3], $args[4]);
2783                                 $found = true;
2784                         }
2785                 }
2786
2787                 # DISPLAYTITLE
2788                 if ( !$found && $argc == 1 && $wgAllowDisplayTitle ) {
2789                         $mwDT =& MagicWord::get( MAG_DISPLAYTITLE );
2790                         if ( $mwDT->matchStartAndRemove( $part1 ) ) {
2791
2792                                 # Set title in parser output object
2793                                 $param = $args[0];
2794                                 $parserOptions = new ParserOptions;
2795                                 $local_parser = new Parser ();
2796                                 $t2 = $local_parser->parse ( $param, $this->mTitle, $parserOptions, false );
2797                                 $this->mOutput->mHTMLtitle = $t2->GetText();
2798
2799                                 # Add subtitle
2800                                 $t = $this->mTitle->getPrefixedText();
2801                                 $this->mOutput->mSubtitle .= wfMsg('displaytitle', $t);
2802                                 $text = "" ;
2803                                 $found = true ;
2804                         }
2805                 }
2806
2807                 # NUMBEROFPAGES, NUMBEROFUSERS, NUMBEROFARTICLES, and NUMBEROFFILES
2808                 if( !$found ) {
2809                         $mwWordsToCheck = array( MAG_NUMBEROFPAGES => 'wfNumberOfPages',
2810                                                                          MAG_NUMBEROFUSERS => 'wfNumberOfUsers',
2811                                                                          MAG_NUMBEROFARTICLES => 'wfNumberOfArticles',
2812                                                                          MAG_NUMBEROFFILES => 'wfNumberOfFiles',
2813                                                                          MAG_NUMBEROFADMINS => 'wfNumberOfAdmins' );
2814                         foreach( $mwWordsToCheck as $word => $func ) {
2815                                 $mwCurrentWord =& MagicWord::get( $word );
2816                                 if( $mwCurrentWord->matchStartAndRemove( $part1 ) ) {
2817                                         $mwRawSuffix =& MagicWord::get( MAG_RAWSUFFIX );
2818                                         if( isset( $args[0] ) && $mwRawSuffix->match( $args[0] ) ) {
2819                                                 # Raw and unformatted
2820                                                 $text = $linestart . call_user_func( $func );
2821                                         } else {
2822                                                 # Formatted according to the content default
2823                                                 $text = $linestart . $wgContLang->formatNum( call_user_func( $func ) );
2824                                         }
2825                                         $found = true;
2826                                 }
2827                         }
2828                 }
2829
2830                 # PAGESINNAMESPACE
2831                 if( !$found ) {
2832                         $mwPagesInNs =& MagicWord::get( MAG_PAGESINNAMESPACE );
2833                         if( $mwPagesInNs->matchStartAndRemove( $part1 ) ) {
2834                                 $found = true;
2835                                 $count = wfPagesInNs( intval( $part1 ) );
2836                                 $mwRawSuffix =& MagicWord::get( MAG_RAWSUFFIX );
2837                                 if( isset( $args[0] ) && $mwRawSuffix->match( $args[0] ) ) {
2838                                         $text = $linestart . $count;
2839                                 } else {
2840                                         $text = $linestart . $wgContLang->formatNum( $count );
2841                                 }
2842                         }
2843                 }
2844
2845                 # #LANGUAGE:
2846                 if( !$found ) {
2847                         $mwLanguage =& MagicWord::get( MAG_LANGUAGE );
2848                         if( $mwLanguage->matchStartAndRemove( $part1 ) ) {
2849                                 $lang = $wgContLang->getLanguageName( strtolower( $part1 ) );
2850                                 $text = $linestart . ( $lang != '' ? $lang : $part1 );
2851                                 $found = true;
2852                         }
2853                 }
2854
2855                 # Extensions
2856                 if ( !$found && substr( $part1, 0, 1 ) == '#' ) {
2857                         $colonPos = strpos( $part1, ':' );
2858                         if ( $colonPos !== false ) {
2859                                 $function = strtolower( substr( $part1, 1, $colonPos - 1 ) );
2860                                 if ( isset( $this->mFunctionSynonyms[$function] ) ) {
2861                                         $function = $this->mFunctionSynonyms[$function];
2862                                 }
2863                                 if ( isset( $this->mFunctionHooks[$function] ) ) {
2864                                         $funcArgs = array_map( 'trim', $args );
2865                                         $funcArgs = array_merge( array( &$this, trim( substr( $part1, $colonPos + 1 ) ) ), $funcArgs );
2866                                         $result = call_user_func_array( $this->mFunctionHooks[$function], $funcArgs );
2867                                         $found = true;
2868
2869                                         // The text is usually already parsed, doesn't need triple-brace tags expanded, etc.
2870                                         //$noargs = true;
2871                                         //$noparse = true;
2872
2873                                         if ( is_array( $result ) ) {
2874                                                 $text = $linestart . $result[0];
2875                                                 unset( $result[0] );
2876
2877                                                 // Extract flags into the local scope
2878                                                 // This allows callers to set flags such as nowiki, noparse, found, etc.
2879                                                 extract( $result );
2880                                         } else {
2881                                                 $text = $linestart . $result;
2882                                         }
2883                                 }
2884                         }
2885                 }
2886
2887                 # Template table test
2888
2889                 # Did we encounter this template already? If yes, it is in the cache
2890                 # and we need to check for loops.
2891                 if ( !$found && isset( $this->mTemplates[$piece['title']] ) ) {
2892                         $found = true;
2893
2894                         # Infinite loop test
2895                         if ( isset( $this->mTemplatePath[$part1] ) ) {
2896                                 $noparse = true;
2897                                 $noargs = true;
2898                                 $found = true;
2899                                 $text = $linestart .
2900                                         '{{' . $part1 . '}}' .
2901                                         '<!-- WARNING: template loop detected -->';
2902                                 wfDebug( "$fname: template loop broken at '$part1'\n" );
2903                         } else {
2904                                 # set $text to cached message.
2905                                 $text = $linestart . $this->mTemplates[$piece['title']];
2906                         }
2907                 }
2908
2909                 # Load from database
2910                 $lastPathLevel = $this->mTemplatePath;
2911                 if ( !$found ) {
2912                         $ns = NS_TEMPLATE;
2913                         # declaring $subpage directly in the function call
2914                         # does not work correctly with references and breaks
2915                         # {{/subpage}}-style inclusions
2916                         $subpage = '';
2917                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
2918                         if ($subpage !== '') {
2919                                 $ns = $this->mTitle->getNamespace();
2920                         }
2921                         $title = Title::newFromText( $part1, $ns );
2922
2923
2924                         if ( !is_null( $title ) ) {
2925                                 $checkVariantLink = sizeof($wgContLang->getVariants())>1;
2926                                 # Check for language variants if the template is not found
2927                                 if($checkVariantLink && $title->getArticleID() == 0){
2928                                         $wgContLang->findVariantLink($part1, $title);
2929                                 }
2930
2931                                 if ( !$title->isExternal() ) {
2932                                         # Check for excessive inclusion
2933                                         $dbk = $title->getPrefixedDBkey();
2934                                         if ( $this->incrementIncludeCount( $dbk ) ) {
2935                                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->mOutputType != OT_WIKI ) {
2936                                                         $text = SpecialPage::capturePath( $title );
2937                                                         if ( is_string( $text ) ) {
2938                                                                 $found = true;
2939                                                                 $noparse = true;
2940                                                                 $noargs = true;
2941                                                                 $isHTML = true;
2942                                                                 $this->disableCache();
2943                                                         }
2944                                                 } else {
2945                                                         $articleContent = $this->fetchTemplate( $title );
2946                                                         if ( $articleContent !== false ) {
2947                                                                 $found = true;
2948                                                                 $text = $articleContent;
2949                                                                 $replaceHeadings = true;
2950                                                         }
2951                                                 }
2952                                         }
2953
2954                                         # If the title is valid but undisplayable, make a link to it
2955                                         if ( $this->mOutputType == OT_HTML && !$found ) {
2956                                                 $text = '[['.$title->getPrefixedText().']]';
2957                                                 $found = true;
2958                                         }
2959                                 } elseif ( $title->isTrans() ) {
2960                                         // Interwiki transclusion
2961                                         if ( $this->mOutputType == OT_HTML && !$forceRawInterwiki ) {
2962                                                 $text = $this->interwikiTransclude( $title, 'render' );
2963                                                 $isHTML = true;
2964                                                 $noparse = true;
2965                                         } else {
2966                                                 $text = $this->interwikiTransclude( $title, 'raw' );
2967                                                 $replaceHeadings = true;
2968                                         }
2969                                         $found = true;
2970                                 }
2971
2972                                 # Template cache array insertion
2973                                 # Use the original $piece['title'] not the mangled $part1, so that
2974                                 # modifiers such as RAW: produce separate cache entries
2975                                 if( $found ) {
2976                                         if( $isHTML ) {
2977                                                 // A special page; don't store it in the template cache.
2978                                         } else {
2979                                                 $this->mTemplates[$piece['title']] = $text;
2980                                         }
2981                                         $text = $linestart . $text;
2982                                 }
2983                         }
2984                 }
2985
2986                 # Recursive parsing, escaping and link table handling
2987                 # Only for HTML output
2988                 if ( $nowiki && $found && $this->mOutputType == OT_HTML ) {
2989                         $text = wfEscapeWikiText( $text );
2990                 } elseif ( ($this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI) && $found ) {
2991                         if ( $noargs ) {
2992                                 $assocArgs = array();
2993                         } else {
2994                                 # Clean up argument array
2995                                 $assocArgs = array();
2996                                 $index = 1;
2997                                 foreach( $args as $arg ) {
2998                                         $eqpos = strpos( $arg, '=' );
2999                                         if ( $eqpos === false ) {
3000                                                 $assocArgs[$index++] = $arg;
3001                                         } else {
3002                                                 $name = trim( substr( $arg, 0, $eqpos ) );
3003                                                 $value = trim( substr( $arg, $eqpos+1 ) );
3004                                                 if ( $value === false ) {
3005                                                         $value = '';
3006                                                 }
3007                                                 if ( $name !== false ) {
3008                                                         $assocArgs[$name] = $value;
3009                                                 }
3010                                         }
3011                                 }
3012
3013                                 # Add a new element to the templace recursion path
3014                                 $this->mTemplatePath[$part1] = 1;
3015                         }
3016
3017                         if ( !$noparse ) {
3018                                 # If there are any <onlyinclude> tags, only include them
3019                                 if ( in_string( '<onlyinclude>', $text ) && in_string( '</onlyinclude>', $text ) ) {
3020                                         preg_match_all( '/<onlyinclude>(.*?)\n?<\/onlyinclude>/s', $text, $m );
3021                                         $text = '';
3022                                         foreach ($m[1] as $piece)
3023                                                 $text .= $piece;
3024                                 }
3025                                 # Remove <noinclude> sections and <includeonly> tags
3026                                 $text = preg_replace( '/<noinclude>.*?<\/noinclude>/s', '', $text );
3027                                 $text = strtr( $text, array( '<includeonly>' => '' , '</includeonly>' => '' ) );
3028
3029                                 if( $this->mOutputType == OT_HTML ) {
3030                                         # Strip <nowiki>, <pre>, etc.
