includes/Parser.php

   1 <?php
   2
   3 /**
   4  *
   5  * File for Parser and related classes
   6  *
   7  * @addtogroup Parser
   8  */
   9
  10
  11 /**
  12  * PHP Parser - Processes wiki markup (which uses a more user-friendly
  13  * syntax, such as "[[link]]" for making links), and provides a one-way
  14  * transformation of that wiki markup it into XHTML output / markup
  15  * (which in turn the browser understands, and can display).
  16  *
  17  * <pre>
  18  * There are five main entry points into the Parser class:
  19  * parse()
  20  *   produces HTML output
  21  * preSaveTransform().
  22  *   produces altered wiki markup.
  23  * preprocess()
  24  *   removes HTML comments and expands templates
  25  * cleanSig()
  26  *   Cleans a signature before saving it to preferences
  27  * extractSections()
  28  *   Extracts sections from an article for section editing
  29  *
  30  * Globals used:
  31  *    objects:   $wgLang, $wgContLang
  32  *
  33  * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
  34  *
  35  * settings:
  36  *  $wgUseTex*, $wgUseDynamicDates*, $wgInterwikiMagic*,
  37  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  38  *  $wgLocaltimezone, $wgAllowSpecialInclusion*,
  39  *  $wgMaxArticleSize*
  40  *
  41  *  * only within ParserOptions
  42  * </pre>
  43  *
  44  * @addtogroup Parser
  45  */
  46 class Parser
  47 {
  48         /**
  49          * Update this version number when the ParserOutput format
  50          * changes in an incompatible way, so the parser cache
  51          * can automatically discard old data.
  52          */
  53         const VERSION = '1.6.4';
  54
  55         # Flags for Parser::setFunctionHook
  56         # Also available as global constants from Defines.php
  57         const SFH_NO_HASH = 1;
  58         const SFH_OBJECT_ARGS = 2;
  59
  60         # Constants needed for external link processing
  61         # Everything except bracket, space, or control characters
  62         const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F]';
  63         const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F]+)
  64                 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sx';
  65
  66         // State constants for the definition list colon extraction
  67         const COLON_STATE_TEXT = 0;
  68         const COLON_STATE_TAG = 1;
  69         const COLON_STATE_TAGSTART = 2;
  70         const COLON_STATE_CLOSETAG = 3;
  71         const COLON_STATE_TAGSLASH = 4;
  72         const COLON_STATE_COMMENT = 5;
  73         const COLON_STATE_COMMENTDASH = 6;
  74         const COLON_STATE_COMMENTDASHDASH = 7;
  75
  76         // Flags for preprocessToDom
  77         const PTD_FOR_INCLUSION = 1;
  78
  79         // Allowed values for $this->mOutputType
  80         // Parameter to startExternalParse().
  81         const OT_HTML = 1;
  82         const OT_WIKI = 2;
  83         const OT_PREPROCESS = 3;
  84         const OT_MSG = 3;
  85
  86         /**#@+
  87          * @private
  88          */
  89         # Persistent:
  90         var $mTagHooks, $mTransparentTagHooks, $mFunctionHooks, $mFunctionSynonyms, $mVariables,
  91                 $mImageParams, $mImageParamsMagicArray, $mStripList, $mMarkerSuffix, $mMarkerIndex,
  92                 $mExtLinkBracketedRegex, $mPreprocessor, $mDefaultStripList, $mVarCache, $mConf;
  93
  94
  95         # Cleared with clearState():
  96         var $mOutput, $mAutonumber, $mDTopen, $mStripState;
  97         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
  98         var $mInterwikiLinkHolders, $mLinkHolders;
  99         var $mIncludeSizes, $mPPNodeCount, $mDefaultSort;
 100         var $mTplExpandCache; // empty-frame expansion cache
 101         var $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
 102
 103         # Temporary
 104         # These are variables reset at least once per parse regardless of $clearState
 105         var $mOptions,      // ParserOptions object
 106                 $mTitle,        // Title context, used for self-link rendering and similar things
 107                 $mOutputType,   // Output type, one of the OT_xxx constants
 108                 $ot,            // Shortcut alias, see setOutputType()
 109                 $mRevisionId,   // ID to display in {{REVISIONID}} tags
 110                 $mRevisionTimestamp, // The timestamp of the specified revision ID
 111                 $mRevIdForTs;   // The revision ID which was used to fetch the timestamp
 112
 113         /**#@-*/
 114
 115         /**
 116          * Constructor
 117          *
 118          * @public
 119          */
 120         function __construct( $conf = array() ) {
 121                 $this->mConf = $conf;
 122                 $this->mTagHooks = array();
 123                 $this->mTransparentTagHooks = array();
 124                 $this->mFunctionHooks = array();
 125                 $this->mFunctionSynonyms = array( 0 => array(), 1 => array() );
 126                 $this->mDefaultStripList = $this->mStripList = array( 'nowiki', 'gallery' );
 127                 $this->mMarkerSuffix = "-QINU\x7f";
 128                 $this->mExtLinkBracketedRegex = '/\[(\b(' . wfUrlProtocols() . ')'.
 129                         '[^][<>"\\x00-\\x20\\x7F]+) *([^\]\\x0a\\x0d]*?)\]/S';
 130                 $this->mVarCache = array();
 131                 if ( isset( $conf['preprocessorClass'] ) ) {
 132                         $this->mPreprocessorClass = $conf['preprocessorClass'];
 133                 } else {
 134                         $this->mPreprocessorClass = 'Preprocessor_DOM';
 135                 }
 136                 $this->mMarkerIndex = 0;
 137                 $this->mFirstCall = true;
 138         }
 139
 140         /**
 141          * Do various kinds of initialisation on the first call of the parser
 142          */
 143         function firstCallInit() {
 144                 if ( !$this->mFirstCall ) {
 145                         return;
 146                 }
 147                 $this->mFirstCall = false;
 148
 149                 wfProfileIn( __METHOD__ );
 150
 151                 $this->setHook( 'pre', array( $this, 'renderPreTag' ) );
 152                 CoreParserFunctions::register( $this );
 153                 $this->initialiseVariables();
 154
 155                 wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
 156                 wfProfileOut( __METHOD__ );
 157         }
 158
 159         /**
 160          * Clear Parser state
 161          *
 162          * @private
 163          */
 164         function clearState() {
 165                 wfProfileIn( __METHOD__ );
 166                 if ( $this->mFirstCall ) {
 167                         $this->firstCallInit();
 168                 }
 169                 $this->mOutput = new ParserOutput;
 170                 $this->mAutonumber = 0;
 171                 $this->mLastSection = '';
 172                 $this->mDTopen = false;
 173                 $this->mIncludeCount = array();
 174                 $this->mStripState = new StripState;
 175                 $this->mArgStack = false;
 176                 $this->mInPre = false;
 177                 $this->mInterwikiLinkHolders = array(
 178                         'texts' => array(),
 179                         'titles' => array()
 180                 );
 181                 $this->mLinkHolders = array(
 182                         'namespaces' => array(),
 183                         'dbkeys' => array(),
 184                         'queries' => array(),
 185                         'texts' => array(),
 186                         'titles' => array()
 187                 );
 188                 $this->mRevisionTimestamp = $this->mRevisionId = null;
 189
 190                 /**
 191                  * Prefix for temporary replacement strings for the multipass parser.
 192                  * \x07 should never appear in input as it's disallowed in XML.
 193                  * Using it at the front also gives us a little extra robustness
 194                  * since it shouldn't match when butted up against identifier-like
 195                  * string constructs.
 196                  *
 197                  * Must not consist of all title characters, or else it will change
 198                  * the behaviour of <nowiki> in a link.
 199                  */
 200                 #$this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 201                 # Changed to \x7f to allow XML double-parsing -- TS
 202                 $this->mUniqPrefix = "\x7fUNIQ" . Parser::getRandomString();
 203
 204
 205                 # Clear these on every parse, bug 4549
 206                 $this->mTplExpandCache = $this->mTplRedirCache = $this->mTplDomCache = array();
 207
 208                 $this->mShowToc = true;
 209                 $this->mForceTocPosition = false;
 210                 $this->mIncludeSizes = array(
 211                         'post-expand' => 0,
 212                         'arg' => 0,
 213                 );
 214                 $this->mPPNodeCount = 0;
 215                 $this->mDefaultSort = false;
 216                 $this->mHeadings = array();
 217                 $this->mDoubleUnderscores = array();
 218
 219                 # Fix cloning
 220                 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
 221                         $this->mPreprocessor = null;
 222                 }
 223
 224                 wfRunHooks( 'ParserClearState', array( &$this ) );
 225                 wfProfileOut( __METHOD__ );
 226         }
 227
 228         function setOutputType( $ot ) {
 229                 $this->mOutputType = $ot;
 230                 // Shortcut alias
 231                 $this->ot = array(
 232                         'html' => $ot == self::OT_HTML,
 233                         'wiki' => $ot == self::OT_WIKI,
 234                         'pre' => $ot == self::OT_PREPROCESS,
 235                 );
 236         }
 237
 238         /**
 239          * Set the context title
 240          */
 241         function setTitle( $t ) {
 242                 if ( !$t || $t instanceof FakeTitle ) {
 243                         $t = Title::newFromText( 'NO TITLE' );
 244                 }
 245                 if ( strval( $t->getFragment() ) !== '' ) {
 246                         # Strip the fragment to avoid various odd effects
 247                         $this->mTitle = clone $t;
 248                         $this->mTitle->setFragment( '' );
 249                 } else {
 250                         $this->mTitle = $t;
 251                 }
 252         }
 253
 254         /**
 255          * Accessor for mUniqPrefix.
 256          *
 257          * @public
 258          */
 259         function uniqPrefix() {
 260                 if( !isset( $this->mUniqPrefix ) ) {
 261                         // @fixme this is probably *horribly wrong*
 262                         // LanguageConverter seems to want $wgParser's uniqPrefix, however
 263                         // if this is called for a parser cache hit, the parser may not
 264                         // have ever been initialized in the first place.
 265                         // Not really sure what the heck is supposed to be going on here.
 266                         return '';
 267                         //throw new MWException( "Accessing uninitialized mUniqPrefix" );
 268                 }
 269                 return $this->mUniqPrefix;
 270         }
 271
 272         /**
 273          * Convert wikitext to HTML
 274          * Do not call this function recursively.
 275          *
 276          * @param string $text Text we want to parse
 277          * @param Title &$title A title object
 278          * @param array $options
 279          * @param boolean $linestart
 280          * @param boolean $clearState
 281          * @param int $revid number to pass in {{REVISIONID}}
 282          * @return ParserOutput a ParserOutput
 283          */
 284         public function parse( $text, &$title, $options, $linestart = true, $clearState = true, $revid = null ) {
 285                 /**
 286                  * First pass--just handle <nowiki> sections, pass the rest off
 287                  * to internalParse() which does all the real work.
 288                  */
 289
 290                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
 291                 $fname = 'Parser::parse-' . wfGetCaller();
 292                 wfProfileIn( __METHOD__ );
 293                 wfProfileIn( $fname );
 294
 295                 if ( $clearState ) {
 296                         $this->clearState();
 297                 }
 298
 299                 $this->mOptions = $options;
 300                 $this->setTitle( $title );
 301                 $oldRevisionId = $this->mRevisionId;
 302                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 303                 if( $revid !== null ) {
 304                         $this->mRevisionId = $revid;
 305                         $this->mRevisionTimestamp = null;
 306                 }
 307                 $this->setOutputType( self::OT_HTML );
 308                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 309                 # No more strip!
 310                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 311                 $text = $this->internalParse( $text );
 312                 $text = $this->mStripState->unstripGeneral( $text );
 313
 314                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 315                 $fixtags = array(
 316                         # french spaces, last one Guillemet-left
 317                         # only if there is something before the space
 318                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&nbsp;\\2',
 319                         # french spaces, Guillemet-right
 320                         '/(\\302\\253) /' => '\\1&nbsp;',
 321                 );
 322                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 323
 324                 # only once and last
 325                 $text = $this->doBlockLevels( $text, $linestart );
 326
 327                 $this->replaceLinkHolders( $text );
 328
 329                 # the position of the parserConvert() call should not be changed. it
 330                 # assumes that the links are all replaced and the only thing left
 331                 # is the <nowiki> mark.
 332                 # Side-effects: this calls $this->mOutput->setTitleText()
 333                 $text = $wgContLang->parserConvert( $text, $this );
 334
 335                 $text = $this->mStripState->unstripNoWiki( $text );
 336
 337                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 338
 339 //!JF Move to its own function
 340
 341                 $uniq_prefix = $this->mUniqPrefix;
 342                 $matches = array();
 343                 $elements = array_keys( $this->mTransparentTagHooks );
 344                 $text = Parser::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 345
 346                 foreach( $matches as $marker => $data ) {
 347                         list( $element, $content, $params, $tag ) = $data;
 348                         $tagName = strtolower( $element );
 349                         if( isset( $this->mTransparentTagHooks[$tagName] ) ) {
 350                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName],
 351                                         array( $content, $params, $this ) );
 352                         } else {
 353                                 $output = $tag;
 354                         }
 355                         $this->mStripState->general->setPair( $marker, $output );
 356                 }
 357                 $text = $this->mStripState->unstripGeneral( $text );
 358
 359                 $text = Sanitizer::normalizeCharReferences( $text );
 360
 361                 if (($wgUseTidy and $this->mOptions->mTidy) or $wgAlwaysUseTidy) {
 362                         $text = Parser::tidy($text);
 363                 } else {
 364                         # attempt to sanitize at least some nesting problems
 365                         # (bug #2702 and quite a few others)
 366                         $tidyregs = array(
 367                                 # ''Something [http://www.cool.com cool''] -->
 368                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 369                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 370                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 371                                 # fix up an anchor inside another anchor, only
 372                                 # at least for a single single nested link (bug 3695)
 373                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 374                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 375                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 376                                 # contains a div, so fix it up here; replace
 377                                 # div with escaped text
 378                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 379                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 380                                 # remove empty italic or bold tag pairs, some
 381                                 # introduced by rules above
 382                                 '/<([bi])><\/\\1>/' => '',
 383                         );
 384
 385                         $text = preg_replace(
 386                                 array_keys( $tidyregs ),
 387                                 array_values( $tidyregs ),
 388                                 $text );
 389                 }
 390
 391                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 392
 393                 # Information on include size limits, for the benefit of users who try to skirt them
 394                 if ( $this->mOptions->getEnableLimitReport() ) {
 395                         $max = $this->mOptions->getMaxIncludeSize();
 396                         $limitReport =
 397                                 "NewPP limit report\n" .
 398                                 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->mMaxPPNodeCount}\n" .
 399                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 400                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n";
 401                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 402                         $text .= "\n<!-- \n$limitReport-->\n";
 403                 }
 404                 $this->mOutput->setText( $text );
 405                 $this->mRevisionId = $oldRevisionId;
 406                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 407                 wfProfileOut( $fname );
 408                 wfProfileOut( __METHOD__ );
 409
 410                 return $this->mOutput;
 411         }
 412
 413         /**
 414          * Recursive parser entry point that can be called from an extension tag
 415          * hook.