3031                                         $text = $this->strip( $text, $this->mStripState );
3032                                         $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'replaceVariables' ), $assocArgs );
3033                                 }
3034                                 $text = $this->replaceVariables( $text, $assocArgs );
3035
3036                                 # If the template begins with a table or block-level
3037                                 # element, it should be treated as beginning a new line.
3038                                 if (!$piece['lineStart'] && preg_match('/^({\\||:|;|#|\*)/', $text)) {
3039                                         $text = "\n" . $text;
3040                                 }
3041                         } elseif ( !$noargs ) {
3042                                 # $noparse and !$noargs
3043                                 # Just replace the arguments, not any double-brace items
3044                                 # This is used for rendered interwiki transclusion
3045                                 $text = $this->replaceVariables( $text, $assocArgs, true );
3046                         }
3047                 }
3048                 # Prune lower levels off the recursion check path
3049                 $this->mTemplatePath = $lastPathLevel;
3050
3051                 if ( !$found ) {
3052                         wfProfileOut( $fname );
3053                         return $piece['text'];
3054                 } else {
3055                         if ( $isHTML ) {
3056                                 # Replace raw HTML by a placeholder
3057                                 # Add a blank line preceding, to prevent it from mucking up
3058                                 # immediately preceding headings
3059                                 $text = "\n\n" . $this->insertStripItem( $text, $this->mStripState );
3060                         } else {
3061                                 # replace ==section headers==
3062                                 # XXX this needs to go away once we have a better parser.
3063                                 if ( $this->mOutputType != OT_WIKI && $replaceHeadings ) {
3064                                         if( !is_null( $title ) )
3065                                                 $encodedname = base64_encode($title->getPrefixedDBkey());
3066                                         else
3067                                                 $encodedname = base64_encode("");
3068                                         $m = preg_split('/(^={1,6}.*?={1,6}\s*?$)/m', $text, -1,
3069                                                 PREG_SPLIT_DELIM_CAPTURE);
3070                                         $text = '';
3071                                         $nsec = 0;
3072                                         for( $i = 0; $i < count($m); $i += 2 ) {
3073                                                 $text .= $m[$i];
3074                                                 if (!isset($m[$i + 1]) || $m[$i + 1] == "") continue;
3075                                                 $hl = $m[$i + 1];
3076                                                 if( strstr($hl, "<!--MWTEMPLATESECTION") ) {
3077                                                         $text .= $hl;
3078                                                         continue;
3079                                                 }
3080                                                 preg_match('/^(={1,6})(.*?)(={1,6})\s*?$/m', $hl, $m2);
3081                                                 $text .= $m2[1] . $m2[2] . "<!--MWTEMPLATESECTION="
3082                                                         . $encodedname . "&" . base64_encode("$nsec") . "-->" . $m2[3];
3083
3084                                                 $nsec++;
3085                                         }
3086                                 }
3087                         }
3088                 }
3089
3090                 # Prune lower levels off the recursion check path
3091                 $this->mTemplatePath = $lastPathLevel;
3092
3093                 if ( !$found ) {
3094                         wfProfileOut( $fname );
3095                         return $piece['text'];
3096                 } else {
3097                         wfProfileOut( $fname );
3098                         return $text;
3099                 }
3100         }
3101
3102         /**
3103          * Fetch the unparsed text of a template and register a reference to it.
3104          */
3105         function fetchTemplate( $title ) {
3106                 $text = false;
3107                 // Loop to fetch the article, with up to 1 redirect
3108                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3109                         $rev = Revision::newFromTitle( $title );
3110                         $this->mOutput->addTemplate( $title, $title->getArticleID() );
3111                         if ( !$rev ) {
3112                                 break;
3113                         }
3114                         $text = $rev->getText();
3115                         if ( $text === false ) {
3116                                 break;
3117                         }
3118                         // Redirect?
3119                         $title = Title::newFromRedirect( $text );
3120                 }
3121                 return $text;
3122         }
3123
3124         /**
3125          * Transclude an interwiki link.
3126          */
3127         function interwikiTransclude( $title, $action ) {
3128                 global $wgEnableScaryTranscluding, $wgCanonicalNamespaceNames;
3129
3130                 if (!$wgEnableScaryTranscluding)
3131                         return wfMsg('scarytranscludedisabled');
3132
3133                 // The namespace will actually only be 0 or 10, depending on whether there was a leading :
3134                 // But we'll handle it generally anyway
3135                 if ( $title->getNamespace() ) {
3136                         // Use the canonical namespace, which should work anywhere
3137                         $articleName = $wgCanonicalNamespaceNames[$title->getNamespace()] . ':' . $title->getDBkey();
3138                 } else {
3139                         $articleName = $title->getDBkey();
3140                 }
3141
3142                 $url = str_replace('$1', urlencode($articleName), Title::getInterwikiLink($title->getInterwiki()));
3143                 $url .= "?action=$action";
3144                 if (strlen($url) > 255)
3145                         return wfMsg('scarytranscludetoolong');
3146                 return $this->fetchScaryTemplateMaybeFromCache($url);
3147         }
3148
3149         function fetchScaryTemplateMaybeFromCache($url) {
3150                 global $wgTranscludeCacheExpiry;
3151                 $dbr =& wfGetDB(DB_SLAVE);
3152                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3153                                 array('tc_url' => $url));
3154                 if ($obj) {
3155                         $time = $obj->tc_time;
3156                         $text = $obj->tc_contents;
3157                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
3158                                 return $text;
3159                         }
3160                 }
3161
3162                 $text = Http::get($url);
3163                 if (!$text)
3164                         return wfMsg('scarytranscludefailed', $url);
3165
3166                 $dbw =& wfGetDB(DB_MASTER);
3167                 $dbw->replace('transcache', array('tc_url'), array(
3168                         'tc_url' => $url,
3169                         'tc_time' => time(),
3170                         'tc_contents' => $text));
3171                 return $text;
3172         }
3173
3174
3175         /**
3176          * Triple brace replacement -- used for template arguments
3177          * @private
3178          */
3179         function argSubstitution( $matches ) {
3180                 $arg = trim( $matches['title'] );
3181                 $text = $matches['text'];
3182                 $inputArgs = end( $this->mArgStack );
3183
3184                 if ( array_key_exists( $arg, $inputArgs ) ) {
3185                         $text = $inputArgs[$arg];
3186                 } else if ($this->mOutputType == OT_HTML && null != $matches['parts'] && count($matches['parts']) > 0) {
3187                         $text = $matches['parts'][0];
3188                 }
3189
3190                 return $text;
3191         }
3192
3193         /**
3194          * Returns true if the function is allowed to include this entity
3195          * @private
3196          */
3197         function incrementIncludeCount( $dbk ) {
3198                 if ( !array_key_exists( $dbk, $this->mIncludeCount ) ) {
3199                         $this->mIncludeCount[$dbk] = 0;
3200                 }
3201                 if ( ++$this->mIncludeCount[$dbk] <= MAX_INCLUDE_REPEAT ) {
3202                         return true;
3203                 } else {
3204                         return false;
3205                 }
3206         }
3207
3208         /**
3209          * Detect __NOGALLERY__ magic word and set a placeholder
3210          */
3211         function stripNoGallery( &$text ) {
3212                 # if the string __NOGALLERY__ (not case-sensitive) occurs in the HTML,
3213                 # do not add TOC
3214                 $mw = MagicWord::get( MAG_NOGALLERY );
3215                 $this->mOutput->mNoGallery = $mw->matchAndRemove( $text ) ;
3216         }
3217
3218         /**
3219          * Detect __TOC__ magic word and set a placeholder
3220          */
3221         function stripToc( $text ) {
3222                 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
3223                 # do not add TOC
3224                 $mw = MagicWord::get( MAG_NOTOC );
3225                 if( $mw->matchAndRemove( $text ) ) {
3226                         $this->mShowToc = false;
3227                 }
3228
3229                 $mw = MagicWord::get( MAG_TOC );
3230                 if( $mw->match( $text ) ) {
3231                         $this->mShowToc = true;
3232                         $this->mForceTocPosition = true;
3233
3234                         // Set a placeholder. At the end we'll fill it in with the TOC.
3235                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3236
3237                         // Only keep the first one.
3238                         $text = $mw->replace( '', $text );
3239                 }
3240                 return $text;
3241         }
3242
3243         /**
3244          * This function accomplishes several tasks:
3245          * 1) Auto-number headings if that option is enabled
3246          * 2) Add an [edit] link to sections for logged in users who have enabled the option
3247          * 3) Add a Table of contents on the top for users who have enabled the option
3248          * 4) Auto-anchor headings
3249          *
3250          * It loops through all headlines, collects the necessary data, then splits up the
3251          * string and re-inserts the newly formatted headlines.
3252          *
3253          * @param string $text
3254          * @param boolean $isMain
3255          * @private
3256          */
3257         function formatHeadings( $text, $isMain=true ) {
3258                 global $wgMaxTocLevel, $wgContLang;
3259
3260                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3261                 if( !$this->mTitle->userCanEdit() ) {
3262                         $showEditLink = 0;
3263                 } else {
3264                         $showEditLink = $this->mOptions->getEditSection();
3265                 }
3266
3267                 # Inhibit editsection links if requested in the page
3268                 $esw =& MagicWord::get( MAG_NOEDITSECTION );
3269                 if( $esw->matchAndRemove( $text ) ) {
3270                         $showEditLink = 0;
3271                 }
3272
3273                 # Get all headlines for numbering them and adding funky stuff like [edit]
3274                 # links - this is for later, but we need the number of headlines right now
3275                 $numMatches = preg_match_all( '/<H([1-6])(.*?'.'>)(.*?)<\/H[1-6] *>/i', $text, $matches );
3276
3277                 # if there are fewer than 4 headlines in the article, do not show TOC
3278                 # unless it's been explicitly enabled.
3279                 $enoughToc = $this->mShowToc &&
3280                         (($numMatches >= 4) || $this->mForceTocPosition);
3281
3282                 # Allow user to stipulate that a page should have a "new section"
3283                 # link added via __NEWSECTIONLINK__
3284                 $mw =& MagicWord::get( MAG_NEWSECTIONLINK );
3285                 if( $mw->matchAndRemove( $text ) )
3286                         $this->mOutput->setNewSection( true );
3287
3288                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3289                 # override above conditions and always show TOC above first header
3290                 $mw =& MagicWord::get( MAG_FORCETOC );
3291                 if ($mw->matchAndRemove( $text ) ) {
3292                         $this->mShowToc = true;
3293                         $enoughToc = true;
3294                 }
3295
3296                 # Never ever show TOC if no headers
3297                 if( $numMatches < 1 ) {
3298                         $enoughToc = false;
3299                 }
3300
3301                 # We need this to perform operations on the HTML
3302                 $sk =& $this->mOptions->getSkin();
3303
3304                 # headline counter
3305                 $headlineCount = 0;
3306                 $sectionCount = 0; # headlineCount excluding template sections
3307
3308                 # Ugh .. the TOC should have neat indentation levels which can be
3309                 # passed to the skin functions. These are determined here
3310                 $toc = '';
3311                 $full = '';
3312                 $head = array();
3313                 $sublevelCount = array();
3314                 $levelCount = array();
3315                 $toclevel = 0;
3316                 $level = 0;
3317                 $prevlevel = 0;
3318                 $toclevel = 0;
3319                 $prevtoclevel = 0;
3320
3321                 foreach( $matches[3] as $headline ) {
3322                         $istemplate = 0;
3323                         $templatetitle = '';
3324                         $templatesection = 0;
3325                         $numbering = '';
3326
3327                         if (preg_match("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", $headline, $mat)) {
3328                                 $istemplate = 1;
3329                                 $templatetitle = base64_decode($mat[1]);
3330                                 $templatesection = 1 + (int)base64_decode($mat[2]);
3331                                 $headline = preg_replace("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", "", $headline);
3332                         }
3333
3334                         if( $toclevel ) {
3335                                 $prevlevel = $level;
3336                                 $prevtoclevel = $toclevel;
3337                         }
3338                         $level = $matches[1][$headlineCount];
3339
3340                         if( $doNumberHeadings || $enoughToc ) {
3341
3342                                 if ( $level > $prevlevel ) {
3343                                         # Increase TOC level
3344                                         $toclevel++;
3345                                         $sublevelCount[$toclevel] = 0;
3346                                         if( $toclevel<$wgMaxTocLevel ) {
3347                                                 $toc .= $sk->tocIndent();
3348                                         }
3349                                 }
3350                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3351                                         # Decrease TOC level, find level to jump to
3352
3353                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
3354                                                 # Can only go down to level 1
3355                                                 $toclevel = 1;
3356                                         } else {
3357                                                 for ($i = $toclevel; $i > 0; $i--) {
3358                                                         if ( $levelCount[$i] == $level ) {
3359                                                                 # Found last matching level
3360                                                                 $toclevel = $i;
3361                                                                 break;
3362                                                         }
3363                                                         elseif ( $levelCount[$i] < $level ) {
3364                                                                 # Found first matching level below current level
3365                                                                 $toclevel = $i + 1;
3366                                                                 break;
3367                                                         }
3368                                                 }
3369                                         }
3370                                         if( $toclevel<$wgMaxTocLevel ) {
3371                                                 $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3372                                         }
3373                                 }
3374                                 else {
3375                                         # No change in level, end TOC line
3376                                         if( $toclevel<$wgMaxTocLevel ) {
3377                                                 $toc .= $sk->tocLineEnd();
3378                                         }
3379                                 }
3380
3381                                 $levelCount[$toclevel] = $level;
3382
3383                                 # count number of headlines for each level
3384                                 @$sublevelCount[$toclevel]++;
3385                                 $dot = 0;
3386                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3387                                         if( !empty( $sublevelCount[$i] ) ) {
3388                                                 if( $dot ) {
3389                                                         $numbering .= '.';
3390                                                 }
3391                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3392                                                 $dot = 1;
3393                                         }
3394                                 }
3395                         }
3396
3397                         # The canonized header is a version of the header text safe to use for links
3398                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3399                         $canonized_headline = $this->unstrip( $headline, $this->mStripState );
3400                         $canonized_headline = $this->unstripNoWiki( $canonized_headline, $this->mStripState );
3401
3402                         # Remove link placeholders by the link text.
3403                         #     <!--LINK number-->
3404                         # turns into
3405                         #     link text with suffix
3406                         $canonized_headline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
3407                                                             "\$this->mLinkHolders['texts'][\$1]",
3408                                                             $canonized_headline );
3409                         $canonized_headline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
3410                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
3411                                                             $canonized_headline );
3412
3413                         # strip out HTML
3414                         $canonized_headline = preg_replace( '/<.*?' . '>/','',$canonized_headline );
3415                         $tocline = trim( $canonized_headline );
3416                         # Save headline for section edit hint before it's escaped
3417                         $headline_hint = trim( $canonized_headline );
3418                         $canonized_headline = Sanitizer::escapeId( $tocline );
3419                         $refers[$headlineCount] = $canonized_headline;
3420
3421                         # count how many in assoc. array so we can track dupes in anchors
3422                         @$refers[$canonized_headline]++;
3423                         $refcount[$headlineCount]=$refers[$canonized_headline];
3424
3425                         # Don't number the heading if it is the only one (looks silly)
3426                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3427                                 # the two are different if the line contains a link
3428                                 $headline=$numbering . ' ' . $headline;
3429                         }
3430
3431                         # Create the anchor for linking from the TOC to the section
3432                         $anchor = $canonized_headline;
3433                         if($refcount[$headlineCount] > 1 ) {
3434                                 $anchor .= '_' . $refcount[$headlineCount];
3435                         }
3436                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3437                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3438                         }
3439                         if( $showEditLink && ( !$istemplate || $templatetitle !== "" ) ) {
3440                                 if ( empty( $head[$headlineCount] ) ) {
3441                                         $head[$headlineCount] = '';
3442                                 }
3443                                 if( $istemplate )
3444                                         $head[$headlineCount] .= $sk->editSectionLinkForOther($templatetitle, $templatesection);
3445                                 else
3446                                         $head[$headlineCount] .= $sk->editSectionLink($this->mTitle, $sectionCount+1, $headline_hint);
3447                         }
3448
3449                         # give headline the correct <h#> tag
3450                         @$head[$headlineCount] .= "<a name=\"$anchor\"></a><h".$level.$matches[2][$headlineCount] .$headline.'</h'.$level.'>';
3451
3452                         $headlineCount++;
3453                         if( !$istemplate )
3454                                 $sectionCount++;
3455                 }
3456
3457                 if( $enoughToc ) {
3458                         if( $toclevel<$wgMaxTocLevel ) {
3459                                 $toc .= $sk->tocUnindent( $toclevel - 1 );
3460                         }
3461                         $toc = $sk->tocList( $toc );
3462                 }
3463
3464                 # split up and insert constructed headlines
3465
3466                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3467                 $i = 0;
3468
3469                 foreach( $blocks as $block ) {
3470                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
3471                                 # This is the [edit] link that appears for the top block of text when
3472                                 # section editing is enabled
3473
3474                                 # Disabled because it broke block formatting
3475                                 # For example, a bullet point in the top line
3476                                 # $full .= $sk->editSectionLink(0);
3477                         }
3478                         $full .= $block;
3479                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3480                                 # Top anchor now in skin
3481                                 $full = $full.$toc;
3482                         }
3483
3484                         if( !empty( $head[$i] ) ) {
3485                                 $full .= $head[$i];
3486                         }
3487                         $i++;
3488                 }
3489                 if( $this->mForceTocPosition ) {
3490                         return str_replace( '<!--MWTOC-->', $toc, $full );
3491                 } else {
3492                         return $full;
3493                 }
3494         }
3495
3496         /**
3497          * Return an HTML link for the "ISBN 123456" text
3498          * @private
3499          */
3500         function magicISBN( $text ) {
3501                 $fname = 'Parser::magicISBN';
3502                 wfProfileIn( $fname );
3503
3504                 $a = split( 'ISBN ', ' '.$text );
3505                 if ( count ( $a ) < 2 ) {
3506                         wfProfileOut( $fname );
3507                         return $text;
3508                 }
3509                 $text = substr( array_shift( $a ), 1);
3510                 $valid = '0123456789-Xx';
3511
3512                 foreach ( $a as $x ) {
3513                         # hack: don't replace inside thumbnail title/alt
3514                         # attributes
3515                         if(preg_match('/<[^>]+(alt|title)="[^">]*$/', $text)) {
3516                                 $text .= "ISBN $x";
3517                                 continue;
3518                         }
3519
3520                         $isbn = $blank = '' ;
3521                         while ( ' ' == $x{0} ) {
3522                                 $blank .= ' ';
3523                                 $x = substr( $x, 1 );
3524                         }
3525                         if ( $x == '' ) { # blank isbn
3526                                 $text .= "ISBN $blank";
3527                                 continue;
3528                         }
3529                         while ( strstr( $valid, $x{0} ) != false ) {
3530                                 $isbn .= $x{0};
3531                                 $x = substr( $x, 1 );
3532                         }
3533                         $num = str_replace( '-', '', $isbn );
3534                         $num = str_replace( ' ', '', $num );
3535                         $num = str_replace( 'x', 'X', $num );
3536
3537                         if ( '' == $num ) {
3538                                 $text .= "ISBN $blank$x";
3539                         } else {
3540                                 $titleObj = Title::makeTitle( NS_SPECIAL, 'Booksources' );
3541                                 $text .= '<a href="' .
3542                                         $titleObj->escapeLocalUrl( 'isbn='.$num ) .