 416          */
 417         function recursiveTagParse( $text ) {
 418                 wfProfileIn( __METHOD__ );
 419                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 420                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 421                 $text = $this->internalParse( $text );
 422                 wfProfileOut( __METHOD__ );
 423                 return $text;
 424         }
 425
 426         /**
 427          * Expand templates and variables in the text, producing valid, static wikitext.
 428          * Also removes comments.
 429          */
 430         function preprocess( $text, $title, $options, $revid = null ) {
 431                 wfProfileIn( __METHOD__ );
 432                 $this->clearState();
 433                 $this->setOutputType( self::OT_PREPROCESS );
 434                 $this->mOptions = $options;
 435                 $this->setTitle( $title );
 436                 if( $revid !== null ) {
 437                         $this->mRevisionId = $revid;
 438                 }
 439                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 440                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 441                 $text = $this->replaceVariables( $text );
 442                 $text = $this->mStripState->unstripBoth( $text );
 443                 wfProfileOut( __METHOD__ );
 444                 return $text;
 445         }
 446
 447         /**
 448          * Get a random string
 449          *
 450          * @private
 451          * @static
 452          */
 453         function getRandomString() {
 454                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 455         }
 456
 457         function &getTitle() { return $this->mTitle; }
 458         function getOptions() { return $this->mOptions; }
 459
 460         function getFunctionLang() {
 461                 global $wgLang, $wgContLang;
 462
 463                 $target = $this->mOptions->getTargetLanguage();
 464                 if ( $target !== null ) {
 465                         return $target;
 466                 } else {
 467                         return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 468                 }
 469         }
 470
 471         /**
 472          * Get a preprocessor object
 473          */
 474         function getPreprocessor() {
 475                 if ( !isset( $this->mPreprocessor ) ) {
 476                         $class = $this->mPreprocessorClass;
 477                         $this->mPreprocessor = new $class( $this );
 478                 }
 479                 return $this->mPreprocessor;
 480         }
 481
 482         /**
 483          * Replaces all occurrences of HTML-style comments and the given tags
 484          * in the text with a random marker and returns the next text. The output
 485          * parameter $matches will be an associative array filled with data in
 486          * the form:
 487          *   'UNIQ-xxxxx' => array(
 488          *     'element',
 489          *     'tag content',
 490          *     array( 'param' => 'x' ),
 491          *     '<element param="x">tag content</element>' ) )
 492          *
 493          * @param $elements list of element names. Comments are always extracted.
 494          * @param $text Source text string.
 495          * @param $uniq_prefix
 496          *
 497          * @public
 498          * @static
 499          */
 500         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 501                 static $n = 1;
 502                 $stripped = '';
 503                 $matches = array();
 504
 505                 $taglist = implode( '|', $elements );
 506                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?>)|<(!--)/i";
 507
 508                 while ( '' != $text ) {
 509                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 510                         $stripped .= $p[0];
 511                         if( count( $p ) < 5 ) {
 512                                 break;
 513                         }
 514                         if( count( $p ) > 5 ) {
 515                                 // comment
 516                                 $element    = $p[4];
 517                                 $attributes = '';
 518                                 $close      = '';
 519                                 $inside     = $p[5];
 520                         } else {
 521                                 // tag
 522                                 $element    = $p[1];
 523                                 $attributes = $p[2];
 524                                 $close      = $p[3];
 525                                 $inside     = $p[4];
 526                         }
 527
 528                         $marker = "$uniq_prefix-$element-" . sprintf('%08X', $n++) . $this->mMarkerSuffix;
 529                         $stripped .= $marker;
 530
 531                         if ( $close === '/>' ) {
 532                                 // Empty element tag, <tag />
 533                                 $content = null;
 534                                 $text = $inside;
 535                                 $tail = null;
 536                         } else {
 537                                 if( $element == '!--' ) {
 538                                         $end = '/(-->)/';
 539                                 } else {
 540                                         $end = "/(<\\/$element\\s*>)/i";
 541                                 }
 542                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 543                                 $content = $q[0];
 544                                 if( count( $q ) < 3 ) {
 545                                         # No end tag -- let it run out to the end of the text.
 546                                         $tail = '';
 547                                         $text = '';
 548                                 } else {
 549                                         $tail = $q[1];
 550                                         $text = $q[2];
 551                                 }
 552                         }
 553
 554                         $matches[$marker] = array( $element,
 555                                 $content,
 556                                 Sanitizer::decodeTagAttributes( $attributes ),
 557                                 "<$element$attributes$close$content$tail" );
 558                 }
 559                 return $stripped;
 560         }
 561
 562         /**
 563          * Get a list of strippable XML-like elements
 564          */
 565         function getStripList() {
 566                 global $wgRawHtml;
 567                 $elements = $this->mStripList;
 568                 if( $wgRawHtml ) {
 569                         $elements[] = 'html';
 570                 }
 571                 if( $this->mOptions->getUseTeX() ) {
 572                         $elements[] = 'math';
 573                 }
 574                 return $elements;
 575         }
 576
 577         /**
 578          * @deprecated use replaceVariables
 579          */
 580         function strip( $text, $state, $stripcomments = false , $dontstrip = array () ) {
 581                 return $text;
 582         }
 583
 584         /**
 585          * Restores pre, math, and other extensions removed by strip()
 586          *
 587          * always call unstripNoWiki() after this one
 588          * @private
 589          * @deprecated use $this->mStripState->unstrip()
 590          */
 591         function unstrip( $text, $state ) {
 592                 return $state->unstripGeneral( $text );
 593         }
 594
 595         /**
 596          * Always call this after unstrip() to preserve the order
 597          *
 598          * @private
 599          * @deprecated use $this->mStripState->unstrip()
 600          */
 601         function unstripNoWiki( $text, $state ) {
 602                 return $state->unstripNoWiki( $text );
 603         }
 604
 605         /**
 606          * @deprecated use $this->mStripState->unstripBoth()
 607          */
 608         function unstripForHTML( $text ) {
 609                 return $this->mStripState->unstripBoth( $text );
 610         }
 611
 612         /**
 613          * Add an item to the strip state
 614          * Returns the unique tag which must be inserted into the stripped text
 615          * The tag will be replaced with the original text in unstrip()
 616          *
 617          * @private
 618          */
 619         function insertStripItem( $text ) {
 620                 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-{$this->mMarkerSuffix}";
 621                 $this->mMarkerIndex++;
 622                 $this->mStripState->general->setPair( $rnd, $text );
 623                 return $rnd;
 624         }
 625
 626         /**
 627          * Interface with html tidy, used if $wgUseTidy = true.
 628          * If tidy isn't able to correct the markup, the original will be
 629          * returned in all its glory with a warning comment appended.
 630          *
 631          * Either the external tidy program or the in-process tidy extension
 632          * will be used depending on availability. Override the default
 633          * $wgTidyInternal setting to disable the internal if it's not working.
 634          *
 635          * @param string $text Hideous HTML input
 636          * @return string Corrected HTML output
 637          * @public
 638          * @static
 639          */
 640         function tidy( $text ) {
 641                 global $wgTidyInternal;
 642                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 643 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 644 '<head><title>test</title></head><body>'.$text.'</body></html>';
 645                 if( $wgTidyInternal ) {
 646                         $correctedtext = Parser::internalTidy( $wrappedtext );
 647                 } else {
 648                         $correctedtext = Parser::externalTidy( $wrappedtext );
 649                 }
 650                 if( is_null( $correctedtext ) ) {
 651                         wfDebug( "Tidy error detected!\n" );
 652                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 653                 }
 654                 return $correctedtext;
 655         }
 656
 657         /**
 658          * Spawn an external HTML tidy process and get corrected markup back from it.
 659          *
 660          * @private
 661          * @static
 662          */
 663         function externalTidy( $text ) {
 664                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 665                 $fname = 'Parser::externalTidy';
 666                 wfProfileIn( $fname );
 667
 668                 $cleansource = '';
 669                 $opts = ' -utf8';
 670
 671                 $descriptorspec = array(
 672                         0 => array('pipe', 'r'),
 673                         1 => array('pipe', 'w'),
 674                         2 => array('file', wfGetNull(), 'a')
 675                 );
 676                 $pipes = array();
 677                 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 678                 if (is_resource($process)) {
 679                         // Theoretically, this style of communication could cause a deadlock
 680                         // here. If the stdout buffer fills up, then writes to stdin could
 681                         // block. This doesn't appear to happen with tidy, because tidy only
 682                         // writes to stdout after it's finished reading from stdin. Search
 683                         // for tidyParseStdin and tidySaveStdout in console/tidy.c
 684                         fwrite($pipes[0], $text);
 685                         fclose($pipes[0]);
 686                         while (!feof($pipes[1])) {
 687                                 $cleansource .= fgets($pipes[1], 1024);
 688                         }
 689                         fclose($pipes[1]);
 690                         proc_close($process);
 691                 }
 692
 693                 wfProfileOut( $fname );
 694
 695                 if( $cleansource == '' && $text != '') {
 696                         // Some kind of error happened, so we couldn't get the corrected text.
 697                         // Just give up; we'll use the source text and append a warning.
 698                         return null;
 699                 } else {
 700                         return $cleansource;
 701                 }
 702         }
 703
 704         /**
 705          * Use the HTML tidy PECL extension to use the tidy library in-process,
 706          * saving the overhead of spawning a new process.
 707          *
 708          * 'pear install tidy' should be able to compile the extension module.
 709          *
 710          * @private
 711          * @static
 712          */
 713         function internalTidy( $text ) {
 714                 global $wgTidyConf, $IP, $wgDebugTidy;
 715                 $fname = 'Parser::internalTidy';
 716                 wfProfileIn( $fname );
 717
 718                 $tidy = new tidy;
 719                 $tidy->parseString( $text, $wgTidyConf, 'utf8' );
 720                 $tidy->cleanRepair();
 721                 if( $tidy->getStatus() == 2 ) {
 722                         // 2 is magic number for fatal error
 723                         // http://www.php.net/manual/en/function.tidy-get-status.php
 724                         $cleansource = null;
 725                 } else {
 726                         $cleansource = tidy_get_output( $tidy );
 727                 }
 728                 if ( $wgDebugTidy && $tidy->getStatus() > 0 ) {
 729                         $cleansource .= "<!--\nTidy reports:\n" .
 730                                 str_replace( '-->', '--&gt;', $tidy->errorBuffer ) .
 731                                 "\n-->";
 732                 }
 733
 734                 wfProfileOut( $fname );
 735                 return $cleansource;
 736         }
 737
 738         /**
 739          * parse the wiki syntax used to render tables
 740          *
 741          * @private
 742          */
 743         function doTableStuff ( $text ) {
 744                 $fname = 'Parser::doTableStuff';
 745                 wfProfileIn( $fname );
 746
 747                 $lines = explode ( "\n" , $text );
 748                 $td_history = array (); // Is currently a td tag open?
 749                 $last_tag_history = array (); // Save history of last lag activated (td, th or caption)
 750                 $tr_history = array (); // Is currently a tr tag open?
 751                 $tr_attributes = array (); // history of tr attributes
 752                 $has_opened_tr = array(); // Did this table open a <tr> element?
 753                 $indent_level = 0; // indent level of the table
 754                 foreach ( $lines as $key => $line )
 755                 {
 756                         $line = trim ( $line );
 757
 758                         if( $line == '' ) { // empty line, go to next line
 759                                 continue;
 760                         }
 761                         $first_character = $line{0};
 762                         $matches = array();
 763
 764                         if ( preg_match( '/^(:*)\{\|(.*)$/' , $line , $matches ) ) {
 765                                 // First check if we are starting a new table
 766                                 $indent_level = strlen( $matches[1] );
 767
 768                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 769                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'table' );
 770
 771                                 $lines[$key] = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 772                                 array_push ( $td_history , false );
 773                                 array_push ( $last_tag_history , '' );
 774                                 array_push ( $tr_history , false );
 775                                 array_push ( $tr_attributes , '' );
 776                                 array_push ( $has_opened_tr , false );
 777                         } else if ( count ( $td_history ) == 0 ) {
 778                                 // Don't do any of the following
 779                                 continue;
 780                         } else if ( substr ( $line , 0 , 2 ) == '|}' ) {
 781                                 // We are ending a table
 782                                 $line = '</table>' . substr ( $line , 2 );
 783                                 $last_tag = array_pop ( $last_tag_history );
 784
 785                                 if ( !array_pop ( $has_opened_tr ) ) {
 786                                         $line = "<tr><td></td></tr>{$line}";
 787                                 }
 788
 789                                 if ( array_pop ( $tr_history ) ) {
 790                                         $line = "</tr>{$line}";
 791                                 }
 792
 793                                 if ( array_pop ( $td_history ) ) {
 794                                         $line = "</{$last_tag}>{$line}";
 795                                 }
 796                                 array_pop ( $tr_attributes );
 797                                 $lines[$key] = $line . str_repeat( '</dd></dl>' , $indent_level );
 798                         } else if ( substr ( $line , 0 , 2 ) == '|-' ) {
 799                                 // Now we have a table row
 800                                 $line = preg_replace( '#^\|-+#', '', $line );
 801
 802                                 // Whats after the tag is now only attributes
 803                                 $attributes = $this->mStripState->unstripBoth( $line );
 804                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'tr' );
 805                                 array_pop ( $tr_attributes );
 806                                 array_push ( $tr_attributes , $attributes );
 807
 808                                 $line = '';
 809                                 $last_tag = array_pop ( $last_tag_history );
 810                                 array_pop ( $has_opened_tr );
 811                                 array_push ( $has_opened_tr , true );
 812
 813                                 if ( array_pop ( $tr_history ) ) {
 814                                         $line = '</tr>';
 815                                 }
 816
 817                                 if ( array_pop ( $td_history ) ) {
 818                                         $line = "</{$last_tag}>{$line}";
 819                                 }
 820
 821                                 $lines[$key] = $line;
 822                                 array_push ( $tr_history , false );
 823                                 array_push ( $td_history , false );
 824                                 array_push ( $last_tag_history , '' );
 825                         }
 826                         else if ( $first_character == '|' || $first_character == '!' || substr ( $line , 0 , 2 )  == '|+' ) {
 827                                 // This might be cell elements, td, th or captions
 828                                 if ( substr ( $line , 0 , 2 ) == '|+' ) {
 829                                         $first_character = '+';
 830                                         $line = substr ( $line , 1 );
 831                                 }
 832
 833                                 $line = substr ( $line , 1 );
 834
 835                                 if ( $first_character == '!' ) {
 836                                         $line = str_replace ( '!!' , '||' , $line );
 837                                 }
 838
 839                                 // Split up multiple cells on the same line.
 840                                 // FIXME : This can result in improper nesting of tags processed
 841                                 // by earlier parser steps, but should avoid splitting up eg
 842                                 // attribute values containing literal "||".