3543                                         "\" class=\"internal\">ISBN $isbn</a>";
3544                                 $text .= $x;
3545                         }
3546                 }
3547                 wfProfileOut( $fname );
3548                 return $text;
3549         }
3550
3551         /**
3552          * Return an HTML link for the "RFC 1234" text
3553          *
3554          * @private
3555          * @param string $text     Text to be processed
3556          * @param string $keyword  Magic keyword to use (default RFC)
3557          * @param string $urlmsg   Interface message to use (default rfcurl)
3558          * @return string
3559          */
3560         function magicRFC( $text, $keyword='RFC ', $urlmsg='rfcurl'  ) {
3561
3562                 $valid = '0123456789';
3563                 $internal = false;
3564
3565                 $a = split( $keyword, ' '.$text );
3566                 if ( count ( $a ) < 2 ) {
3567                         return $text;
3568                 }
3569                 $text = substr( array_shift( $a ), 1);
3570
3571                 /* Check if keyword is preceed by [[.
3572                  * This test is made here cause of the array_shift above
3573                  * that prevent the test to be done in the foreach.
3574                  */
3575                 if ( substr( $text, -2 ) == '[[' ) {
3576                         $internal = true;
3577                 }
3578
3579                 foreach ( $a as $x ) {
3580                         /* token might be empty if we have RFC RFC 1234 */
3581                         if ( $x=='' ) {
3582                                 $text.=$keyword;
3583                                 continue;
3584                                 }
3585
3586                         # hack: don't replace inside thumbnail title/alt
3587                         # attributes
3588                         if(preg_match('/<[^>]+(alt|title)="[^">]*$/', $text)) {
3589                                 $text .= $keyword . $x;
3590                                 continue;
3591                         }
3592
3593                         $id = $blank = '' ;
3594
3595                         /** remove and save whitespaces in $blank */
3596                         while ( $x{0} == ' ' ) {
3597                                 $blank .= ' ';
3598                                 $x = substr( $x, 1 );
3599                         }
3600
3601                         /** remove and save the rfc number in $id */
3602                         while ( strstr( $valid, $x{0} ) != false ) {
3603                                 $id .= $x{0};
3604                                 $x = substr( $x, 1 );
3605                         }
3606
3607                         if ( $id == '' ) {
3608                                 /* call back stripped spaces*/
3609                                 $text .= $keyword.$blank.$x;
3610                         } elseif( $internal ) {
3611                                 /* normal link */
3612                                 $text .= $keyword.$id.$x;
3613                         } else {
3614                                 /* build the external link*/
3615                                 $url = wfMsg( $urlmsg, $id);
3616                                 $sk =& $this->mOptions->getSkin();
3617                                 $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
3618                                 $text .= "<a href=\"{$url}\"{$la}>{$keyword}{$id}</a>{$x}";
3619                         }
3620
3621                         /* Check if the next RFC keyword is preceed by [[ */
3622                         $internal = ( substr($x,-2) == '[[' );
3623                 }
3624                 return $text;
3625         }
3626
3627         /**
3628          * Transform wiki markup when saving a page by doing \r\n -> \n
3629          * conversion, substitting signatures, {{subst:}} templates, etc.
3630          *
3631          * @param string $text the text to transform
3632          * @param Title &$title the Title object for the current article
3633          * @param User &$user the User object describing the current user
3634          * @param ParserOptions $options parsing options
3635          * @param bool $clearState whether to clear the parser state first
3636          * @return string the altered wiki markup
3637          * @public
3638          */
3639         function preSaveTransform( $text, &$title, &$user, $options, $clearState = true ) {
3640                 $this->mOptions = $options;
3641                 $this->mTitle =& $title;
3642                 $this->mOutputType = OT_WIKI;
3643
3644                 if ( $clearState ) {
3645                         $this->clearState();
3646                 }
3647
3648                 $stripState = false;
3649                 $pairs = array(
3650                         "\r\n" => "\n",
3651                 );
3652                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3653                 $text = $this->strip( $text, $stripState, true, array( 'gallery' ) );
3654                 $text = $this->pstPass2( $text, $stripState, $user );
3655                 $text = $this->unstrip( $text, $stripState );
3656                 $text = $this->unstripNoWiki( $text, $stripState );
3657                 return $text;
3658         }
3659
3660         /**
3661          * Pre-save transform helper function
3662          * @private
3663          */
3664         function pstPass2( $text, &$stripState, &$user ) {
3665                 global $wgContLang, $wgLocaltimezone;
3666
3667                 /* Note: This is the timestamp saved as hardcoded wikitext to
3668                  * the database, we use $wgContLang here in order to give
3669                  * everyone the same signature and use the default one rather
3670                  * than the one selected in each user's preferences.
3671                  */
3672                 if ( isset( $wgLocaltimezone ) ) {
3673                         $oldtz = getenv( 'TZ' );
3674                         putenv( 'TZ='.$wgLocaltimezone );
3675                 }
3676                 $d = $wgContLang->timeanddate( date( 'YmdHis' ), false, false) .
3677                   ' (' . date( 'T' ) . ')';
3678                 if ( isset( $wgLocaltimezone ) ) {
3679                         putenv( 'TZ='.$oldtz );
3680                 }
3681
3682                 # Variable replacement
3683                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3684                 $text = $this->replaceVariables( $text );
3685
3686                 # Strip out <nowiki> etc. added via replaceVariables
3687                 $text = $this->strip( $text, $stripState, false, array( 'gallery' ) );
3688
3689                 # Signatures
3690                 $sigText = $this->getUserSig( $user );
3691                 $text = strtr( $text, array(
3692                         '~~~~~' => $d,
3693                         '~~~~' => "$sigText $d",
3694                         '~~~' => $sigText
3695                 ) );
3696
3697                 # Context links: [[|name]] and [[name (context)|]]
3698                 #
3699                 global $wgLegalTitleChars;
3700                 $tc = "[$wgLegalTitleChars]";
3701                 $np = str_replace( array( '(', ')' ), array( '', '' ), $tc ); # No parens
3702
3703                 $namespacechar = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
3704                 $conpat = "/^({$np}+) \\(({$tc}+)\\)$/";
3705
3706                 $p1 = "/\[\[({$np}+) \\(({$np}+)\\)\\|]]/";             # [[page (context)|]]
3707                 $p2 = "/\[\[\\|({$tc}+)]]/";                                    # [[|page]]
3708                 $p3 = "/\[\[(:*$namespacechar+):({$np}+)\\|]]/";                # [[namespace:page|]] and [[:namespace:page|]]
3709                 $p4 = "/\[\[(:*$namespacechar+):({$np}+) \\(({$np}+)\\)\\|]]/"; # [[ns:page (cont)|]] and [[:ns:page (cont)|]]
3710                 $context = '';
3711                 $t = $this->mTitle->getText();
3712                 if ( preg_match( $conpat, $t, $m ) ) {
3713                         $context = $m[2];
3714                 }
3715                 $text = preg_replace( $p4, '[[\\1:\\2 (\\3)|\\2]]', $text );
3716                 $text = preg_replace( $p1, '[[\\1 (\\2)|\\1]]', $text );
3717                 $text = preg_replace( $p3, '[[\\1:\\2|\\2]]', $text );
3718
3719                 if ( '' == $context ) {
3720                         $text = preg_replace( $p2, '[[\\1]]', $text );
3721                 } else {
3722                         $text = preg_replace( $p2, "[[\\1 ({$context})|\\1]]", $text );
3723                 }
3724
3725                 # Trim trailing whitespace
3726                 # MAG_END (__END__) tag allows for trailing
3727                 # whitespace to be deliberately included
3728                 $text = rtrim( $text );
3729                 $mw =& MagicWord::get( MAG_END );
3730                 $mw->matchAndRemove( $text );
3731
3732                 return $text;
3733         }
3734
3735         /**
3736          * Fetch the user's signature text, if any, and normalize to
3737          * validated, ready-to-insert wikitext.
3738          *
3739          * @param User $user
3740          * @return string
3741          * @private
3742          */
3743         function getUserSig( &$user ) {
3744                 $username = $user->getName();
3745                 $nickname = $user->getOption( 'nickname' );
3746                 $nickname = $nickname === '' ? $username : $nickname;
3747
3748                 if( $user->getBoolOption( 'fancysig' ) !== false ) {
3749                         # Sig. might contain markup; validate this
3750                         if( $this->validateSig( $nickname ) !== false ) {
3751                                 # Validated; clean up (if needed) and return it
3752                                 return $this->cleanSig( $nickname, true );
3753                         } else {
3754                                 # Failed to validate; fall back to the default
3755                                 $nickname = $username;
3756                                 wfDebug( "Parser::getUserSig: $username has bad XML tags in signature.\n" );
3757                         }
3758                 }
3759
3760                 // Make sure nickname doesnt get a sig in a sig
3761                 $nickname = $this->cleanSigInSig( $nickname );
3762
3763                 # If we're still here, make it a link to the user page
3764                 $userpage = $user->getUserPage();
3765                 return( '[[' . $userpage->getPrefixedText() . '|' . wfEscapeWikiText( $nickname ) . ']]' );
3766         }
3767
3768         /**
3769          * Check that the user's signature contains no bad XML
3770          *
3771          * @param string $text
3772          * @return mixed An expanded string, or false if invalid.
3773          */
3774         function validateSig( $text ) {
3775                 return( wfIsWellFormedXmlFragment( $text ) ? $text : false );
3776         }
3777
3778         /**
3779          * Clean up signature text
3780          *
3781          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
3782          * 2) Substitute all transclusions
3783          *
3784          * @param string $text
3785          * @param $parsing Whether we're cleaning (preferences save) or parsing
3786          * @return string Signature text
3787          */
3788         function cleanSig( $text, $parsing = false ) {
3789                 global $wgTitle;
3790                 $this->startExternalParse( $wgTitle, new ParserOptions(), $parsing ? OT_WIKI : OT_MSG );
3791
3792                 $substWord = MagicWord::get( MAG_SUBST );
3793                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3794                 $substText = '{{' . $substWord->getSynonym( 0 );
3795
3796                 $text = preg_replace( $substRegex, $substText, $text );
3797                 $text = $this->cleanSigInSig( $text );
3798                 $text = $this->replaceVariables( $text );
3799
3800                 $this->clearState();
3801                 return $text;
3802         }
3803
3804         /**
3805          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
3806          * @param string $text
3807          * @return string Signature text with /~{3,5}/ removed
3808          */
3809         function cleanSigInSig( $text ) {
3810                 $text = preg_replace( '/~{3,5}/', '', $text );
3811                 return $text;
3812         }
3813
3814         /**
3815          * Set up some variables which are usually set up in parse()
3816          * so that an external function can call some class members with confidence
3817          * @public
3818          */
3819         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
3820                 $this->mTitle =& $title;
3821                 $this->mOptions = $options;
3822                 $this->mOutputType = $outputType;
3823                 if ( $clearState ) {
3824                         $this->clearState();
3825                 }
3826         }
3827
3828         /**
3829          * Transform a MediaWiki message by replacing magic variables.