 843                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 844
 845                                 $lines[$key] = '';
 846
 847                                 // Loop through each table cell
 848                                 foreach ( $cells as $cell )
 849                                 {
 850                                         $previous = '';
 851                                         if ( $first_character != '+' )
 852                                         {
 853                                                 $tr_after = array_pop ( $tr_attributes );
 854                                                 if ( !array_pop ( $tr_history ) ) {
 855                                                         $previous = "<tr{$tr_after}>\n";
 856                                                 }
 857                                                 array_push ( $tr_history , true );
 858                                                 array_push ( $tr_attributes , '' );
 859                                                 array_pop ( $has_opened_tr );
 860                                                 array_push ( $has_opened_tr , true );
 861                                         }
 862
 863                                         $last_tag = array_pop ( $last_tag_history );
 864
 865                                         if ( array_pop ( $td_history ) ) {
 866                                                 $previous = "</{$last_tag}>{$previous}";
 867                                         }
 868
 869                                         if ( $first_character == '|' ) {
 870                                                 $last_tag = 'td';
 871                                         } else if ( $first_character == '!' ) {
 872                                                 $last_tag = 'th';
 873                                         } else if ( $first_character == '+' ) {
 874                                                 $last_tag = 'caption';
 875                                         } else {
 876                                                 $last_tag = '';
 877                                         }
 878
 879                                         array_push ( $last_tag_history , $last_tag );
 880
 881                                         // A cell could contain both parameters and data
 882                                         $cell_data = explode ( '|' , $cell , 2 );
 883
 884                                         // Bug 553: Note that a '|' inside an invalid link should not
 885                                         // be mistaken as delimiting cell parameters
 886                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 887                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 888                                         } else if ( count ( $cell_data ) == 1 )
 889                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 890                                         else {
 891                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 892                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 893                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 894                                         }
 895
 896                                         $lines[$key] .= $cell;
 897                                         array_push ( $td_history , true );
 898                                 }
 899                         }
 900                 }
 901
 902                 // Closing open td, tr && table
 903                 while ( count ( $td_history ) > 0 )
 904                 {
 905                         if ( array_pop ( $td_history ) ) {
 906                                 $lines[] = '</td>' ;
 907                         }
 908                         if ( array_pop ( $tr_history ) ) {
 909                                 $lines[] = '</tr>' ;
 910                         }
 911                         if ( !array_pop ( $has_opened_tr ) ) {
 912                                 $lines[] = "<tr><td></td></tr>" ;
 913                         }
 914
 915                         $lines[] = '</table>' ;
 916                 }
 917
 918                 $output = implode ( "\n" , $lines ) ;
 919
 920                 // special case: don't return empty table
 921                 if( $output == "<table>\n<tr><td></td></tr>\n</table>" ) {
 922                         $output = '';
 923                 }
 924
 925                 wfProfileOut( $fname );
 926
 927                 return $output;
 928         }
 929
 930         /**
 931          * Helper function for parse() that transforms wiki markup into
 932          * HTML. Only called for $mOutputType == self::OT_HTML.
 933          *
 934          * @private
 935          */
 936         function internalParse( $text ) {
 937                 $isMain = true;
 938                 $fname = 'Parser::internalParse';
 939                 wfProfileIn( $fname );
 940
 941                 # Hook to suspend the parser in this state
 942                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
 943                         wfProfileOut( $fname );
 944                         return $text ;
 945                 }
 946
 947                 $text = $this->replaceVariables( $text );
 948                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
 949                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
 950
 951                 // Tables need to come after variable replacement for things to work
 952                 // properly; putting them before other transformations should keep
 953                 // exciting things like link expansions from showing up in surprising
 954                 // places.
 955                 $text = $this->doTableStuff( $text );
 956
 957                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 958
 959                 $text = $this->doDoubleUnderscore( $text );
 960                 $text = $this->doHeadings( $text );
 961                 if($this->mOptions->getUseDynamicDates()) {
 962                         $df =& DateFormatter::getInstance();
 963                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 964                 }
 965                 $text = $this->doAllQuotes( $text );
 966                 $text = $this->replaceInternalLinks( $text );
 967                 $text = $this->replaceExternalLinks( $text );
 968
 969                 # replaceInternalLinks may sometimes leave behind
 970                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 971                 $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
 972
 973                 $text = $this->doMagicLinks( $text );
 974                 $text = $this->formatHeadings( $text, $isMain );
 975
 976                 wfProfileOut( $fname );
 977                 return $text;
 978         }
 979
 980         /**
 981          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 982          * magic external links.
 983          *
 984          * @private
 985          */
 986         function doMagicLinks( $text ) {
 987                 wfProfileIn( __METHOD__ );
 988                 $text = preg_replace_callback(
 989                         '!(?:                           # Start cases
 990                             <a.*?</a> |                 # Skip link text
 991                             <.*?> |                     # Skip stuff inside HTML elements
 992                             (?:RFC|PMID)\s+([0-9]+) |   # RFC or PMID, capture number as m[1]
 993                             ISBN\s+(\b                  # ISBN, capture number as m[2]
 994                                      (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
 995                                      (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
 996                                      [0-9Xx]                 # check digit
 997                                    \b)
 998                         )!x', array( &$this, 'magicLinkCallback' ), $text );
 999                 wfProfileOut( __METHOD__ );
1000                 return $text;
1001         }
1002
1003         function magicLinkCallback( $m ) {
1004                 if ( substr( $m[0], 0, 1 ) == '<' ) {
1005                         # Skip HTML element
1006                         return $m[0];
1007                 } elseif ( substr( $m[0], 0, 4 ) == 'ISBN' ) {
1008                         $isbn = $m[2];
1009                         $num = strtr( $isbn, array(
1010                                 '-' => '',
1011                                 ' ' => '',
1012                                 'x' => 'X',
1013                         ));
1014                         $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
1015                         $text = '<a href="' .
1016                                 $titleObj->escapeLocalUrl() .
1017                                 "\" class=\"internal\">ISBN $isbn</a>";
1018                 } else {
1019                         if ( substr( $m[0], 0, 3 ) == 'RFC' ) {
1020                                 $keyword = 'RFC';
1021                                 $urlmsg = 'rfcurl';
1022                                 $id = $m[1];
1023                         } elseif ( substr( $m[0], 0, 4 ) == 'PMID' ) {
1024                                 $keyword = 'PMID';
1025                                 $urlmsg = 'pubmedurl';
1026                                 $id = $m[1];
1027                         } else {
1028                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1029                                         substr($m[0], 0, 20 ) . '"' );
1030                         }
1031
1032                         $url = wfMsg( $urlmsg, $id);
1033                         $sk = $this->mOptions->getSkin();
1034                         $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
1035                         $text = "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
1036                 }
1037                 return $text;
1038         }
1039
1040         /**
1041          * Parse headers and return html
1042          *
1043          * @private
1044          */
1045         function doHeadings( $text ) {
1046                 $fname = 'Parser::doHeadings';
1047                 wfProfileIn( $fname );
1048                 for ( $i = 6; $i >= 1; --$i ) {
1049                         $h = str_repeat( '=', $i );
1050                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1051                           "<h$i>\\1</h$i>", $text );
1052                 }
1053                 wfProfileOut( $fname );
1054                 return $text;
1055         }
1056
1057         /**
1058          * Replace single quotes with HTML markup
1059          * @private
1060          * @return string the altered text
1061          */
1062         function doAllQuotes( $text ) {
1063                 $fname = 'Parser::doAllQuotes';
1064                 wfProfileIn( $fname );
1065                 $outtext = '';
1066                 $lines = explode( "\n", $text );
1067                 foreach ( $lines as $line ) {
1068                         $outtext .= $this->doQuotes ( $line ) . "\n";
1069                 }
1070                 $outtext = substr($outtext, 0,-1);
1071                 wfProfileOut( $fname );
1072                 return $outtext;
1073         }
1074
1075         /**
1076          * Helper function for doAllQuotes()
1077          */
1078         public function doQuotes( $text ) {
1079                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1080                 if ( count( $arr ) == 1 )
1081                         return $text;
1082                 else
1083                 {
1084                         # First, do some preliminary work. This may shift some apostrophes from
1085                         # being mark-up to being text. It also counts the number of occurrences
1086                         # of bold and italics mark-ups.
1087                         $i = 0;
1088                         $numbold = 0;
1089                         $numitalics = 0;
1090                         foreach ( $arr as $r )
1091                         {
1092                                 if ( ( $i % 2 ) == 1 )
1093                                 {
1094                                         # If there are ever four apostrophes, assume the first is supposed to
1095                                         # be text, and the remaining three constitute mark-up for bold text.
1096                                         if ( strlen( $arr[$i] ) == 4 )
1097                                         {
1098                                                 $arr[$i-1] .= "'";
1099                                                 $arr[$i] = "'''";
1100                                         }
1101                                         # If there are more than 5 apostrophes in a row, assume they're all
1102                                         # text except for the last 5.
1103                                         else if ( strlen( $arr[$i] ) > 5 )
1104                                         {
1105                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1106                                                 $arr[$i] = "'''''";
1107                                         }
1108                                         # Count the number of occurrences of bold and italics mark-ups.
1109                                         # We are not counting sequences of five apostrophes.
1110                                         if ( strlen( $arr[$i] ) == 2 )      { $numitalics++;             }
1111                                         else if ( strlen( $arr[$i] ) == 3 ) { $numbold++;                }
1112                                         else if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
1113                                 }
1114                                 $i++;
1115                         }
1116
1117                         # If there is an odd number of both bold and italics, it is likely
1118                         # that one of the bold ones was meant to be an apostrophe followed
1119                         # by italics. Which one we cannot know for certain, but it is more
1120                         # likely to be one that has a single-letter word before it.
1121                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1122                         {
1123                                 $i = 0;
1124                                 $firstsingleletterword = -1;
1125                                 $firstmultiletterword = -1;
1126                                 $firstspace = -1;
1127                                 foreach ( $arr as $r )
1128                                 {
1129                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1130                                         {
1131                                                 $x1 = substr ($arr[$i-1], -1);
1132                                                 $x2 = substr ($arr[$i-1], -2, 1);
1133                                                 if ($x1 == ' ') {
1134                                                         if ($firstspace == -1) $firstspace = $i;
1135                                                 } else if ($x2 == ' ') {
1136                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1137                                                 } else {
1138                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1139                                                 }
1140                                         }
1141                                         $i++;
1142                                 }
1143
1144                                 # If there is a single-letter word, use it!
1145                                 if ($firstsingleletterword > -1)
1146                                 {
1147                                         $arr [ $firstsingleletterword ] = "''";
1148                                         $arr [ $firstsingleletterword-1 ] .= "'";
1149                                 }
1150                                 # If not, but there's a multi-letter word, use that one.
1151                                 else if ($firstmultiletterword > -1)
1152                                 {
1153                                         $arr [ $firstmultiletterword ] = "''";
1154                                         $arr [ $firstmultiletterword-1 ] .= "'";
1155                                 }
1156                                 # ... otherwise use the first one that has neither.
1157                                 # (notice that it is possible for all three to be -1 if, for example,
1158                                 # there is only one pentuple-apostrophe in the line)
1159                                 else if ($firstspace > -1)
1160                                 {
1161                                         $arr [ $firstspace ] = "''";
1162                                         $arr [ $firstspace-1 ] .= "'";
1163                                 }
1164                         }
1165
1166                         # Now let's actually convert our apostrophic mush to HTML!
1167                         $output = '';
1168                         $buffer = '';
1169                         $state = '';
1170                         $i = 0;
1171                         foreach ($arr as $r)
1172                         {
1173                                 if (($i % 2) == 0)
1174                                 {
1175                                         if ($state == 'both')
1176                                                 $buffer .= $r;
1177                                         else
1178                                                 $output .= $r;
1179                                 }
1180                                 else
1181                                 {
1182                                         if (strlen ($r) == 2)
1183                                         {
1184                                                 if ($state == 'i')
1185                                                 { $output .= '</i>'; $state = ''; }
1186                                                 else if ($state == 'bi')
1187                                                 { $output .= '</i>'; $state = 'b'; }
1188                                                 else if ($state == 'ib')
1189                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1190                                                 else if ($state == 'both')
1191                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1192                                                 else # $state can be 'b' or ''
1193                                                 { $output .= '<i>'; $state .= 'i'; }
1194                                         }
1195                                         else if (strlen ($r) == 3)
1196                                         {
1197                                                 if ($state == 'b')
1198                                                 { $output .= '</b>'; $state = ''; }
1199                                                 else if ($state == 'bi')
1200                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1201                                                 else if ($state == 'ib')
1202                                                 { $output .= '</b>'; $state = 'i'; }
1203                                                 else if ($state == 'both')
1204                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1205                                                 else # $state can be 'i' or ''
1206                                                 { $output .= '<b>'; $state .= 'b'; }
1207                                         }
1208                                         else if (strlen ($r) == 5)
1209                                         {
1210                                                 if ($state == 'b')
1211                                                 { $output .= '</b><i>'; $state = 'i'; }
1212                                                 else if ($state == 'i')
1213                                                 { $output .= '</i><b>'; $state = 'b'; }
1214                                                 else if ($state == 'bi')
1215                                                 { $output .= '</i></b>'; $state = ''; }
1216                                                 else if ($state == 'ib')
1217                                                 { $output .= '</b></i>'; $state = ''; }
1218                                                 else if ($state == 'both')
1219                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1220                                                 else # ($state == '')
1221                                                 { $buffer = ''; $state = 'both'; }
1222                                         }
1223                                 }
1224                                 $i++;
1225                         }
1226                         # Now close all remaining tags.  Notice that the order is important.
1227                         if ($state == 'b' || $state == 'ib')
1228                                 $output .= '</b>';
1229                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
1230                                 $output .= '</i>';
1231                         if ($state == 'bi')
1232                                 $output .= '</b>';
1233                         # There might be lonely ''''', so make sure we have a buffer
1234                         if ($state == 'both' && $buffer)
1235                                 $output .= '<b><i>'.$buffer.'</i></b>';
1236                         return $output;
1237                 }
1238         }
1239
1240         /**
1241          * Replace external links
1242          *
1243          * Note: this is all very hackish and the order of execution matters a lot.
1244          * Make sure to run maintenance/parserTests.php if you change this code.
1245          *
1246          * @private
1247          */
1248         function replaceExternalLinks( $text ) {
1249                 global $wgContLang;
1250                 $fname = 'Parser::replaceExternalLinks';
1251                 wfProfileIn( $fname );
1252
1253                 $sk = $this->mOptions->getSkin();
1254
1255                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1256
1257                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
1258
1259                 $i = 0;
1260                 while ( $i<count( $bits ) ) {
1261                         $url = $bits[$i++];
1262                         $protocol = $bits[$i++];
1263                         $text = $bits[$i++];
1264                         $trail = $bits[$i++];
1265
1266                         # The characters '<' and '>' (which were escaped by
1267                         # removeHTMLtags()) should not be included in
1268                         # URLs, per RFC 2396.
1269                         $m2 = array();
1270                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1271                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1272                                 $url = substr($url, 0, $m2[0][1]);
1273                         }
1274
1275                         # If the link text is an image URL, replace it with an <img> tag
1276                         # This happened by accident in the original parser, but some people used it extensively
1277                         $img = $this->maybeMakeExternalImage( $text );
1278                         if ( $img !== false ) {
1279                                 $text = $img;
1280                         }
1281
1282                         $dtrail = '';
1283
1284                         # Set linktype for CSS - if URL==text, link is essentially free
1285                         $linktype = ($text == $url) ? 'free' : 'text';
1286
1287                         # No link text, e.g. [http://domain.tld/some.link]
1288                         if ( $text == '' ) {
1289                                 # Autonumber if allowed. See bug #5918
1290                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1291                                         $text = '[' . ++$this->mAutonumber . ']';
1292                                         $linktype = 'autonumber';
1293                                 } else {
1294                                         # Otherwise just use the URL
1295                                         $text = htmlspecialchars( $url );
1296                                         $linktype = 'free';
1297                                 }
1298                         } else {
1299                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1300                                 # Check for trail
1301                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1302                         }
1303
1304                         $text = $wgContLang->markNoConversion($text);
1305
1306                         $url = Sanitizer::cleanUrl( $url );
1307
1308                         # Process the trail (i.e. everything after this link up until start of the next link),
1309                         # replacing any non-bracketed links
1310                         $trail = $this->replaceFreeExternalLinks( $trail );
1311
1312                         # Use the encoded URL
1313                         # This means that users can paste URLs directly into the text
1314                         # Funny characters like &ouml; aren't valid in URLs anyway
1315                         # This was changed in August 2004
1316                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype, $this->mTitle->getNamespace() ) . $dtrail . $trail;
1317
1318                         # Register link in the output object.
1319                         # Replace unnecessary URL escape codes with the referenced character
1320                         # This prevents spammers from hiding links from the filters
1321                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1322                         $this->mOutput->addExternalLink( $pasteurized );
1323                 }
1324
1325                 wfProfileOut( $fname );
1326                 return $s;
1327         }
1328
1329         /**
1330          * Replace anything that looks like a URL with a link
1331          * @private
1332          */
1333         function replaceFreeExternalLinks( $text ) {
1334                 global $wgContLang;
1335                 $fname = 'Parser::replaceFreeExternalLinks';
1336                 wfProfileIn( $fname );
1337
1338                 $bits = preg_split( '/(\b(?:' . wfUrlProtocols() . '))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1339                 $s = array_shift( $bits );
1340                 $i = 0;
1341
1342                 $sk = $this->mOptions->getSkin();
1343
1344                 while ( $i < count( $bits ) ){
1345                         $protocol = $bits[$i++];
1346                         $remainder = $bits[$i++];
1347
1348                         $m = array();
1349                         if ( preg_match( '/^('.self::EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1350                                 # Found some characters after the protocol that look promising
1351                                 $url = $protocol . $m[1];
1352                                 $trail = $m[2];
1353
1354                                 # special case: handle urls as url args:
1355                                 # http://www.example.com/foo?=http://www.example.com/bar
1356                                 if(strlen($trail) == 0 &&
1357                                         isset($bits[$i]) &&
1358                                         preg_match('/^'. wfUrlProtocols() . '$/S', $bits[$i]) &&
1359                                         preg_match( '/^('.self::EXT_LINK_URL_CLASS.'+)(.*)$/s', $bits[$i + 1], $m ))
1360                                 {
1361                                         # add protocol, arg
1362                                         $url .= $bits[$i] . $m[1]; # protocol, url as arg to previous link
1363                                         $i += 2;
1364                                         $trail = $m[2];
1365                                 }
1366
1367                                 # The characters '<' and '>' (which were escaped by
1368                                 # removeHTMLtags()) should not be included in
1369                                 # URLs, per RFC 2396.
1370                                 $m2 = array();
1371                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1372                                         $trail = substr($url, $m2[0][1]) . $trail;
1373                                         $url = substr($url, 0, $m2[0][1]);
1374                                 }
1375
1376                                 # Move trailing punctuation to $trail
1377                                 $sep = ',;\.:!?';
1378                                 # If there is no left bracket, then consider right brackets fair game too
1379                                 if ( strpos( $url, '(' ) === false ) {
1380                                         $sep .= ')';
1381                                 }
1382
1383                                 $numSepChars = strspn( strrev( $url ), $sep );
1384                                 if ( $numSepChars ) {
1385                                         $trail = substr( $url, -$numSepChars ) . $trail;
1386                                         $url = substr( $url, 0, -$numSepChars );
1387                                 }
1388
1389                                 $url = Sanitizer::cleanUrl( $url );
1390
1391                                 # Is this an external image?
1392                                 $text = $this->maybeMakeExternalImage( $url );
1393                                 if ( $text === false ) {
1394                                         # Not an image, make a link
1395                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free', $this->mTitle->getNamespace() );
1396                                         # Register it in the output object...
1397                                         # Replace unnecessary URL escape codes with their equivalent characters
1398                                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1399                                         $this->mOutput->addExternalLink( $pasteurized );
1400                                 }
1401                                 $s .= $text . $trail;
1402                         } else {
1403                                 $s .= $protocol . $remainder;
1404                         }
1405                 }
1406                 wfProfileOut( $fname );
1407                 return $s;
1408         }
1409
1410         /**
1411          * Replace unusual URL escape codes with their equivalent characters
1412          * @param string
1413          * @return string
1414          * @static
1415          * @todo  This can merge genuinely required bits in the path or query string,
1416          *        breaking legit URLs. A proper fix would treat the various parts of
1417          *        the URL differently; as a workaround, just use the output for
1418          *        statistical records, not for actual linking/output.
1419          */
1420         static function replaceUnusualEscapes( $url ) {
1421                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1422                         array( 'Parser', 'replaceUnusualEscapesCallback' ), $url );
1423         }
1424
1425         /**
1426          * Callback function used in replaceUnusualEscapes().
1427          * Replaces unusual URL escape codes with their equivalent character
1428          * @static
1429          * @private
1430          */
1431         private static function replaceUnusualEscapesCallback( $matches ) {
1432                 $char = urldecode( $matches[0] );
1433                 $ord = ord( $char );
1434                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1435                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1436                         // No, shouldn't be escaped
1437                         return $char;
1438                 } else {
1439                         // Yes, leave it escaped
1440                         return $matches[0];
1441                 }
1442         }
1443
1444         /**
1445          * make an image if it's allowed, either through the global
1446          * option or through the exception
1447          * @private
1448          */
1449         function maybeMakeExternalImage( $url ) {
1450                 $sk = $this->mOptions->getSkin();
1451                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1452                 $imagesexception = !empty($imagesfrom);
1453                 $text = false;
1454                 if ( $this->mOptions->getAllowExternalImages()
1455                      || ( $imagesexception && strpos( $url, $imagesfrom ) === 0 ) ) {
1456                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1457                                 # Image found
1458                                 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
1459                         }
1460                 }
1461                 return $text;
1462         }
1463
1464         /**
1465          * Process [[ ]] wikilinks
1466          *
1467          * @private
1468          */
1469         function replaceInternalLinks( $s ) {
1470                 global $wgContLang;
1471                 static $fname = 'Parser::replaceInternalLinks' ;
1472
1473                 wfProfileIn( $fname );
1474
1475                 wfProfileIn( $fname.'-setup' );
1476                 static $tc = FALSE;
1477                 # the % is needed to support urlencoded titles as well
1478                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1479
1480                 $sk = $this->mOptions->getSkin();
1481
1482                 #split the entire text string on occurences of [[
1483                 $a = explode( '[[', ' ' . $s );
1484                 #get the first element (all text up to first [[), and remove the space we added
1485                 $s = array_shift( $a );
1486                 $s = substr( $s, 1 );
1487
1488                 # Match a link having the form [[namespace:link|alternate]]trail
1489                 static $e1 = FALSE;
1490                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1491                 # Match cases where there is no "]]", which might still be images
1492                 static $e1_img = FALSE;
1493                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1494
1495                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1496                 $e2 = null;
1497                 if ( $useLinkPrefixExtension ) {
1498                         # Match the end of a line for a word that's not followed by whitespace,
1499                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1500                         $e2 = wfMsgForContent( 'linkprefix' );
1501                 }
1502
1503                 if( is_null( $this->mTitle ) ) {
1504                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1505                 }
1506                 $nottalk = !$this->mTitle->isTalkPage();
1507
1508                 if ( $useLinkPrefixExtension ) {
1509                         $m = array();
1510                         if ( preg_match( $e2, $s, $m ) ) {
1511                                 $first_prefix = $m[2];
1512                         } else {
1513                                 $first_prefix = false;
1514                         }
1515                 } else {
1516                         $prefix = '';
1517                 }
1518
1519                 if($wgContLang->hasVariants()) {
1520                         $selflink = $wgContLang->convertLinkToAllVariants($this->mTitle->getPrefixedText());
1521                 } else {
1522                         $selflink = array($this->mTitle->getPrefixedText());
1523                 }
1524                 $useSubpages = $this->areSubpagesAllowed();
1525                 wfProfileOut( $fname.'-setup' );
1526
1527                 # Loop for each link
1528                 for ($k = 0; isset( $a[$k] ); $k++) {
1529                         $line = $a[$k];
1530                         if ( $useLinkPrefixExtension ) {
1531                                 wfProfileIn( $fname.'-prefixhandling' );
1532                                 if ( preg_match( $e2, $s, $m ) ) {
1533                                         $prefix = $m[2];
1534                                         $s = $m[1];
1535                                 } else {
1536                                         $prefix='';
1537                                 }
1538                                 # first link
1539                                 if($first_prefix) {
1540                                         $prefix = $first_prefix;
1541                                         $first_prefix = false;
1542                                 }
1543                                 wfProfileOut( $fname.'-prefixhandling' );
1544                         }
1545
1546                         $might_be_img = false;
1547
1548                         wfProfileIn( "$fname-e1" );
1549                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1550                                 $text = $m[2];
1551                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1552                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1553                                 # the real problem is with the $e1 regex
1554                                 # See bug 1300.
1555                                 #
1556                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1557                                 # and no image is in sight. See bug 2095.
1558                                 #
1559                                 if( $text !== '' &&
1560                                         substr( $m[3], 0, 1 ) === ']' &&
1561                                         strpos($text, '[') !== false
1562                                 )
1563                                 {
1564                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1565                                         $m[3] = substr( $m[3], 1 );
1566                                 }
1567                                 # fix up urlencoded title texts
1568                                 if( strpos( $m[1], '%' ) !== false ) {
1569                                         # Should anchors '#' also be rejected?
1570                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1571                                 }
1572                                 $trail = $m[3];
1573                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1574                                 $might_be_img = true;
1575                                 $text = $m[2];
1576                                 if ( strpos( $m[1], '%' ) !== false ) {
1577                                         $m[1] = urldecode($m[1]);
1578                                 }
1579                                 $trail = "";
1580                         } else { # Invalid form; output directly
1581                                 $s .= $prefix . '[[' . $line ;
1582                                 wfProfileOut( "$fname-e1" );
1583                                 continue;
1584                         }
1585                         wfProfileOut( "$fname-e1" );
1586                         wfProfileIn( "$fname-misc" );
1587
1588                         # Don't allow internal links to pages containing
1589                         # PROTO: where PROTO is a valid URL protocol; these
1590                         # should be external links.
1591                         if (preg_match('/^\b(?:' . wfUrlProtocols() . ')/', $m[1])) {
1592                                 $s .= $prefix . '[[' . $line ;
1593                                 continue;
1594                         }
1595
1596                         # Make subpage if necessary
1597                         if( $useSubpages ) {
1598                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1599                         } else {
1600                                 $link = $m[1];
1601                         }
1602
1603                         $noforce = (substr($m[1], 0, 1) != ':');
1604                         if (!$noforce) {
1605                                 # Strip off leading ':'
1606                                 $link = substr($link, 1);
1607                         }
1608
1609                         wfProfileOut( "$fname-misc" );
1610                         wfProfileIn( "$fname-title" );
1611                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki($link) );
1612                         if( !$nt ) {
1613                                 $s .= $prefix . '[[' . $line;
1614                                 wfProfileOut( "$fname-title" );
1615                                 continue;
1616                         }
1617
1618                         $ns = $nt->getNamespace();
1619                         $iw = $nt->getInterWiki();
1620                         wfProfileOut( "$fname-title" );
1621
1622                         if ($might_be_img) { # if this is actually an invalid link
1623                                 wfProfileIn( "$fname-might_be_img" );
1624                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1625                                         $found = false;
1626                                         while (isset ($a[$k+1]) ) {
1627                                                 #look at the next 'line' to see if we can close it there
1628                                                 $spliced = array_splice( $a, $k + 1, 1 );
1629                                                 $next_line = array_shift( $spliced );
1630                                                 $m = explode( ']]', $next_line, 3 );
1631                                                 if ( count( $m ) == 3 ) {
1632                                                         # the first ]] closes the inner link, the second the image
1633                                                         $found = true;
1634                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1635                                                         $trail = $m[2];
1636                                                         break;
1637                                                 } elseif ( count( $m ) == 2 ) {
1638                                                         #if there's exactly one ]] that's fine, we'll keep looking
1639                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1640                                                 } else {
1641                                                         #if $next_line is invalid too, we need look no further
1642                                                         $text .= '[[' . $next_line;
1643                                                         break;
1644                                                 }
1645                                         }
1646                                         if ( !$found ) {
1647                                                 # we couldn't find the end of this imageLink, so output it raw
1648                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1649                                                 $text = $this->replaceInternalLinks($text);
1650                                                 $s .= "{$prefix}[[$link|$text";
1651                                                 # note: no $trail, because without an end, there *is* no trail
1652                                                 wfProfileOut( "$fname-might_be_img" );
1653                                                 continue;
1654                                         }
1655                                 } else { #it's not an image, so output it raw
1656                                         $s .= "{$prefix}[[$link|$text";
1657                                         # note: no $trail, because without an end, there *is* no trail
1658                                         wfProfileOut( "$fname-might_be_img" );
1659                                         continue;
1660                                 }
1661                                 wfProfileOut( "$fname-might_be_img" );
1662                         }
1663
1664                         $wasblank = ( '' == $text );
1665                         if( $wasblank ) $text = $link;
1666
1667                         # Link not escaped by : , create the various objects
1668                         if( $noforce ) {
1669
1670                                 # Interwikis
1671                                 wfProfileIn( "$fname-interwiki" );
1672                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1673                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1674                                         $s = rtrim($s . $prefix);
1675                                         $s .= trim($trail, "\n") == '' ? '': $prefix . $trail;
1676                                         wfProfileOut( "$fname-interwiki" );
1677                                         continue;
1678                                 }
1679                                 wfProfileOut( "$fname-interwiki" );
1680
1681                                 if ( $ns == NS_IMAGE ) {
1682                                         wfProfileIn( "$fname-image" );
1683                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1684                                                 # recursively parse links inside the image caption
1685                                                 # actually, this will parse them in any other parameters, too,
1686                                                 # but it might be hard to fix that, and it doesn't matter ATM
1687                                                 $text = $this->replaceExternalLinks($text);
1688                                                 $text = $this->replaceInternalLinks($text);
1689
1690                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1691                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text ) ) . $trail;
1692                                                 $this->mOutput->addImage( $nt->getDBkey() );
1693
1694                                                 wfProfileOut( "$fname-image" );
1695                                                 continue;
1696                                         } else {
1697                                                 # We still need to record the image's presence on the page
1698                                                 $this->mOutput->addImage( $nt->getDBkey() );
1699                                         }
1700                                         wfProfileOut( "$fname-image" );
1701
1702                                 }
1703
1704                                 if ( $ns == NS_CATEGORY ) {
1705                                         wfProfileIn( "$fname-category" );
1706                                         $s = rtrim($s . "\n"); # bug 87
1707
1708                                         if ( $wasblank ) {
1709                                                 $sortkey = $this->getDefaultSort();
1710                                         } else {
1711                                                 $sortkey = $text;
1712                                         }
1713                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1714                                         $sortkey = str_replace( "\n", '', $sortkey );
1715                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1716                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1717
1718                                         /**
1719                                          * Strip the whitespace Category links produce, see bug 87
1720                                          * @todo We might want to use trim($tmp, "\n") here.