3830          *
3831          * @param string $text the text to transform
3832          * @param ParserOptions $options  options
3833          * @return string the text with variables substituted
3834          * @public
3835          */
3836         function transformMsg( $text, $options ) {
3837                 global $wgTitle;
3838                 static $executing = false;
3839
3840                 $fname = "Parser::transformMsg";
3841
3842                 # Guard against infinite recursion
3843                 if ( $executing ) {
3844                         return $text;
3845                 }
3846                 $executing = true;
3847
3848                 wfProfileIn($fname);
3849
3850                 $this->mTitle = $wgTitle;
3851                 $this->mOptions = $options;
3852                 $this->mOutputType = OT_MSG;
3853                 $this->clearState();
3854                 $text = $this->replaceVariables( $text );
3855
3856                 $executing = false;
3857                 wfProfileOut($fname);
3858                 return $text;
3859         }
3860
3861         /**
3862          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
3863          * The callback should have the following form:
3864          *    function myParserHook( $text, $params, &$parser ) { ... }
3865          *
3866          * Transform and return $text. Use $parser for any required context, e.g. use
3867          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
3868          *
3869          * @public
3870          *
3871          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
3872          * @param mixed $callback The callback function (and object) to use for the tag
3873          *
3874          * @return The old value of the mTagHooks array associated with the hook
3875          */
3876         function setHook( $tag, $callback ) {
3877                 $tag = strtolower( $tag );
3878                 $oldVal = @$this->mTagHooks[$tag];
3879                 $this->mTagHooks[$tag] = $callback;
3880
3881                 return $oldVal;
3882         }
3883
3884         /**
3885          * Create a function, e.g. {{sum:1|2|3}}
3886          * The callback function should have the form:
3887          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
3888          *
3889          * The callback may either return the text result of the function, or an array with the text
3890          * in element 0, and a number of flags in the other elements. The names of the flags are
3891          * specified in the keys. Valid flags are:
3892          *   found                     The text returned is valid, stop processing the template. This
3893          *                             is on by default.
3894          *   nowiki                    Wiki markup in the return value should be escaped
3895          *   noparse                   Unsafe HTML tags should not be stripped, etc.
3896          *   noargs                    Don't replace triple-brace arguments in the return value
3897          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
3898          *
3899          * @public
3900          *
3901          * @param mixed $id The magic word ID, or (deprecated) the function name. Function names are case-insensitive.
3902          * @param mixed $callback The callback function (and object) to use
3903          *
3904          * @return The old callback function for this name, if any
3905          */
3906         function setFunctionHook( $id, $callback ) {
3907                 if( is_string( $id ) ) {
3908                         $id = strtolower( $id );
3909                 }
3910                 $oldVal = @$this->mFunctionHooks[$id];
3911                 $this->mFunctionHooks[$id] = $callback;
3912                 # Invalidate function synonym cache
3913                 $this->mFunctionSynonymsValid = false;
3914                 return $oldVal;
3915         }
3916
3917         /**
3918          * Make sure the function synonym cache is up to date
3919          */
3920         function refreshFunctionCache() {
3921                 if ( !$this->mFunctionSynonymsValid ) {
3922                         $this->mFunctionSynonyms = array();
3923                         foreach( $this->mFunctionHooks as $key => $value ) {
3924                                 if ( is_int( $key ) ) {
3925                                         $mw = MagicWord::get( $key );
3926                                         $synonyms = $mw->getSynonyms();
3927                                         foreach( $synonyms as $synonym ) {
3928                                                 $this->mFunctionSynonyms[strtolower($synonym)] = $key;
3929                                         }
3930                                 }
3931                         }
3932                         $this->mFunctionSynonymsValid = true;
3933                 }
3934         }
3935
3936         /**
3937          * Replace <!--LINK--> link placeholders with actual links, in the buffer
3938          * Placeholders created in Skin::makeLinkObj()
3939          * Returns an array of links found, indexed by PDBK:
3940          *  0 - broken
3941          *  1 - normal link
3942          *  2 - stub
3943          * $options is a bit field, RLH_FOR_UPDATE to select for update
3944          */
3945         function replaceLinkHolders( &$text, $options = 0 ) {
3946                 global $wgUser;
3947                 global $wgOutputReplace;
3948
3949                 $fname = 'Parser::replaceLinkHolders';
3950                 wfProfileIn( $fname );
3951
3952                 $pdbks = array();
3953                 $colours = array();
3954                 $sk =& $this->mOptions->getSkin();
3955                 $linkCache =& LinkCache::singleton();
3956
3957                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
3958                         wfProfileIn( $fname.'-check' );
3959                         $dbr =& wfGetDB( DB_SLAVE );
3960                         $page = $dbr->tableName( 'page' );
3961                         $threshold = $wgUser->getOption('stubthreshold');
3962
3963                         # Sort by namespace
3964                         asort( $this->mLinkHolders['namespaces'] );
3965
3966                         # Generate query
3967                         $query = false;
3968                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3969                                 # Make title object
3970                                 $title = $this->mLinkHolders['titles'][$key];
3971
3972                                 # Skip invalid entries.
3973                                 # Result will be ugly, but prevents crash.
3974                                 if ( is_null( $title ) ) {
3975                                         continue;
3976                                 }
3977                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
3978
3979                                 # Check if it's a static known link, e.g. interwiki
3980                                 if ( $title->isAlwaysKnown() ) {
3981                                         $colours[$pdbk] = 1;
3982                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
3983                                         $colours[$pdbk] = 1;
3984                                         $this->mOutput->addLink( $title, $id );
3985                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
3986                                         $colours[$pdbk] = 0;
3987                                 } else {
3988                                         # Not in the link cache, add it to the query
3989                                         if ( !isset( $current ) ) {
3990                                                 $current = $ns;
3991                                                 $query =  "SELECT page_id, page_namespace, page_title";
3992                                                 if ( $threshold > 0 ) {
3993                                                         $query .= ', page_len, page_is_redirect';
3994                                                 }
3995                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
3996                                         } elseif ( $current != $ns ) {
3997                                                 $current = $ns;
3998                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
3999                                         } else {
4000                                                 $query .= ', ';
4001                                         }
4002
4003                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
4004                                 }
4005                         }
4006                         if ( $query ) {
4007                                 $query .= '))';
4008                                 if ( $options & RLH_FOR_UPDATE ) {
4009                                         $query .= ' FOR UPDATE';
4010                                 }
4011
4012                                 $res = $dbr->query( $query, $fname );
4013
4014                                 # Fetch data and form into an associative array
4015                                 # non-existent = broken
4016                                 # 1 = known
4017                                 # 2 = stub
4018                                 while ( $s = $dbr->fetchObject($res) ) {
4019                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
4020                                         $pdbk = $title->getPrefixedDBkey();
4021                                         $linkCache->addGoodLinkObj( $s->page_id, $title );
4022                                         $this->mOutput->addLink( $title, $s->page_id );
4023
4024                                         if ( $threshold >  0 ) {
4025                                                 $size = $s->page_len;
4026                                                 if ( $s->page_is_redirect || $s->page_namespace != 0 || $size >= $threshold ) {
4027                                                         $colours[$pdbk] = 1;
4028                                                 } else {
4029                                                         $colours[$pdbk] = 2;
4030                                                 }
4031                                         } else {
4032                                                 $colours[$pdbk] = 1;
4033                                         }
4034                                 }
4035                         }
4036                         wfProfileOut( $fname.'-check' );
4037
4038                         # Construct search and replace arrays
4039                         wfProfileIn( $fname.'-construct' );
4040                         $wgOutputReplace = array();
4041                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4042                                 $pdbk = $pdbks[$key];
4043                                 $searchkey = "<!--LINK $key-->";
4044                                 $title = $this->mLinkHolders['titles'][$key];
4045                                 if ( empty( $colours[$pdbk] ) ) {
4046                                         $linkCache->addBadLinkObj( $title );
4047                                         $colours[$pdbk] = 0;
4048                                         $this->mOutput->addLink( $title, 0 );
4049                                         $wgOutputReplace[$searchkey] = $sk->makeBrokenLinkObj( $title,
4050                                                                         $this->mLinkHolders['texts'][$key],
4051                                                                         $this->mLinkHolders['queries'][$key] );
4052                                 } elseif ( $colours[$pdbk] == 1 ) {
4053                                         $wgOutputReplace[$searchkey] = $sk->makeKnownLinkObj( $title,
4054                                                                         $this->mLinkHolders['texts'][$key],
4055                                                                         $this->mLinkHolders['queries'][$key] );
4056                                 } elseif ( $colours[$pdbk] == 2 ) {
4057                                         $wgOutputReplace[$searchkey] = $sk->makeStubLinkObj( $title,
4058                                                                         $this->mLinkHolders['texts'][$key],
4059                                                                         $this->mLinkHolders['queries'][$key] );
4060                                 }
4061                         }
4062                         wfProfileOut( $fname.'-construct' );
4063
4064                         # Do the thing
4065                         wfProfileIn( $fname.'-replace' );
4066
4067                         $text = preg_replace_callback(
4068                                 '/(<!--LINK .*?-->)/',
4069                                 "wfOutputReplaceMatches",
4070                                 $text);
4071
4072                         wfProfileOut( $fname.'-replace' );
4073                 }
4074
4075                 # Now process interwiki link holders
4076                 # This is quite a bit simpler than internal links
4077                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
4078                         wfProfileIn( $fname.'-interwiki' );
4079                         # Make interwiki link HTML
4080                         $wgOutputReplace = array();
4081                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
4082                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
4083                                 $wgOutputReplace[$key] = $sk->makeLinkObj( $title, $link );
4084                         }
4085
4086                         $text = preg_replace_callback(
4087                                 '/<!--IWLINK (.*?)-->/',
4088                                 "wfOutputReplaceMatches",
4089                                 $text );
4090                         wfProfileOut( $fname.'-interwiki' );
4091                 }
4092
4093                 wfProfileOut( $fname );
4094                 return $colours;
4095         }
4096
4097         /**
4098          * Replace <!--LINK--> link placeholders with plain text of links
4099          * (not HTML-formatted).