1721                                          */
1722                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1723
1724                                         wfProfileOut( "$fname-category" );
1725                                         continue;
1726                                 }
1727                         }
1728
1729                         # Self-link checking
1730                         if( $nt->getFragment() === '' ) {
1731                                 if( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1732                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1733                                         continue;
1734                                 }
1735                         }
1736
1737                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1738                         if( $ns == NS_MEDIA ) {
1739                                 $link = $sk->makeMediaLinkObj( $nt, $text );
1740                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1741                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1742                                 $this->mOutput->addImage( $nt->getDBkey() );
1743                                 continue;
1744                         } elseif( $ns == NS_SPECIAL ) {
1745                                 if( SpecialPage::exists( $nt->getDBkey() ) ) {
1746                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1747                                 } else {
1748                                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1749                                 }
1750                                 continue;
1751                         } elseif( $ns == NS_IMAGE ) {
1752                                 $img = wfFindFile( $nt );
1753                                 if( $img ) {
1754                                         // Force a blue link if the file exists; may be a remote
1755                                         // upload on the shared repository, and we want to see its
1756                                         // auto-generated page.
1757                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1758                                         $this->mOutput->addLink( $nt );
1759                                         continue;
1760                                 }
1761                         }
1762                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1763                 }
1764                 wfProfileOut( $fname );
1765                 return $s;
1766         }
1767
1768         /**
1769          * Make a link placeholder. The text returned can be later resolved to a real link with
1770          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1771          * parsing of interwiki links, and secondly to allow all existence checks and
1772          * article length checks (for stub links) to be bundled into a single query.
1773          *
1774          */
1775         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1776                 wfProfileIn( __METHOD__ );
1777                 if ( ! is_object($nt) ) {
1778                         # Fail gracefully
1779                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1780                 } else {
1781                         # Separate the link trail from the rest of the link
1782                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1783
1784                         if ( $nt->isExternal() ) {
1785                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1786                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1787                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1788                         } else {
1789                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1790                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1791                                 $this->mLinkHolders['queries'][] = $query;
1792                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1793                                 $this->mLinkHolders['titles'][] = $nt;
1794
1795                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1796                         }
1797                 }
1798                 wfProfileOut( __METHOD__ );
1799                 return $retVal;
1800         }
1801
1802         /**
1803          * Render a forced-blue link inline; protect against double expansion of
1804          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1805          * Since this little disaster has to split off the trail text to avoid
1806          * breaking URLs in the following text without breaking trails on the
1807          * wiki links, it's been made into a horrible function.
1808          *
1809          * @param Title $nt
1810          * @param string $text
1811          * @param string $query
1812          * @param string $trail
1813          * @param string $prefix
1814          * @return string HTML-wikitext mix oh yuck
1815          */
1816         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1817                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1818                 $sk = $this->mOptions->getSkin();
1819                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1820                 return $this->armorLinks( $link ) . $trail;
1821         }
1822
1823         /**
1824          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1825          * going to go through further parsing steps before inline URL expansion.
1826          *
1827          * In particular this is important when using action=render, which causes
1828          * full URLs to be included.
1829          *
1830          * Oh man I hate our multi-layer parser!
1831          *
1832          * @param string more-or-less HTML
1833          * @return string less-or-more HTML with NOPARSE bits
1834          */
1835         function armorLinks( $text ) {
1836                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1837                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1838         }
1839
1840         /**
1841          * Return true if subpage links should be expanded on this page.
1842          * @return bool
1843          */
1844         function areSubpagesAllowed() {
1845                 # Some namespaces don't allow subpages
1846                 global $wgNamespacesWithSubpages;
1847                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1848         }
1849
1850         /**
1851          * Handle link to subpage if necessary
1852          * @param string $target the source of the link
1853          * @param string &$text the link text, modified as necessary
1854          * @return string the full name of the link
1855          * @private
1856          */
1857         function maybeDoSubpageLink($target, &$text) {
1858                 # Valid link forms:
1859                 # Foobar -- normal
1860                 # :Foobar -- override special treatment of prefix (images, language links)
1861                 # /Foobar -- convert to CurrentPage/Foobar
1862                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1863                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1864                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1865
1866                 $fname = 'Parser::maybeDoSubpageLink';
1867                 wfProfileIn( $fname );
1868                 $ret = $target; # default return value is no change
1869
1870                 # Some namespaces don't allow subpages,
1871                 # so only perform processing if subpages are allowed
1872                 if( $this->areSubpagesAllowed() ) {
1873                         $hash = strpos( $target, '#' );
1874                         if( $hash !== false ) {
1875                                 $suffix = substr( $target, $hash );
1876                                 $target = substr( $target, 0, $hash );
1877                         } else {
1878                                 $suffix = '';
1879                         }
1880                         # bug 7425
1881                         $target = trim( $target );
1882                         # Look at the first character
1883                         if( $target != '' && $target{0} == '/' ) {
1884                                 # / at end means we don't want the slash to be shown
1885                                 $m = array();
1886                                 $trailingSlashes = preg_match_all( '%(/+)$%', $target, $m );
1887                                 if( $trailingSlashes ) {
1888                                         $noslash = $target = substr( $target, 1, -strlen($m[0][0]) );
1889                                 } else {
1890                                         $noslash = substr( $target, 1 );
1891                                 }
1892
1893                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash) . $suffix;
1894                                 if( '' === $text ) {
1895                                         $text = $target . $suffix;
1896                                 } # this might be changed for ugliness reasons
1897                         } else {
1898                                 # check for .. subpage backlinks
1899                                 $dotdotcount = 0;
1900                                 $nodotdot = $target;
1901                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1902                                         ++$dotdotcount;
1903                                         $nodotdot = substr( $nodotdot, 3 );
1904                                 }
1905                                 if($dotdotcount > 0) {
1906                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1907                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1908                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1909                                                 # / at the end means don't show full path
1910                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1911                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1912                                                         if( '' === $text ) {
1913                                                                 $text = $nodotdot . $suffix;
1914                                                         }
1915                                                 }
1916                                                 $nodotdot = trim( $nodotdot );
1917                                                 if( $nodotdot != '' ) {
1918                                                         $ret .= '/' . $nodotdot;
1919                                                 }
1920                                                 $ret .= $suffix;
1921                                         }
1922                                 }
1923                         }
1924                 }
1925
1926                 wfProfileOut( $fname );
1927                 return $ret;
1928         }
1929
1930         /**#@+
1931          * Used by doBlockLevels()
1932          * @private
1933          */
1934         /* private */ function closeParagraph() {
1935                 $result = '';
1936                 if ( '' != $this->mLastSection ) {
1937                         $result = '</' . $this->mLastSection  . ">\n";
1938                 }
1939                 $this->mInPre = false;
1940                 $this->mLastSection = '';
1941                 return $result;
1942         }
1943         # getCommon() returns the length of the longest common substring
1944         # of both arguments, starting at the beginning of both.
1945         #
1946         /* private */ function getCommon( $st1, $st2 ) {
1947                 $fl = strlen( $st1 );
1948                 $shorter = strlen( $st2 );
1949                 if ( $fl < $shorter ) { $shorter = $fl; }
1950
1951                 for ( $i = 0; $i < $shorter; ++$i ) {
1952                         if ( $st1{$i} != $st2{$i} ) { break; }
1953                 }
1954                 return $i;
1955         }
1956         # These next three functions open, continue, and close the list
1957         # element appropriate to the prefix character passed into them.
1958         #
1959         /* private */ function openList( $char ) {
1960                 $result = $this->closeParagraph();
1961
1962                 if ( '*' == $char ) { $result .= '<ul><li>'; }
1963                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
1964                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
1965                 else if ( ';' == $char ) {
1966                         $result .= '<dl><dt>';
1967                         $this->mDTopen = true;
1968                 }
1969                 else { $result = '<!-- ERR 1 -->'; }
1970
1971                 return $result;
1972         }
1973
1974         /* private */ function nextItem( $char ) {
1975                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
1976                 else if ( ':' == $char || ';' == $char ) {
1977                         $close = '</dd>';
1978                         if ( $this->mDTopen ) { $close = '</dt>'; }
1979                         if ( ';' == $char ) {
1980                                 $this->mDTopen = true;
1981                                 return $close . '<dt>';
1982                         } else {
1983                                 $this->mDTopen = false;
1984                                 return $close . '<dd>';
1985                         }
1986                 }
1987                 return '<!-- ERR 2 -->';
1988         }
1989
1990         /* private */ function closeList( $char ) {
1991                 if ( '*' == $char ) { $text = '</li></ul>'; }
1992                 else if ( '#' == $char ) { $text = '</li></ol>'; }
1993                 else if ( ':' == $char ) {
1994                         if ( $this->mDTopen ) {
1995                                 $this->mDTopen = false;
1996                                 $text = '</dt></dl>';
1997                         } else {
1998                                 $text = '</dd></dl>';
1999                         }
2000                 }
2001                 else {  return '<!-- ERR 3 -->'; }
2002                 return $text."\n";
2003         }
2004         /**#@-*/
2005
2006         /**
2007          * Make lists from lines starting with ':', '*', '#', etc.
2008          *
2009          * @private
2010          * @return string the lists rendered as HTML
2011          */
2012         function doBlockLevels( $text, $linestart ) {
2013                 $fname = 'Parser::doBlockLevels';
2014                 wfProfileIn( $fname );
2015
2016                 # Parsing through the text line by line.  The main thing
2017                 # happening here is handling of block-level elements p, pre,
2018                 # and making lists from lines starting with * # : etc.
2019                 #
2020                 $textLines = explode( "\n", $text );
2021
2022                 $lastPrefix = $output = '';
2023                 $this->mDTopen = $inBlockElem = false;
2024                 $prefixLength = 0;
2025                 $paragraphStack = false;
2026
2027                 if ( !$linestart ) {
2028                         $output .= array_shift( $textLines );
2029                 }
2030                 foreach ( $textLines as $oLine ) {
2031                         $lastPrefixLength = strlen( $lastPrefix );
2032                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
2033                         $preOpenMatch = preg_match('/<pre/i', $oLine );
2034                         if ( !$this->mInPre ) {
2035                                 # Multiple prefixes may abut each other for nested lists.
2036                                 $prefixLength = strspn( $oLine, '*#:;' );
2037                                 $pref = substr( $oLine, 0, $prefixLength );
2038
2039                                 # eh?
2040                                 $pref2 = str_replace( ';', ':', $pref );
2041                                 $t = substr( $oLine, $prefixLength );
2042                                 $this->mInPre = !empty($preOpenMatch);
2043                         } else {
2044                                 # Don't interpret any other prefixes in preformatted text
2045                                 $prefixLength = 0;
2046                                 $pref = $pref2 = '';
2047                                 $t = $oLine;
2048                         }
2049
2050                         # List generation
2051                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
2052                                 # Same as the last item, so no need to deal with nesting or opening stuff
2053                                 $output .= $this->nextItem( substr( $pref, -1 ) );
2054                                 $paragraphStack = false;
2055
2056                                 if ( substr( $pref, -1 ) == ';') {
2057                                         # The one nasty exception: definition lists work like this:
2058                                         # ; title : definition text
2059                                         # So we check for : in the remainder text to split up the
2060                                         # title and definition, without b0rking links.
2061                                         $term = $t2 = '';
2062                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2063                                                 $t = $t2;
2064                                                 $output .= $term . $this->nextItem( ':' );
2065                                         }
2066                                 }
2067                         } elseif( $prefixLength || $lastPrefixLength ) {
2068                                 # Either open or close a level...
2069                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
2070                                 $paragraphStack = false;
2071
2072                                 while( $commonPrefixLength < $lastPrefixLength ) {
2073                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
2074                                         --$lastPrefixLength;
2075                                 }
2076                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2077                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
2078                                 }
2079                                 while ( $prefixLength > $commonPrefixLength ) {
2080                                         $char = substr( $pref, $commonPrefixLength, 1 );
2081                                         $output .= $this->openList( $char );
2082
2083                                         if ( ';' == $char ) {
2084                                                 # FIXME: This is dupe of code above
2085                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2086                                                         $t = $t2;
2087                                                         $output .= $term . $this->nextItem( ':' );
2088                                                 }
2089                                         }
2090                                         ++$commonPrefixLength;
2091                                 }
2092                                 $lastPrefix = $pref2;
2093                         }
2094                         if( 0 == $prefixLength ) {
2095                                 wfProfileIn( "$fname-paragraph" );
2096                                 # No prefix (not in list)--go to paragraph mode
2097                                 // XXX: use a stack for nestable elements like span, table and div
2098                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2099                                 $closematch = preg_match(
2100                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2101                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2102                                 if ( $openmatch or $closematch ) {
2103                                         $paragraphStack = false;
2104                                         # TODO bug 5718: paragraph closed
2105                                         $output .= $this->closeParagraph();
2106                                         if ( $preOpenMatch and !$preCloseMatch ) {
2107                                                 $this->mInPre = true;
2108                                         }
2109                                         if ( $closematch ) {
2110                                                 $inBlockElem = false;
2111                                         } else {
2112                                                 $inBlockElem = true;
2113                                         }
2114                                 } else if ( !$inBlockElem && !$this->mInPre ) {
2115                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
2116                                                 // pre
2117                                                 if ($this->mLastSection != 'pre') {
2118                                                         $paragraphStack = false;
2119                                                         $output .= $this->closeParagraph().'<pre>';
2120                                                         $this->mLastSection = 'pre';
2121                                                 }
2122                                                 $t = substr( $t, 1 );
2123                                         } else {
2124                                                 // paragraph
2125                                                 if ( '' == trim($t) ) {
2126                                                         if ( $paragraphStack ) {
2127                                                                 $output .= $paragraphStack.'<br />';
2128                                                                 $paragraphStack = false;
2129                                                                 $this->mLastSection = 'p';
2130                                                         } else {
2131                                                                 if ($this->mLastSection != 'p' ) {
2132                                                                         $output .= $this->closeParagraph();
2133                                                                         $this->mLastSection = '';
2134                                                                         $paragraphStack = '<p>';
2135                                                                 } else {
2136                                                                         $paragraphStack = '</p><p>';
2137                                                                 }
2138                                                         }
2139                                                 } else {
2140                                                         if ( $paragraphStack ) {
2141                                                                 $output .= $paragraphStack;
2142                                                                 $paragraphStack = false;
2143                                                                 $this->mLastSection = 'p';
2144                                                         } else if ($this->mLastSection != 'p') {
2145                                                                 $output .= $this->closeParagraph().'<p>';
2146                                                                 $this->mLastSection = 'p';
2147                                                         }
2148                                                 }
2149                                         }
2150                                 }
2151                                 wfProfileOut( "$fname-paragraph" );
2152                         }
2153                         // somewhere above we forget to get out of pre block (bug 785)
2154                         if($preCloseMatch && $this->mInPre) {
2155                                 $this->mInPre = false;
2156                         }
2157                         if ($paragraphStack === false) {
2158                                 $output .= $t."\n";
2159                         }
2160                 }
2161                 while ( $prefixLength ) {
2162                         $output .= $this->closeList( $pref2{$prefixLength-1} );
2163                         --$prefixLength;
2164                 }
2165                 if ( '' != $this->mLastSection ) {
2166                         $output .= '</' . $this->mLastSection . '>';
2167                         $this->mLastSection = '';
2168                 }
2169
2170                 wfProfileOut( $fname );
2171                 return $output;
2172         }
2173
2174         /**
2175          * Split up a string on ':', ignoring any occurences inside tags
2176          * to prevent illegal overlapping.
2177          * @param string $str the string to split
2178          * @param string &$before set to everything before the ':'
2179          * @param string &$after set to everything after the ':'
2180          * return string the position of the ':', or false if none found
2181          */
2182         function findColonNoLinks($str, &$before, &$after) {
2183                 $fname = 'Parser::findColonNoLinks';
2184                 wfProfileIn( $fname );
2185
2186                 $pos = strpos( $str, ':' );
2187                 if( $pos === false ) {
2188                         // Nothing to find!
2189                         wfProfileOut( $fname );
2190                         return false;
2191                 }
2192
2193                 $lt = strpos( $str, '<' );
2194                 if( $lt === false || $lt > $pos ) {
2195                         // Easy; no tag nesting to worry about
2196                         $before = substr( $str, 0, $pos );
2197                         $after = substr( $str, $pos+1 );
2198                         wfProfileOut( $fname );
2199                         return $pos;
2200                 }
2201
2202                 // Ugly state machine to walk through avoiding tags.
2203                 $state = self::COLON_STATE_TEXT;
2204                 $stack = 0;
2205                 $len = strlen( $str );
2206                 for( $i = 0; $i < $len; $i++ ) {
2207                         $c = $str{$i};
2208
2209                         switch( $state ) {
2210                         // (Using the number is a performance hack for common cases)
2211                         case 0: // self::COLON_STATE_TEXT:
2212                                 switch( $c ) {
2213                                 case "<":
2214                                         // Could be either a <start> tag or an </end> tag
2215                                         $state = self::COLON_STATE_TAGSTART;
2216                                         break;
2217                                 case ":":
2218                                         if( $stack == 0 ) {
2219                                                 // We found it!
2220                                                 $before = substr( $str, 0, $i );
2221                                                 $after = substr( $str, $i + 1 );
2222                                                 wfProfileOut( $fname );
2223                                                 return $i;
2224                                         }
2225                                         // Embedded in a tag; don't break it.
2226                                         break;
2227                                 default:
2228                                         // Skip ahead looking for something interesting
2229                                         $colon = strpos( $str, ':', $i );
2230                                         if( $colon === false ) {
2231                                                 // Nothing else interesting
2232                                                 wfProfileOut( $fname );
2233                                                 return false;
2234                                         }
2235                                         $lt = strpos( $str, '<', $i );
2236                                         if( $stack === 0 ) {
2237                                                 if( $lt === false || $colon < $lt ) {
2238                                                         // We found it!
2239                                                         $before = substr( $str, 0, $colon );
2240                                                         $after = substr( $str, $colon + 1 );
2241                                                         wfProfileOut( $fname );
2242                                                         return $i;
2243                                                 }
2244                                         }
2245                                         if( $lt === false ) {
2246                                                 // Nothing else interesting to find; abort!
2247                                                 // We're nested, but there's no close tags left. Abort!
2248                                                 break 2;
2249                                         }
2250                                         // Skip ahead to next tag start
2251                                         $i = $lt;
2252                                         $state = self::COLON_STATE_TAGSTART;
2253                                 }
2254                                 break;
2255                         case 1: // self::COLON_STATE_TAG:
2256                                 // In a <tag>
2257                                 switch( $c ) {
2258                                 case ">":
2259                                         $stack++;
2260                                         $state = self::COLON_STATE_TEXT;
2261                                         break;
2262                                 case "/":
2263                                         // Slash may be followed by >?
2264                                         $state = self::COLON_STATE_TAGSLASH;
2265                                         break;
2266                                 default:
2267                                         // ignore
2268                                 }
2269                                 break;
2270                         case 2: // self::COLON_STATE_TAGSTART:
2271                                 switch( $c ) {
2272                                 case "/":
2273                                         $state = self::COLON_STATE_CLOSETAG;
2274                                         break;
2275                                 case "!":
2276                                         $state = self::COLON_STATE_COMMENT;
2277                                         break;
2278                                 case ">":
2279                                         // Illegal early close? This shouldn't happen D:
2280                                         $state = self::COLON_STATE_TEXT;
2281                                         break;
2282                                 default:
2283                                         $state = self::COLON_STATE_TAG;
2284                                 }
2285                                 break;
2286                         case 3: // self::COLON_STATE_CLOSETAG:
2287                                 // In a </tag>
2288                                 if( $c == ">" ) {
2289                                         $stack--;
2290                                         if( $stack < 0 ) {
2291                                                 wfDebug( "Invalid input in $fname; too many close tags\n" );
2292                                                 wfProfileOut( $fname );
2293                                                 return false;
2294                                         }
2295                                         $state = self::COLON_STATE_TEXT;
2296                                 }
2297                                 break;
2298                         case self::COLON_STATE_TAGSLASH:
2299                                 if( $c == ">" ) {
2300                                         // Yes, a self-closed tag <blah/>
2301                                         $state = self::COLON_STATE_TEXT;
2302                                 } else {
2303                                         // Probably we're jumping the gun, and this is an attribute
2304                                         $state = self::COLON_STATE_TAG;
2305                                 }
2306                                 break;
2307                         case 5: // self::COLON_STATE_COMMENT:
2308                                 if( $c == "-" ) {
2309                                         $state = self::COLON_STATE_COMMENTDASH;
2310                                 }
2311                                 break;
2312                         case self::COLON_STATE_COMMENTDASH:
2313                                 if( $c == "-" ) {
2314                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2315                                 } else {
2316                                         $state = self::COLON_STATE_COMMENT;
2317                                 }
2318                                 break;
2319                         case self::COLON_STATE_COMMENTDASHDASH:
2320                                 if( $c == ">" ) {
2321                                         $state = self::COLON_STATE_TEXT;
2322                                 } else {
2323                                         $state = self::COLON_STATE_COMMENT;
2324                                 }
2325                                 break;
2326                         default:
2327                                 throw new MWException( "State machine error in $fname" );
2328                         }
2329                 }
2330                 if( $stack > 0 ) {
2331                         wfDebug( "Invalid input in $fname; not enough close tags (stack $stack, state $state)\n" );
2332                         return false;
2333                 }
2334                 wfProfileOut( $fname );
2335                 return false;
2336         }
2337
2338         /**
2339          * Return value of a magic variable (like PAGENAME)
2340          *
2341          * @private
2342          */
2343         function getVariableValue( $index ) {
2344                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2345
2346                 /**
2347                  * Some of these require message or data lookups and can be
2348                  * expensive to check many times.
2349                  */
2350                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2351                         if ( isset( $this->mVarCache[$index] ) ) {
2352                                 return $this->mVarCache[$index];
2353                         }
2354                 }
2355
2356                 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2357                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2358
2359                 # Use the time zone
2360                 global $wgLocaltimezone;
2361                 if ( isset( $wgLocaltimezone ) ) {
2362                         $oldtz = getenv( 'TZ' );
2363                         putenv( 'TZ='.$wgLocaltimezone );
2364                 }
2365
2366                 wfSuppressWarnings(); // E_STRICT system time bitching
2367                 $localTimestamp = date( 'YmdHis', $ts );
2368                 $localMonth = date( 'm', $ts );
2369                 $localMonthName = date( 'n', $ts );
2370                 $localDay = date( 'j', $ts );
2371                 $localDay2 = date( 'd', $ts );
2372                 $localDayOfWeek = date( 'w', $ts );
2373                 $localWeek = date( 'W', $ts );
2374                 $localYear = date( 'Y', $ts );
2375                 $localHour = date( 'H', $ts );
2376                 if ( isset( $wgLocaltimezone ) ) {
2377                         putenv( 'TZ='.$oldtz );
2378                 }
2379                 wfRestoreWarnings();
2380
2381                 switch ( $index ) {
2382                         case 'currentmonth':
2383                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2384                         case 'currentmonthname':
2385                                 return $this->mVarCache[$index] = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2386                         case 'currentmonthnamegen':
2387                                 return $this->mVarCache[$index] = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2388                         case 'currentmonthabbrev':
2389                                 return $this->mVarCache[$index] = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2390                         case 'currentday':
2391                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2392                         case 'currentday2':
2393                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2394                         case 'localmonth':
2395                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localMonth );
2396                         case 'localmonthname':
2397                                 return $this->mVarCache[$index] = $wgContLang->getMonthName( $localMonthName );
2398                         case 'localmonthnamegen':
2399                                 return $this->mVarCache[$index] = $wgContLang->getMonthNameGen( $localMonthName );
2400                         case 'localmonthabbrev':
2401                                 return $this->mVarCache[$index] = $wgContLang->getMonthAbbreviation( $localMonthName );
2402                         case 'localday':
2403                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localDay );
2404                         case 'localday2':
2405                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localDay2 );
2406                         case 'pagename':
2407                                 return wfEscapeWikiText( $this->mTitle->getText() );
2408                         case 'pagenamee':
2409                                 return $this->mTitle->getPartialURL();
2410                         case 'fullpagename':
2411                                 return wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2412                         case 'fullpagenamee':
2413                                 return $this->mTitle->getPrefixedURL();
2414                         case 'subpagename':
2415                                 return wfEscapeWikiText( $this->mTitle->getSubpageText() );
2416                         case 'subpagenamee':
2417                                 return $this->mTitle->getSubpageUrlForm();
2418                         case 'basepagename':
2419                                 return wfEscapeWikiText( $this->mTitle->getBaseText() );
2420                         case 'basepagenamee':
2421                                 return wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2422                         case 'talkpagename':
2423                                 if( $this->mTitle->canTalk() ) {
2424                                         $talkPage = $this->mTitle->getTalkPage();
2425                                         return wfEscapeWikiText( $talkPage->getPrefixedText() );
2426                                 } else {
2427                                         return '';
2428                                 }
2429                         case 'talkpagenamee':
2430                                 if( $this->mTitle->canTalk() ) {
2431                                         $talkPage = $this->mTitle->getTalkPage();
2432                                         return $talkPage->getPrefixedUrl();
2433                                 } else {
2434                                         return '';
2435                                 }
2436                         case 'subjectpagename':
2437                                 $subjPage = $this->mTitle->getSubjectPage();
2438                                 return wfEscapeWikiText( $subjPage->getPrefixedText() );
2439                         case 'subjectpagenamee':
2440                                 $subjPage = $this->mTitle->getSubjectPage();
2441                                 return $subjPage->getPrefixedUrl();
2442                         case 'revisionid':
2443                                 // Let the edit saving system know we should parse the page
2444                                 // *after* a revision ID has been assigned.
2445                                 $this->mOutput->setFlag( 'vary-revision' );
2446                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2447                                 return $this->mRevisionId;
2448                         case 'revisionday':
2449                                 // Let the edit saving system know we should parse the page
2450                                 // *after* a revision ID has been assigned. This is for null edits.
2451                                 $this->mOutput->setFlag( 'vary-revision' );
2452                                 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2453                                 return intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2454                         case 'revisionday2':
2455                                 // Let the edit saving system know we should parse the page
2456                                 // *after* a revision ID has been assigned. This is for null edits.
2457                                 $this->mOutput->setFlag( 'vary-revision' );
2458                                 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2459                                 return substr( $this->getRevisionTimestamp(), 6, 2 );
2460                         case 'revisionmonth':
2461                                 // Let the edit saving system know we should parse the page
2462                                 // *after* a revision ID has been assigned. This is for null edits.
2463                                 $this->mOutput->setFlag( 'vary-revision' );
2464                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2465                                 return intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2466                         case 'revisionyear':
2467                                 // Let the edit saving system know we should parse the page
2468                                 // *after* a revision ID has been assigned. This is for null edits.
2469                                 $this->mOutput->setFlag( 'vary-revision' );
2470                                 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2471                                 return substr( $this->getRevisionTimestamp(), 0, 4 );
2472                         case 'revisiontimestamp':
2473                                 // Let the edit saving system know we should parse the page
2474                                 // *after* a revision ID has been assigned. This is for null edits.