4100          * @param string $text
4101          * @return string
4102          */
4103         function replaceLinkHoldersText( $text ) {
4104                 $fname = 'Parser::replaceLinkHoldersText';
4105                 wfProfileIn( $fname );
4106
4107                 $text = preg_replace_callback(
4108                         '/<!--(LINK|IWLINK) (.*?)-->/',
4109                         array( &$this, 'replaceLinkHoldersTextCallback' ),
4110                         $text );
4111
4112                 wfProfileOut( $fname );
4113                 return $text;
4114         }
4115
4116         /**
4117          * @param array $matches
4118          * @return string
4119          * @private
4120          */
4121         function replaceLinkHoldersTextCallback( $matches ) {
4122                 $type = $matches[1];
4123                 $key  = $matches[2];
4124                 if( $type == 'LINK' ) {
4125                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
4126                                 return $this->mLinkHolders['texts'][$key];
4127                         }
4128                 } elseif( $type == 'IWLINK' ) {
4129                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
4130                                 return $this->mInterwikiLinkHolders['texts'][$key];
4131                         }
4132                 }
4133                 return $matches[0];
4134         }
4135
4136         /**
4137          * Tag hook handler for 'pre'.
4138          */
4139         function renderPreTag( $text, $attribs, $parser ) {
4140                 // Backwards-compatibility hack
4141                 $content = preg_replace( '!<nowiki>(.*?)</nowiki>!is', '\\1', $text );
4142
4143                 $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
4144                 return wfOpenElement( 'pre', $attribs ) .
4145                         wfEscapeHTMLTagsOnly( $content ) .
4146                         '</pre>';
4147         }
4148
4149         /**
4150          * Renders an image gallery from a text with one line per image.
4151          * text labels may be given by using |-style alternative text. E.g.
4152          *   Image:one.jpg|The number "1"
4153          *   Image:tree.jpg|A tree
4154          * given as text will return the HTML of a gallery with two images,
4155          * labeled 'The number "1"' and
4156          * 'A tree'.
4157          */
4158         function renderImageGallery( $text, $params ) {
4159                 $ig = new ImageGallery();
4160                 $ig->setShowBytes( false );
4161                 $ig->setShowFilename( false );
4162                 $ig->setParsing();
4163                 $ig->useSkin( $this->mOptions->getSkin() );
4164
4165                 if( isset( $params['caption'] ) )
4166                         $ig->setCaption( $params['caption'] );
4167
4168                 $lines = explode( "\n", $text );
4169                 foreach ( $lines as $line ) {
4170                         # match lines like these:
4171                         # Image:someimage.jpg|This is some image
4172                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4173                         # Skip empty lines
4174                         if ( count( $matches ) == 0 ) {
4175                                 continue;
4176                         }
4177                         $nt =& Title::newFromText( $matches[1] );
4178                         if( is_null( $nt ) ) {
4179                                 # Bogus title. Ignore these so we don't bomb out later.
4180                                 continue;
4181                         }
4182                         if ( isset( $matches[3] ) ) {
4183                                 $label = $matches[3];
4184                         } else {
4185                                 $label = '';
4186                         }
4187
4188                         $pout = $this->parse( $label,
4189                                 $this->mTitle,
4190                                 $this->mOptions,
4191                                 false, // Strip whitespace...?
4192                                 false  // Don't clear state!
4193                         );
4194                         $html = $pout->getText();
4195
4196                         $ig->add( new Image( $nt ), $html );
4197
4198                         # Only add real images (bug #5586)
4199                         if ( $nt->getNamespace() == NS_IMAGE ) {
4200                                 $this->mOutput->addImage( $nt->getDBkey() );
4201                         }
4202                 }
4203                 return $ig->toHTML();
4204         }
4205
4206         /**
4207          * Parse image options text and use it to make an image
4208          */
4209         function makeImage( &$nt, $options ) {
4210                 global $wgUseImageResize;
4211
4212                 $align = '';
4213
4214                 # Check if the options text is of the form "options|alt text"
4215                 # Options are:
4216                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
4217                 #  * left               no resizing, just left align. label is used for alt= only
4218                 #  * right              same, but right aligned
4219                 #  * none               same, but not aligned
4220                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
4221                 #  * center             center the image
4222                 #  * framed             Keep original image size, no magnify-button.
4223
4224                 $part = explode( '|', $options);
4225
4226                 $mwThumb  =& MagicWord::get( MAG_IMG_THUMBNAIL );
4227                 $mwManualThumb =& MagicWord::get( MAG_IMG_MANUALTHUMB );
4228                 $mwLeft   =& MagicWord::get( MAG_IMG_LEFT );
4229                 $mwRight  =& MagicWord::get( MAG_IMG_RIGHT );
4230                 $mwNone   =& MagicWord::get( MAG_IMG_NONE );
4231                 $mwWidth  =& MagicWord::get( MAG_IMG_WIDTH );
4232                 $mwCenter =& MagicWord::get( MAG_IMG_CENTER );
4233                 $mwFramed =& MagicWord::get( MAG_IMG_FRAMED );
4234                 $caption = '';
4235
4236                 $width = $height = $framed = $thumb = false;
4237                 $manual_thumb = '' ;
4238
4239                 foreach( $part as $key => $val ) {
4240                         if ( $wgUseImageResize && ! is_null( $mwThumb->matchVariableStartToEnd($val) ) ) {
4241                                 $thumb=true;
4242                         } elseif ( ! is_null( $match = $mwManualThumb->matchVariableStartToEnd($val) ) ) {
4243                                 # use manually specified thumbnail
4244                                 $thumb=true;
4245                                 $manual_thumb = $match;
4246                         } elseif ( ! is_null( $mwRight->matchVariableStartToEnd($val) ) ) {
4247                                 # remember to set an alignment, don't render immediately
4248                                 $align = 'right';
4249                         } elseif ( ! is_null( $mwLeft->matchVariableStartToEnd($val) ) ) {
4250                                 # remember to set an alignment, don't render immediately
4251                                 $align = 'left';
4252                         } elseif ( ! is_null( $mwCenter->matchVariableStartToEnd($val) ) ) {
4253                                 # remember to set an alignment, don't render immediately
4254                                 $align = 'center';
4255                         } elseif ( ! is_null( $mwNone->matchVariableStartToEnd($val) ) ) {
4256                                 # remember to set an alignment, don't render immediately
4257                                 $align = 'none';
4258                         } elseif ( $wgUseImageResize && ! is_null( $match = $mwWidth->matchVariableStartToEnd($val) ) ) {
4259                                 wfDebug( "MAG_IMG_WIDTH match: $match\n" );
4260                                 # $match is the image width in pixels
4261                                 if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $match, $m ) ) {
4262                                         $width = intval( $m[1] );
4263                                         $height = intval( $m[2] );
4264                                 } else {
4265                                         $width = intval($match);
4266                                 }
4267                         } elseif ( ! is_null( $mwFramed->matchVariableStartToEnd($val) ) ) {
4268                                 $framed=true;
4269                         } else {
4270                                 $caption = $val;
4271                         }
4272                 }
4273                 # Strip bad stuff out of the alt text
4274                 $alt = $this->replaceLinkHoldersText( $caption );
4275
4276                 # make sure there are no placeholders in thumbnail attributes
4277                 # that are later expanded to html- so expand them now and
4278                 # remove the tags
4279                 $alt = $this->unstrip($alt, $this->mStripState);
4280                 $alt = Sanitizer::stripAllTags( $alt );
4281
4282                 # Linker does the rest
4283                 $sk =& $this->mOptions->getSkin();
4284                 return $sk->makeImageLinkObj( $nt, $caption, $alt, $align, $width, $height, $framed, $thumb, $manual_thumb );
4285         }
4286
4287         /**
4288          * Set a flag in the output object indicating that the content is dynamic and
4289          * shouldn't be cached.
4290          */
4291         function disableCache() {
4292                 wfDebug( "Parser output marked as uncacheable.\n" );
4293                 $this->mOutput->mCacheTime = -1;
4294         }
4295
4296         /**#@+
4297          * Callback from the Sanitizer for expanding items found in HTML attribute
4298          * values, so they can be safely tested and escaped.
4299          * @param string $text
4300          * @param array $args
4301          * @return string
4302          * @private
4303          */
4304         function attributeStripCallback( &$text, $args ) {
4305                 $text = $this->replaceVariables( $text, $args );
4306                 $text = $this->unstripForHTML( $text );
4307                 return $text;
4308         }
4309
4310         function unstripForHTML( $text ) {
4311                 $text = $this->unstrip( $text, $this->mStripState );
4312                 $text = $this->unstripNoWiki( $text, $this->mStripState );
4313                 return $text;
4314         }
4315         /**#@-*/
4316
4317         /**#@+
4318          * Accessor/mutator
4319          */
4320         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4321         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4322         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4323         /**#@-*/
4324
4325         /**#@+
4326          * Accessor
4327          */
4328         function getTags() { return array_keys( $this->mTagHooks ); }
4329         /**#@-*/
4330
4331
4332         /**
4333          * Break wikitext input into sections, and either pull or replace
4334          * some particular section's text.
4335          *
4336          * External callers should use the getSection and replaceSection methods.
4337          *
4338          * @param $text Page wikitext
4339          * @param $section Numbered section. 0 pulls the text before the first
4340          *                 heading; other numbers will pull the given section
4341          *                 along with its lower-level subsections.
4342          * @param $mode One of "get" or "replace"
4343          * @param $newtext Replacement text for section data.
4344          * @return string for "get", the extracted section text.
4345          *                for "replace", the whole page with the section replaced.