2475                                 $this->mOutput->setFlag( 'vary-revision' );
2476                                 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2477                                 return $this->getRevisionTimestamp();
2478                         case 'namespace':
2479                                 return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2480                         case 'namespacee':
2481                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2482                         case 'talkspace':
2483                                 return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2484                         case 'talkspacee':
2485                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2486                         case 'subjectspace':
2487                                 return $this->mTitle->getSubjectNsText();
2488                         case 'subjectspacee':
2489                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2490                         case 'currentdayname':
2491                                 return $this->mVarCache[$index] = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2492                         case 'currentyear':
2493                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2494                         case 'currenttime':
2495                                 return $this->mVarCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2496                         case 'currenthour':
2497                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2498                         case 'currentweek':
2499                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2500                                 // int to remove the padding
2501                                 return $this->mVarCache[$index] = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2502                         case 'currentdow':
2503                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2504                         case 'localdayname':
2505                                 return $this->mVarCache[$index] = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2506                         case 'localyear':
2507                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localYear, true );
2508                         case 'localtime':
2509                                 return $this->mVarCache[$index] = $wgContLang->time( $localTimestamp, false, false );
2510                         case 'localhour':
2511                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localHour, true );
2512                         case 'localweek':
2513                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2514                                 // int to remove the padding
2515                                 return $this->mVarCache[$index] = $wgContLang->formatNum( (int)$localWeek );
2516                         case 'localdow':
2517                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localDayOfWeek );
2518                         case 'numberofarticles':
2519                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::articles() );
2520                         case 'numberoffiles':
2521                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::images() );
2522                         case 'numberofusers':
2523                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::users() );
2524                         case 'numberofpages':
2525                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::pages() );
2526                         case 'numberofadmins':
2527                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::admins() );
2528                         case 'numberofedits':
2529                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::edits() );
2530                         case 'currenttimestamp':
2531                                 return $this->mVarCache[$index] = wfTimestamp( TS_MW, $ts );
2532                         case 'localtimestamp':
2533                                 return $this->mVarCache[$index] = $localTimestamp;
2534                         case 'currentversion':
2535                                 return $this->mVarCache[$index] = SpecialVersion::getVersion();
2536                         case 'sitename':
2537                                 return $wgSitename;
2538                         case 'server':
2539                                 return $wgServer;
2540                         case 'servername':
2541                                 return $wgServerName;
2542                         case 'scriptpath':
2543                                 return $wgScriptPath;
2544                         case 'directionmark':
2545                                 return $wgContLang->getDirMark();
2546                         case 'contentlanguage':
2547                                 global $wgContLanguageCode;
2548                                 return $wgContLanguageCode;
2549                         default:
2550                                 $ret = null;
2551                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret ) ) )
2552                                         return $ret;
2553                                 else
2554                                         return null;
2555                 }
2556         }
2557
2558         /**
2559          * initialise the magic variables (like CURRENTMONTHNAME)
2560          *
2561          * @private
2562          */
2563         function initialiseVariables() {
2564                 $fname = 'Parser::initialiseVariables';
2565                 wfProfileIn( $fname );
2566                 $variableIDs = MagicWord::getVariableIDs();
2567
2568                 $this->mVariables = new MagicWordArray( $variableIDs );
2569                 wfProfileOut( $fname );
2570         }
2571
2572         /**
2573          * Preprocess some wikitext and return the document tree.
2574          * This is the ghost of replace_variables().
2575          *
2576          * @param string $text The text to parse
2577          * @param integer flags Bitwise combination of:
2578          *          self::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
2579          *                                     included. Default is to assume a direct page view.
2580          *
2581          * The generated DOM tree must depend only on the input text and the flags.
2582          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2583          *
2584          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2585          * change in the DOM tree for a given text, must be passed through the section identifier
2586          * in the section edit link and thus back to extractSections().
2587          *
2588          * The output of this function is currently only cached in process memory, but a persistent
2589          * cache may be implemented at a later date which takes further advantage of these strict
2590          * dependency requirements.
2591          *
2592          * @private
2593          */
2594         function preprocessToDom ( $text, $flags = 0 ) {
2595                 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2596                 return $dom;
2597         }
2598
2599         /*
2600          * Return a three-element array: leading whitespace, string contents, trailing whitespace
2601          */
2602         public static function splitWhitespace( $s ) {
2603                 $ltrimmed = ltrim( $s );
2604                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2605                 $trimmed = rtrim( $ltrimmed );
2606                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2607                 if ( $diff > 0 ) {
2608                         $w2 = substr( $ltrimmed, -$diff );
2609                 } else {
2610                         $w2 = '';
2611                 }
2612                 return array( $w1, $trimmed, $w2 );
2613         }
2614
2615         /**
2616          * Replace magic variables, templates, and template arguments
2617          * with the appropriate text. Templates are substituted recursively,
2618          * taking care to avoid infinite loops.
2619          *
2620          * Note that the substitution depends on value of $mOutputType:
2621          *  self::OT_WIKI: only {{subst:}} templates
2622          *  self::OT_PREPROCESS: templates but not extension tags
2623          *  self::OT_HTML: all templates and extension tags
2624          *
2625          * @param string $tex The text to transform
2626          * @param PPFrame $frame Object describing the arguments passed to the template
2627          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2628          * @private
2629          */
2630         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2631                 # Prevent too big inclusions
2632                 if( strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2633                         return $text;
2634                 }
2635
2636                 $fname = __METHOD__;
2637                 wfProfileIn( $fname );
2638
2639                 if ( $frame === false ) {
2640                         $frame = $this->getPreprocessor()->newFrame();
2641                 } elseif ( !( $frame instanceof PPFrame ) ) {
2642                         throw new MWException( __METHOD__ . ' called using the old argument format' );
2643                 }
2644
2645                 $dom = $this->preprocessToDom( $text );
2646                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2647                 $text = $frame->expand( $dom, $flags );
2648
2649                 wfProfileOut( $fname );
2650                 return $text;
2651         }
2652
2653         /// Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2654         static function createAssocArgs( $args ) {
2655                 $assocArgs = array();
2656                 $index = 1;
2657                 foreach( $args as $arg ) {
2658                         $eqpos = strpos( $arg, '=' );
2659                         if ( $eqpos === false ) {
2660                                 $assocArgs[$index++] = $arg;
2661                         } else {
2662                                 $name = trim( substr( $arg, 0, $eqpos ) );
2663                                 $value = trim( substr( $arg, $eqpos+1 ) );
2664                                 if ( $value === false ) {
2665                                         $value = '';
2666                                 }
2667                                 if ( $name !== false ) {
2668                                         $assocArgs[$name] = $value;
2669                                 }
2670                         }
2671                 }
2672
2673                 return $assocArgs;
2674         }
2675
2676         /**
2677          * Return the text of a template, after recursively
2678          * replacing any variables or templates within the template.
2679          *
2680          * @param array $piece The parts of the template
2681          *  $piece['title']: the title, i.e. the part before the |
2682          *  $piece['parts']: the parameter array
2683          *  $piece['lineStart']: whether the brace was at the start of a line
2684          * @param PPFrame The current frame, contains template arguments
2685          * @return string the text of the template
2686          * @private
2687          */
2688         function braceSubstitution( $piece, $frame ) {
2689                 global $wgContLang, $wgLang, $wgAllowDisplayTitle, $wgNonincludableNamespaces;
2690                 $fname = __METHOD__;
2691                 wfProfileIn( $fname );
2692                 wfProfileIn( __METHOD__.'-setup' );
2693
2694                 # Flags
2695                 $found = false;             # $text has been filled
2696                 $nowiki = false;            # wiki markup in $text should be escaped
2697                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2698                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2699                 $isChildObj = false;        # $text is a DOM node needing expansion in a child frame
2700                 $isLocalObj = false;        # $text is a DOM node needing expansion in the current frame
2701
2702                 # Title object, where $text came from
2703                 $title = NULL;
2704
2705                 # $part1 is the bit before the first |, and must contain only title characters.
2706                 # Various prefixes will be stripped from it later.
2707                 $titleWithSpaces = $frame->expand( $piece['title'] );
2708                 $part1 = trim( $titleWithSpaces );
2709                 $titleText = false;
2710
2711                 # Original title text preserved for various purposes
2712                 $originalTitle = $part1;
2713
2714                 # $args is a list of argument nodes, starting from index 0, not including $part1
2715                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2716                 wfProfileOut( __METHOD__.'-setup' );
2717
2718                 # SUBST
2719                 wfProfileIn( __METHOD__.'-modifiers' );
2720                 if ( !$found ) {
2721                         $mwSubst =& MagicWord::get( 'subst' );
2722                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor $this->ot['wiki'] ) {
2723                                 # One of two possibilities is true:
2724                                 # 1) Found SUBST but not in the PST phase
2725                                 # 2) Didn't find SUBST and in the PST phase
2726                                 # In either case, return without further processing
2727                                 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2728                                 $isLocalObj = true;
2729                                 $found = true;
2730                         }
2731                 }
2732
2733                 # Variables
2734                 if ( !$found && $args->getLength() == 0 ) {
2735                         $id = $this->mVariables->matchStartToEnd( $part1 );
2736                         if ( $id !== false ) {
2737                                 $text = $this->getVariableValue( $id );
2738                                 if (MagicWord::getCacheTTL($id)>-1)
2739                                         $this->mOutput->mContainsOldMagic = true;
2740                                 $found = true;
2741                         }
2742                 }
2743
2744                 # MSG, MSGNW and RAW
2745                 if ( !$found ) {
2746                         # Check for MSGNW:
2747                         $mwMsgnw =& MagicWord::get( 'msgnw' );
2748                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2749                                 $nowiki = true;
2750                         } else {
2751                                 # Remove obsolete MSG:
2752                                 $mwMsg =& MagicWord::get( 'msg' );
2753                                 $mwMsg->matchStartAndRemove( $part1 );
2754                         }
2755
2756                         # Check for RAW:
2757                         $mwRaw =& MagicWord::get( 'raw' );
2758                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2759                                 $forceRawInterwiki = true;
2760                         }
2761                 }
2762                 wfProfileOut( __METHOD__.'-modifiers' );
2763
2764                 # Parser functions
2765                 if ( !$found ) {
2766                         wfProfileIn( __METHOD__ . '-pfunc' );
2767
2768                         $colonPos = strpos( $part1, ':' );
2769                         if ( $colonPos !== false ) {
2770                                 # Case sensitive functions
2771                                 $function = substr( $part1, 0, $colonPos );
2772                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
2773                                         $function = $this->mFunctionSynonyms[1][$function];
2774                                 } else {
2775                                         # Case insensitive functions
2776                                         $function = strtolower( $function );
2777                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
2778                                                 $function = $this->mFunctionSynonyms[0][$function];
2779                                         } else {
2780                                                 $function = false;
2781                                         }
2782                                 }
2783                                 if ( $function ) {
2784                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
2785                                         $initialArgs = array( &$this );
2786                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
2787                                         if ( $flags & SFH_OBJECT_ARGS ) {
2788                                                 # Add a frame parameter, and pass the arguments as an array
2789                                                 $allArgs = $initialArgs;
2790                                                 $allArgs[] = $frame;
2791                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2792                                                         $funcArgs[] = $args->item( $i );
2793                                                 }
2794                                                 $allArgs[] = $funcArgs;
2795                                         } else {
2796                                                 # Convert arguments to plain text
2797                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2798                                                         $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
2799                                                 }
2800                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
2801                                         }
2802
2803                                         # Workaround for PHP bug 35229 and similar
2804                                         if ( !is_callable( $callback ) ) {
2805                                                 throw new MWException( "Tag hook for $name is not callable\n" );
2806                                         }
2807                                         $result = call_user_func_array( $callback, $allArgs );
2808                                         $found = true;
2809
2810                                         if ( is_array( $result ) ) {
2811                                                 if ( isset( $result[0] ) ) {
2812                                                         $text = $result[0];
2813                                                         unset( $result[0] );
2814                                                 }
2815
2816                                                 // Extract flags into the local scope
2817                                                 // This allows callers to set flags such as nowiki, found, etc.
2818                                                 extract( $result );
2819                                         } else {
2820                                                 $text = $result;
2821                                         }
2822                                 }
2823                         }
2824                         wfProfileOut( __METHOD__ . '-pfunc' );
2825                 }
2826
2827                 # Finish mangling title and then check for loops.
2828                 # Set $title to a Title object and $titleText to the PDBK
2829                 if ( !$found ) {
2830                         $ns = NS_TEMPLATE;
2831                         # Split the title into page and subpage
2832                         $subpage = '';
2833                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
2834                         if ($subpage !== '') {
2835                                 $ns = $this->mTitle->getNamespace();
2836                         }
2837                         $title = Title::newFromText( $part1, $ns );
2838                         if ( $title ) {
2839                                 $titleText = $title->getPrefixedText();
2840                                 # Check for language variants if the template is not found
2841                                 if($wgContLang->hasVariants() && $title->getArticleID() == 0){
2842                                         $wgContLang->findVariantLink($part1, $title);
2843                                 }
2844                                 # Do infinite loop check
2845                                 if ( !$frame->loopCheck( $title ) ) {
2846                                         $found = true;
2847                                         $text = "<span class=\"error\">Template loop detected: [[$titleText]]</span>";
2848                                         wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
2849                                 }
2850                                 # Do recursion depth check
2851                                 $limit = $this->mOptions->getMaxTemplateDepth();
2852                                 if ( $frame->depth >= $limit ) {
2853                                         $found = true;
2854                                         $text = "<span class=\"error\">Template recursion depth limit exceeded ($limit)</span>";
2855                                 }
2856                         }
2857                 }
2858
2859                 # Load from database
2860                 if ( !$found && $title ) {
2861                         wfProfileIn( __METHOD__ . '-loadtpl' );
2862                         if ( !$title->isExternal() ) {
2863                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
2864                                         $text = SpecialPage::capturePath( $title );
2865                                         if ( is_string( $text ) ) {
2866                                                 $found = true;
2867                                                 $isHTML = true;
2868                                                 $this->disableCache();
2869                                         }
2870                                 } else if ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
2871                                         $found = false; //access denied
2872                                         wfDebug( "$fname: template inclusion denied for " . $title->getPrefixedDBkey() );
2873                                 } else {
2874                                         list( $text, $title ) = $this->getTemplateDom( $title );
2875                                         if ( $text !== false ) {
2876                                                 $found = true;
2877                                                 $isChildObj = true;
2878                                         }
2879                                 }
2880
2881                                 # If the title is valid but undisplayable, make a link to it
2882                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
2883                                         $text = "[[:$titleText]]";
2884                                         $found = true;
2885                                 }
2886                         } elseif ( $title->isTrans() ) {
2887                                 // Interwiki transclusion
2888                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
2889                                         $text = $this->interwikiTransclude( $title, 'render' );
2890                                         $isHTML = true;
2891                                 } else {
2892                                         $text = $this->interwikiTransclude( $title, 'raw' );
2893                                         // Preprocess it like a template
2894                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
2895                                         $isChildObj = true;
2896                                 }
2897                                 $found = true;
2898                         }
2899                         wfProfileOut( __METHOD__ . '-loadtpl' );
2900                 }
2901
2902                 # If we haven't found text to substitute by now, we're done
2903                 # Recover the source wikitext and return it
2904                 if ( !$found ) {
2905                         $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2906                         wfProfileOut( $fname );
2907                         return array( 'object' => $text );
2908                 }
2909
2910                 # Expand DOM-style return values in a child frame
2911                 if ( $isChildObj ) {
2912                         # Clean up argument array
2913                         $newFrame = $frame->newChild( $args, $title );
2914
2915                         if ( $nowiki ) {
2916                                 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
2917                         } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
2918                                 # Expansion is eligible for the empty-frame cache
2919                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
2920                                         $text = $this->mTplExpandCache[$titleText];
2921                                 } else {
2922                                         $text = $newFrame->expand( $text );
2923                                         $this->mTplExpandCache[$titleText] = $text;
2924                                 }
2925                         } else {
2926                                 # Uncached expansion
2927                                 $text = $newFrame->expand( $text );
2928                         }
2929                 }
2930                 if ( $isLocalObj && $nowiki ) {
2931                         $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
2932                         $isLocalObj = false;
2933                 }
2934
2935                 # Replace raw HTML by a placeholder
2936                 # Add a blank line preceding, to prevent it from mucking up
2937                 # immediately preceding headings
2938                 if ( $isHTML ) {
2939                         $text = "\n\n" . $this->insertStripItem( $text );
2940                 }
2941                 # Escape nowiki-style return values
2942                 elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
2943                         $text = wfEscapeWikiText( $text );
2944                 }
2945                 # Bug 529: if the template begins with a table or block-level
2946                 # element, it should be treated as beginning a new line.