4346          */
4347         private function extractSections( $text, $section, $mode, $newtext='' ) {
4348                 # strip NOWIKI etc. to avoid confusion (true-parameter causes HTML
4349                 # comments to be stripped as well)
4350                 $striparray = array();
4351
4352                 $oldOutputType = $this->mOutputType;
4353                 $oldOptions = $this->mOptions;
4354                 $this->mOptions = new ParserOptions();
4355                 $this->mOutputType = OT_WIKI;
4356
4357                 $striptext = $this->strip( $text, $striparray, true );
4358
4359                 $this->mOutputType = $oldOutputType;
4360                 $this->mOptions = $oldOptions;
4361
4362                 # now that we can be sure that no pseudo-sections are in the source,
4363                 # split it up by section
4364                 $uniq = preg_quote( $this->uniqPrefix(), '/' );
4365                 $comment = "(?:$uniq-!--.*?QINU)";
4366                 $secs = preg_split(
4367                 /*
4368                         "/
4369                         ^(
4370                         (?:$comment|<\/?noinclude>)* # Initial comments will be stripped
4371                         (?:
4372                                 (=+) # Should this be limited to 6?
4373                                 .+?  # Section title...
4374                                 \\2  # Ending = count must match start
4375                         |
4376                                 ^
4377                                 <h([1-6])\b.*?>
4378                                 .*?
4379                                 <\/h\\3\s*>
4380                         )
4381                         (?:$comment|<\/?noinclude>|\s+)* # Trailing whitespace ok
4382                         )$
4383                         /mix",
4384                 */
4385                         "/
4386                         (
4387                                 ^
4388                                 (?:$comment|<\/?noinclude>)* # Initial comments will be stripped
4389                                 (=+) # Should this be limited to 6?
4390                                 .+?  # Section title...
4391                                 \\2  # Ending = count must match start
4392                                 (?:$comment|<\/?noinclude>|[ \\t]+)* # Trailing whitespace ok
4393                                 $
4394                         |
4395                                 <h([1-6])\b.*?>
4396                                 .*?
4397                                 <\/h\\3\s*>
4398                         )
4399                         /mix",
4400                         $striptext, -1,
4401                         PREG_SPLIT_DELIM_CAPTURE);
4402
4403                 if( $mode == "get" ) {
4404                         if( $section == 0 ) {
4405                                 // "Section 0" returns the content before any other section.
4406                                 $rv = $secs[0];
4407                         } else {
4408                                 $rv = "";
4409                         }
4410                 } elseif( $mode == "replace" ) {
4411                         if( $section == 0 ) {
4412                                 $rv = $newtext . "\n\n";
4413                                 $remainder = true;
4414                         } else {
4415                                 $rv = $secs[0];
4416                                 $remainder = false;
4417                         }
4418                 }
4419                 $count = 0;
4420                 $sectionLevel = 0;
4421                 for( $index = 1; $index < count( $secs ); ) {
4422                         $headerLine = $secs[$index++];
4423                         if( $secs[$index] ) {
4424                                 // A wiki header
4425                                 $headerLevel = strlen( $secs[$index++] );
4426                         } else {
4427                                 // An HTML header
4428                                 $index++;
4429                                 $headerLevel = intval( $secs[$index++] );
4430                         }
4431                         $content = $secs[$index++];
4432
4433                         $count++;
4434                         if( $mode == "get" ) {
4435                                 if( $count == $section ) {
4436                                         $rv = $headerLine . $content;
4437                                         $sectionLevel = $headerLevel;
4438                                 } elseif( $count > $section ) {
4439                                         if( $sectionLevel && $headerLevel > $sectionLevel ) {
4440                                                 $rv .= $headerLine . $content;
4441                                         } else {
4442                                                 // Broke out to a higher-level section
4443                                                 break;
4444                                         }
4445                                 }
4446                         } elseif( $mode == "replace" ) {
4447                                 if( $count < $section ) {
4448                                         $rv .= $headerLine . $content;
4449                                 } elseif( $count == $section ) {
4450                                         $rv .= $newtext . "\n\n";
4451                                         $sectionLevel = $headerLevel;
4452                                 } elseif( $count > $section ) {
4453                                         if( $headerLevel <= $sectionLevel ) {
4454                                                 // Passed the section's sub-parts.
4455                                                 $remainder = true;
4456                                         }
4457                                         if( $remainder ) {
4458                                                 $rv .= $headerLine . $content;
4459                                         }
4460                                 }
4461                         }
4462                 }
4463                 # reinsert stripped tags
4464                 $rv = $this->unstrip( $rv, $striparray );
4465                 $rv = $this->unstripNoWiki( $rv, $striparray );
4466                 $rv = trim( $rv );
4467                 return $rv;
4468         }
4469
4470         /**
4471          * This function returns the text of a section, specified by a number ($section).
4472          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4473          * the first section before any such heading (section 0).
4474          *
4475          * If a section contains subsections, these are also returned.
4476          *
4477          * @param $text String: text to look in
4478          * @param $section Integer: section number
4479          * @return string text of the requested section
4480          */
4481         function getSection( $text, $section ) {
4482                 return $this->extractSections( $text, $section, "get" );
4483         }
4484
4485         function replaceSection( $oldtext, $section, $text ) {
4486                 return $this->extractSections( $oldtext, $section, "replace", $text );
4487         }
4488
4489 }
4490
4491 /**
4492  * @todo document
4493  * @package MediaWiki
4494  */
4495 class ParserOutput
4496 {
4497         var $mText,             # The output text
4498                 $mLanguageLinks,    # List of the full text of language links, in the order they appear
4499                 $mCategories,       # Map of category names to sort keys
4500                 $mContainsOldMagic, # Boolean variable indicating if the input contained variables like {{CURRENTDAY}}
4501                 $mCacheTime,        # Time when this object was generated, or -1 for uncacheable. Used in ParserCache.
4502                 $mVersion,          # Compatibility check
4503                 $mTitleText,        # title text of the chosen language variant
4504                 $mLinks,            # 2-D map of NS/DBK to ID for the links in the document. ID=zero for broken.
4505                 $mTemplates,        # 2-D map of NS/DBK to ID for the template references. ID=zero for broken.
4506                 $mImages,           # DB keys of the images used, in the array key only
4507                 $mExternalLinks,    # External link URLs, in the key only
4508                 $mHTMLtitle,            # Display HTML title
4509                 $mSubtitle,                     # Additional subtitle
4510                 $mNewSection,           # Show a new section link?
4511                 $mNoGallery;            # No gallery on category page? (__NOGALLERY__)
4512
4513         function ParserOutput( $text = '', $languageLinks = array(), $categoryLinks = array(),
4514                 $containsOldMagic = false, $titletext = '' )
4515         {
4516                 $this->mText = $text;
4517                 $this->mLanguageLinks = $languageLinks;
4518                 $this->mCategories = $categoryLinks;
4519                 $this->mContainsOldMagic = $containsOldMagic;
4520                 $this->mCacheTime = '';
4521                 $this->mVersion = MW_PARSER_VERSION;
4522                 $this->mTitleText = $titletext;
4523                 $this->mLinks = array();
4524                 $this->mTemplates = array();
4525                 $this->mImages = array();
4526                 $this->mExternalLinks = array();
4527                 $this->mHTMLtitle = "" ;
4528                 $this->mSubtitle = "" ;
4529                 $this->mNewSection = false;
4530                 $this->mNoGallery = false;
4531         }
4532
4533         function getText()                   { return $this->mText; }
4534         function &getLanguageLinks()          { return $this->mLanguageLinks; }
4535         function getCategoryLinks()          { return array_keys( $this->mCategories ); }
4536         function &getCategories()            { return $this->mCategories; }
4537         function getCacheTime()              { return $this->mCacheTime; }
4538         function getTitleText()              { return $this->mTitleText; }
4539         function &getLinks()                 { return $this->mLinks; }
4540         function &getTemplates()             { return $this->mTemplates; }
4541         function &getImages()                { return $this->mImages; }
4542         function &getExternalLinks()         { return $this->mExternalLinks; }
4543         function getNoGallery()              { return $this->mNoGallery; }
4544
4545         function containsOldMagic()          { return $this->mContainsOldMagic; }
4546         function setText( $text )            { return wfSetVar( $this->mText, $text ); }
4547         function setLanguageLinks( $ll )     { return wfSetVar( $this->mLanguageLinks, $ll ); }
4548         function setCategoryLinks( $cl )     { return wfSetVar( $this->mCategories, $cl ); }
4549         function setContainsOldMagic( $com ) { return wfSetVar( $this->mContainsOldMagic, $com ); }
4550         function setCacheTime( $t )          { return wfSetVar( $this->mCacheTime, $t ); }
4551         function setTitleText( $t )          { return wfSetVar ($this->mTitleText, $t); }
4552
4553         function addCategory( $c, $sort )    { $this->mCategories[$c] = $sort; }
4554         function addImage( $name )           { $this->mImages[$name] = 1; }
4555         function addLanguageLink( $t )       { $this->mLanguageLinks[] = $t; }
4556         function addExternalLink( $url )     { $this->mExternalLinks[$url] = 1; }
4557
4558         function setNewSection( $value ) {
4559                 $this->mNewSection = (bool)$value;
4560         }
4561         function getNewSection() {
4562                 return (bool)$this->mNewSection;
4563         }
4564
4565         function addLink( $title, $id ) {
4566                 $ns = $title->getNamespace();
4567                 $dbk = $title->getDBkey();
4568                 if ( !isset( $this->mLinks[$ns] ) ) {
4569                         $this->mLinks[$ns] = array();
4570                 }
4571                 $this->mLinks[$ns][$dbk] = $id;
4572         }
4573
4574         function addTemplate( $title, $id ) {
4575                 $ns = $title->getNamespace();
4576                 $dbk = $title->getDBkey();
4577                 if ( !isset( $this->mTemplates[$ns] ) ) {
4578                         $this->mTemplates[$ns] = array();
4579                 }
4580                 $this->mTemplates[$ns][$dbk] = $id;
4581         }
4582
4583         /**
4584          * Return true if this cached output object predates the global or
4585          * per-article cache invalidation timestamps, or if it comes from
4586          * an incompatible older version.
4587          *
4588          * @param string $touched the affected article's last touched timestamp
4589          * @return bool
4590          * @public
4591          */
4592         function expired( $touched ) {
4593                 global $wgCacheEpoch;
4594                 return $this->getCacheTime() == -1 || // parser says it's uncacheable
4595                        $this->getCacheTime() < $touched ||
4596                        $this->getCacheTime() <= $wgCacheEpoch ||
4597                        !isset( $this->mVersion ) ||
4598                        version_compare( $this->mVersion, MW_PARSER_VERSION, "lt" );
4599         }
4600 }
4601
4602 /**
4603  * Set options of the Parser
4604  * @todo document
4605  * @package MediaWiki
4606  */
4607 class ParserOptions
4608 {
4609         # All variables are private
4610         var $mUseTeX;                    # Use texvc to expand <math> tags
4611         var $mUseDynamicDates;           # Use DateFormatter to format dates
4612         var $mInterwikiMagic;            # Interlanguage links are removed and returned in an array
4613         var $mAllowExternalImages;       # Allow external images inline
4614         var $mAllowExternalImagesFrom;   # If not, any exception?
4615         var $mSkin;                      # Reference to the preferred skin
4616         var $mDateFormat;                # Date format index
4617         var $mEditSection;               # Create "edit section" links
4618         var $mNumberHeadings;            # Automatically number headings
4619         var $mAllowSpecialInclusion;     # Allow inclusion of special pages
4620         var $mTidy;                      # Ask for tidy cleanup
4621         var $mInterfaceMessage;          # Which lang to call for PLURAL and GRAMMAR
4622
4623         var $mUser;                      # Stored user object, just used to initialise the skin
4624
4625         function getUseTeX()                        { return $this->mUseTeX; }
4626         function getUseDynamicDates()               { return $this->mUseDynamicDates; }
4627         function getInterwikiMagic()                { return $this->mInterwikiMagic; }
4628         function getAllowExternalImages()           { return $this->mAllowExternalImages; }
4629         function getAllowExternalImagesFrom()       { return $this->mAllowExternalImagesFrom; }
4630         function getDateFormat()                    { return $this->mDateFormat; }
4631         function getEditSection()                   { return $this->mEditSection; }
4632         function getNumberHeadings()                { return $this->mNumberHeadings; }
4633         function getAllowSpecialInclusion()         { return $this->mAllowSpecialInclusion; }
4634         function getTidy()                          { return $this->mTidy; }
4635         function getInterfaceMessage()              { return $this->mInterfaceMessage; }
4636
4637         function &getSkin() {
4638                 if ( !isset( $this->mSkin ) ) {
4639                         $this->mSkin = $this->mUser->getSkin();
4640                 }
4641                 return $this->mSkin;
4642         }
4643
4644         function setUseTeX( $x )                    { return wfSetVar( $this->mUseTeX, $x ); }
4645         function setUseDynamicDates( $x )           { return wfSetVar( $this->mUseDynamicDates, $x ); }
4646         function setInterwikiMagic( $x )            { return wfSetVar( $this->mInterwikiMagic, $x ); }
4647         function setAllowExternalImages( $x )       { return wfSetVar( $this->mAllowExternalImages, $x ); }
4648         function setAllowExternalImagesFrom( $x )   { return wfSetVar( $this->mAllowExternalImagesFrom, $x ); }
4649         function setDateFormat( $x )                { return wfSetVar( $this->mDateFormat, $x ); }
4650         function setEditSection( $x )               { return wfSetVar( $this->mEditSection, $x ); }
4651         function setNumberHeadings( $x )            { return wfSetVar( $this->mNumberHeadings, $x ); }
4652         function setAllowSpecialInclusion( $x )     { return wfSetVar( $this->mAllowSpecialInclusion, $x ); }
4653         function setTidy( $x )                      { return wfSetVar( $this->mTidy, $x); }
4654         function setSkin( &$x ) { $this->mSkin =& $x; }
4655         function setInterfaceMessage( $x )          { return wfSetVar( $this->mInterfaceMessage, $x); }
4656
4657         function ParserOptions( $user = null ) {
4658                 $this->initialiseFromUser( $user );
4659         }
4660
4661         /**
4662          * Get parser options
4663          * @static
4664          */
4665         function newFromUser( &$user ) {
4666                 return new ParserOptions( $user );
4667         }
4668
4669         /** Get user options */
4670         function initialiseFromUser( &$userInput ) {
4671                 global $wgUseTeX, $wgUseDynamicDates, $wgInterwikiMagic, $wgAllowExternalImages;
4672                 global $wgAllowExternalImagesFrom, $wgAllowSpecialInclusion;
4673                 $fname = 'ParserOptions::initialiseFromUser';
4674                 wfProfileIn( $fname );
4675                 if ( !$userInput ) {
4676                         global $wgUser;
4677                         if ( isset( $wgUser ) ) {
4678                                 $user = $wgUser;
4679                         } else {
4680                                 $user = new User;
4681                                 $user->setLoaded( true );
4682                         }
4683                 } else {
4684                         $user =& $userInput;
4685                 }
4686
4687                 $this->mUser = $user;
4688
4689                 $this->mUseTeX = $wgUseTeX;
4690                 $this->mUseDynamicDates = $wgUseDynamicDates;
4691                 $this->mInterwikiMagic = $wgInterwikiMagic;
4692                 $this->mAllowExternalImages = $wgAllowExternalImages;
4693                 $this->mAllowExternalImagesFrom = $wgAllowExternalImagesFrom;
4694                 $this->mSkin = null; # Deferred
4695                 $this->mDateFormat = $user->getOption( 'date' );
4696                 $this->mEditSection = true;
4697                 $this->mNumberHeadings = $user->getOption( 'numberheadings' );
4698                 $this->mAllowSpecialInclusion = $wgAllowSpecialInclusion;
4699                 $this->mTidy = false;
4700                 $this->mInterfaceMessage = false;
4701                 wfProfileOut( $fname );
4702         }
4703 }
4704
4705 /**
4706  * Callback function used by Parser::replaceLinkHolders()
4707  * to substitute link placeholders.
4708  */
4709 function &wfOutputReplaceMatches( $matches ) {
4710         global $wgOutputReplace;
4711         return $wgOutputReplace[$matches[1]];
4712 }
4713
4714 /**
4715  * Return the total number of articles
4716  */
4717 function wfNumberOfArticles() {
4718         global $wgNumberOfArticles;
4719
4720         wfLoadSiteStats();
4721         return $wgNumberOfArticles;
4722 }
4723
4724 /**
4725  * Return the number of files
4726  */
4727 function wfNumberOfFiles() {
4728         $fname = 'wfNumberOfFiles';
4729
4730         wfProfileIn( $fname );
4731         $dbr =& wfGetDB( DB_SLAVE );
4732         $numImages = $dbr->selectField('site_stats', 'ss_images', array(), $fname );
4733         wfProfileOut( $fname );
4734
4735         return $numImages;
4736 }
4737
4738 /**
4739  * Return the number of user accounts
4740  * @return integer
4741  */
4742 function wfNumberOfUsers() {
4743         wfProfileIn( 'wfNumberOfUsers' );
4744         $dbr =& wfGetDB( DB_SLAVE );
4745         $count = $dbr->selectField( 'site_stats', 'ss_users', array(), 'wfNumberOfUsers' );
4746         wfProfileOut( 'wfNumberOfUsers' );
4747         return (int)$count;
4748 }
4749
4750 /**
4751  * Return the total number of pages
4752  * @return integer
4753  */
4754 function wfNumberOfPages() {
4755         wfProfileIn( 'wfNumberOfPages' );
4756         $dbr =& wfGetDB( DB_SLAVE );
4757         $count = $dbr->selectField( 'site_stats', 'ss_total_pages', array(), 'wfNumberOfPages' );
4758         wfProfileOut( 'wfNumberOfPages' );
4759         return (int)$count;
4760 }
4761
4762 /**
4763  * Return the total number of admins
4764  *
4765  * @return integer
4766  */
4767 function wfNumberOfAdmins() {
4768         static $admins = -1;
4769         wfProfileIn( 'wfNumberOfAdmins' );
4770         if( $admins == -1 ) {
4771                 $dbr =& wfGetDB( DB_SLAVE );
4772                 $admins = $dbr->selectField( 'user_groups', 'COUNT(*)', array( 'ug_group' => 'sysop' ), 'wfNumberOfAdmins' );
4773         }
4774         wfProfileOut( 'wfNumberOfAdmins' );
4775         return (int)$admins;
4776 }
4777
4778 /**
4779  * Count the number of pages in a particular namespace
4780  *
4781  * @param $ns Namespace
4782  * @return integer
4783  */
4784 function wfPagesInNs( $ns ) {
4785         static $pageCount = array();
4786         wfProfileIn( 'wfPagesInNs' );
4787         if( !isset( $pageCount[$ns] ) ) {
4788                 $dbr =& wfGetDB( DB_SLAVE );
4789                 $pageCount[$ns] = $dbr->selectField( 'page', 'COUNT(*)', array( 'page_namespace' => $ns ), 'wfPagesInNs' );
4790         }
4791         wfProfileOut( 'wfPagesInNs' );
4792         return (int)$pageCount[$ns];
4793 }
4794
4795 /**
4796  * Get various statistics from the database
4797  * @private
4798  */
4799 function wfLoadSiteStats() {
4800         global $wgNumberOfArticles, $wgTotalViews, $wgTotalEdits;
4801         $fname = 'wfLoadSiteStats';
4802
4803         if ( -1 != $wgNumberOfArticles ) return;
4804         $dbr =& wfGetDB( DB_SLAVE );
4805         $s = $dbr->selectRow( 'site_stats',
4806                 array( 'ss_total_views', 'ss_total_edits', 'ss_good_articles' ),
4807                 array( 'ss_row_id' => 1 ), $fname
4808         );
4809
4810         if ( $s === false ) {
4811                 return;
4812         } else {
4813                 $wgTotalViews = $s->ss_total_views;
4814                 $wgTotalEdits = $s->ss_total_edits;
4815                 $wgNumberOfArticles = $s->ss_good_articles;
4816         }
4817 }
4818
4819 /**
4820  * Escape html tags
4821  * Basically replacing " > and < with HTML entities ( &quot;, &gt;, &lt;)
4822  *
4823  * @param $in String: text that might contain HTML tags.
4824  * @return string Escaped string
4825  */
4826 function wfEscapeHTMLTagsOnly( $in ) {
4827         return str_replace(
4828                 array( '"', '>', '<' ),
4829                 array( '&quot;', '&gt;', '&lt;' ),
4830                 $in );
4831 }
4832
4833 ?>