2947                 # This behaviour is somewhat controversial.
2948                 elseif ( is_string( $text ) && !$piece['lineStart'] && preg_match('/^(?:{\\||:|;|#|\*)/', $text)) /*}*/{
2949                         $text = "\n" . $text;
2950                 }
2951
2952                 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
2953                         # Error, oversize inclusion
2954                         $text = "[[$originalTitle]]" .
2955                                 $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
2956                 }
2957
2958                 if ( $isLocalObj ) {
2959                         $ret = array( 'object' => $text );
2960                 } else {
2961                         $ret = array( 'text' => $text );
2962                 }
2963
2964                 wfProfileOut( $fname );
2965                 return $ret;
2966         }
2967
2968         /**
2969          * Get the semi-parsed DOM representation of a template with a given title,
2970          * and its redirect destination title. Cached.
2971          */
2972         function getTemplateDom( $title ) {
2973                 $cacheTitle = $title;
2974                 $titleText = $title->getPrefixedDBkey();
2975
2976                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
2977                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
2978                         $title = Title::makeTitle( $ns, $dbk );
2979                         $titleText = $title->getPrefixedDBkey();
2980                 }
2981                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
2982                         return array( $this->mTplDomCache[$titleText], $title );
2983                 }
2984
2985                 // Cache miss, go to the database
2986                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
2987
2988                 if ( $text === false ) {
2989                         $this->mTplDomCache[$titleText] = false;
2990                         return array( false, $title );
2991                 }
2992
2993                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
2994                 $this->mTplDomCache[ $titleText ] = $dom;
2995
2996                 if (! $title->equals($cacheTitle)) {
2997                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
2998                                 array( $title->getNamespace(),$cdb = $title->getDBkey() );
2999                 }
3000
3001                 return array( $dom, $title );
3002         }
3003
3004         /**
3005          * Fetch the unparsed text of a template and register a reference to it.
3006          */
3007         function fetchTemplateAndTitle( $title ) {
3008                 $templateCb = $this->mOptions->getTemplateCallback();
3009                 $stuff = call_user_func( $templateCb, $title );
3010                 $text = $stuff['text'];
3011                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3012                 if ( isset( $stuff['deps'] ) ) {
3013                         foreach ( $stuff['deps'] as $dep ) {
3014                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3015                         }
3016                 }
3017                 return array($text,$finalTitle);
3018         }
3019
3020         function fetchTemplate( $title ) {
3021                 $rv = $this->fetchTemplateAndTitle($title);
3022                 return $rv[0];
3023         }
3024
3025         /**
3026          * Static function to get a template
3027          * Can be overridden via ParserOptions::setTemplateCallback().
3028          */
3029         static function statelessFetchTemplate( $title ) {
3030                 $text = $skip = false;
3031                 $finalTitle = $title;
3032                 $deps = array();
3033
3034                 // Loop to fetch the article, with up to 1 redirect
3035                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3036                         # Give extensions a chance to select the revision instead
3037                         $id = false; // Assume current
3038                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( false, &$title, &$skip, &$id ) );
3039
3040                         if( $skip ) {
3041                                 $text = false;
3042                                 $deps[] = array(
3043                                         'title' => $title,
3044                                         'page_id' => $title->getArticleID(),
3045                                         'rev_id' => null );
3046                                 break;
3047                         }
3048                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3049                         $rev_id = $rev ? $rev->getId() : 0;
3050
3051                         $deps[] = array(
3052                                 'title' => $title,
3053                                 'page_id' => $title->getArticleID(),
3054                                 'rev_id' => $rev_id );
3055
3056                         if( $rev ) {
3057                                 $text = $rev->getText();
3058                         } elseif( $title->getNamespace() == NS_MEDIAWIKI ) {
3059                                 global $wgLang;
3060                                 $message = $wgLang->lcfirst( $title->getText() );
3061                                 $text = wfMsgForContentNoTrans( $message );
3062                                 if( wfEmptyMsg( $message, $text ) ) {
3063                                         $text = false;
3064                                         break;
3065                                 }
3066                         } else {
3067                                 break;
3068                         }
3069                         if ( $text === false ) {
3070                                 break;
3071                         }
3072                         // Redirect?
3073                         $finalTitle = $title;
3074                         $title = Title::newFromRedirect( $text );
3075                 }
3076                 return array(
3077                         'text' => $text,
3078                         'finalTitle' => $finalTitle,
3079                         'deps' => $deps );
3080         }
3081
3082         /**
3083          * Transclude an interwiki link.
3084          */
3085         function interwikiTransclude( $title, $action ) {
3086                 global $wgEnableScaryTranscluding;
3087
3088                 if (!$wgEnableScaryTranscluding)
3089                         return wfMsg('scarytranscludedisabled');
3090
3091                 $url = $title->getFullUrl( "action=$action" );
3092
3093                 if (strlen($url) > 255)
3094                         return wfMsg('scarytranscludetoolong');
3095                 return $this->fetchScaryTemplateMaybeFromCache($url);
3096         }
3097
3098         function fetchScaryTemplateMaybeFromCache($url) {
3099                 global $wgTranscludeCacheExpiry;
3100                 $dbr = wfGetDB(DB_SLAVE);
3101                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3102                                 array('tc_url' => $url));
3103                 if ($obj) {
3104                         $time = $obj->tc_time;
3105                         $text = $obj->tc_contents;
3106                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
3107                                 return $text;
3108                         }
3109                 }
3110
3111                 $text = Http::get($url);
3112                 if (!$text)
3113                         return wfMsg('scarytranscludefailed', $url);
3114
3115                 $dbw = wfGetDB(DB_MASTER);
3116                 $dbw->replace('transcache', array('tc_url'), array(
3117                         'tc_url' => $url,
3118                         'tc_time' => time(),
3119                         'tc_contents' => $text));
3120                 return $text;
3121         }
3122
3123
3124         /**
3125          * Triple brace replacement -- used for template arguments
3126          * @private
3127          */
3128         function argSubstitution( $piece, $frame ) {
3129                 wfProfileIn( __METHOD__ );
3130
3131                 $error = false;
3132                 $parts = $piece['parts'];
3133                 $nameWithSpaces = $frame->expand( $piece['title'] );
3134                 $argName = trim( $nameWithSpaces );
3135                 $object = false;
3136                 $text = $frame->getArgument( $argName );
3137                 if (  $text === false && $parts->getLength() > 0
3138                   && (
3139                     $this->ot['html']
3140                     || $this->ot['pre']
3141                     || ( $this->ot['wiki'] && $frame->isTemplate() )
3142                   )
3143                 ) {
3144                         # No match in frame, use the supplied default
3145                         $object = $parts->item( 0 )->getChildren();
3146                 }
3147                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3148                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3149                 }
3150
3151                 if ( $text === false && $object === false ) {
3152                         # No match anywhere
3153                         $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3154                 }
3155                 if ( $error !== false ) {
3156                         $text .= $error;
3157                 }
3158                 if ( $object !== false ) {
3159                         $ret = array( 'object' => $object );
3160                 } else {
3161                         $ret = array( 'text' => $text );
3162                 }
3163
3164                 wfProfileOut( __METHOD__ );
3165                 return $ret;
3166         }
3167
3168         /**
3169          * Return the text to be used for a given extension tag.
3170          * This is the ghost of strip().
3171          *
3172          * @param array $params Associative array of parameters:
3173          *     name       PPNode for the tag name
3174          *     attr       PPNode for unparsed text where tag attributes are thought to be
3175          *     attributes Optional associative array of parsed attributes
3176          *     inner      Contents of extension element
3177          *     noClose    Original text did not have a close tag
3178          * @param PPFrame $frame
3179          */
3180         function extensionSubstitution( $params, $frame ) {
3181                 global $wgRawHtml, $wgContLang;
3182
3183                 $name = $frame->expand( $params['name'] );
3184                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3185                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3186
3187                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf('%08X', $this->mMarkerIndex++) . $this->mMarkerSuffix;
3188
3189                 if ( $this->ot['html'] ) {
3190                         $name = strtolower( $name );
3191
3192                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3193                         if ( isset( $params['attributes'] ) ) {
3194                                 $attributes = $attributes + $params['attributes'];
3195                         }
3196                         switch ( $name ) {
3197                                 case 'html':
3198                                         if( $wgRawHtml ) {
3199                                                 $output = $content;
3200                                                 break;
3201                                         } else {
3202                                                 throw new MWException( '<html> extension tag encountered unexpectedly' );
3203                                         }
3204                                 case 'nowiki':
3205                                         $output = Xml::escapeTagsOnly( $content );
3206                                         break;
3207                                 case 'math':
3208                                         $output = $wgContLang->armourMath(
3209                                                 MathRenderer::renderMath( $content, $attributes ) );
3210                                         break;
3211                                 case 'gallery':
3212                                         $output = $this->renderImageGallery( $content, $attributes );
3213                                         break;
3214                                 default:
3215                                         if( isset( $this->mTagHooks[$name] ) ) {
3216                                                 # Workaround for PHP bug 35229 and similar
3217                                                 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3218                                                         throw new MWException( "Tag hook for $name is not callable\n" );
3219                                                 }
3220                                                 $output = call_user_func_array( $this->mTagHooks[$name],
3221                                                         array( $content, $attributes, $this ) );
3222                                         } else {
3223                                                 throw new MWException( "Invalid call hook $name" );
3224                                         }
3225                         }
3226                 } else {
3227                         if ( is_null( $attrText ) ) {
3228                                 $attrText = '';
3229                         }
3230                         if ( isset( $params['attributes'] ) ) {
3231                                 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3232                                         $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3233                                                 htmlspecialchars( $attrValue ) . '"';
3234                                 }
3235                         }
3236                         if ( $content === null ) {
3237                                 $output = "<$name$attrText/>";
3238                         } else {
3239                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3240                                 $output = "<$name$attrText>$content$close";
3241                         }
3242                 }
3243
3244                 if ( $name == 'html' || $name == 'nowiki' ) {
3245                         $this->mStripState->nowiki->setPair( $marker, $output );
3246                 } else {
3247                         $this->mStripState->general->setPair( $marker, $output );
3248                 }
3249                 return $marker;
3250         }
3251
3252         /**
3253          * Increment an include size counter
3254          *
3255          * @param string $type The type of expansion
3256          * @param integer $size The size of the text
3257          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3258          */
3259         function incrementIncludeSize( $type, $size ) {
3260                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize( $type ) ) {
3261                         return false;
3262                 } else {
3263                         $this->mIncludeSizes[$type] += $size;
3264                         return true;
3265                 }
3266         }
3267
3268         /**
3269          * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3270          * Fills $this->mDoubleUnderscores, returns the modified text
3271          */
3272         function doDoubleUnderscore( $text ) {
3273                 // The position of __TOC__ needs to be recorded
3274                 $mw = MagicWord::get( 'toc' );
3275                 if( $mw->match( $text ) ) {
3276                         $this->mShowToc = true;
3277                         $this->mForceTocPosition = true;
3278
3279                         // Set a placeholder. At the end we'll fill it in with the TOC.
3280                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3281
3282                         // Only keep the first one.
3283                         $text = $mw->replace( '', $text );
3284                 }
3285
3286                 // Now match and remove the rest of them
3287                 $mwa = MagicWord::getDoubleUnderscoreArray();
3288                 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3289
3290                 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3291                         $this->mOutput->mNoGallery = true;
3292                 }
3293                 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3294                         $this->mShowToc = false;
3295                 }
3296                 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
3297                         $this->mOutput->setProperty( 'hiddencat', 'y' );
3298
3299                         $containerCategory = Title::makeTitleSafe( NS_CATEGORY, wfMsgForContent( 'hidden-category-category' ) );
3300                         if ( $containerCategory ) {
3301                                 $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3302                         } else {
3303                                 wfDebug( __METHOD__.": [[MediaWiki:hidden-category-category]] is not a valid title!\n" );
3304                         }
3305                 }
3306                 return $text;
3307         }
3308
3309         /**
3310          * This function accomplishes several tasks:
3311          * 1) Auto-number headings if that option is enabled
3312          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3313          * 3) Add a Table of contents on the top for users who have enabled the option
3314          * 4) Auto-anchor headings
3315          *
3316          * It loops through all headlines, collects the necessary data, then splits up the
3317          * string and re-inserts the newly formatted headlines.
3318          *
3319          * @param string $text
3320          * @param boolean $isMain
3321          * @private
3322          */
3323         function formatHeadings( $text, $isMain=true ) {
3324                 global $wgMaxTocLevel, $wgContLang;
3325
3326                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3327                 if( !$this->mTitle->quickUserCan( 'edit' ) ) {
3328                         $showEditLink = 0;
3329                 } else {
3330                         $showEditLink = $this->mOptions->getEditSection();
3331                 }
3332
3333                 # Inhibit editsection links if requested in the page
3334                 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
3335                         $showEditLink = 0;
3336                 }
3337
3338                 # Get all headlines for numbering them and adding funky stuff like [edit]
3339                 # links - this is for later, but we need the number of headlines right now
3340                 $matches = array();
3341                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3342
3343                 # if there are fewer than 4 headlines in the article, do not show TOC
3344                 # unless it's been explicitly enabled.
3345                 $enoughToc = $this->mShowToc &&
3346                         (($numMatches >= 4) || $this->mForceTocPosition);
3347
3348                 # Allow user to stipulate that a page should have a "new section"
3349                 # link added via __NEWSECTIONLINK__
3350                 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
3351                         $this->mOutput->setNewSection( true );
3352                 }
3353
3354                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3355                 # override above conditions and always show TOC above first header
3356                 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
3357                         $this->mShowToc = true;
3358                         $enoughToc = true;
3359                 }
3360
3361                 # We need this to perform operations on the HTML
3362                 $sk = $this->mOptions->getSkin();
3363
3364                 # headline counter
3365                 $headlineCount = 0;
3366                 $numVisible = 0;
3367
3368                 # Ugh .. the TOC should have neat indentation levels which can be
3369                 # passed to the skin functions. These are determined here
3370                 $toc = '';
3371                 $full = '';
3372                 $head = array();
3373                 $sublevelCount = array();
3374                 $levelCount = array();
3375                 $toclevel = 0;
3376                 $level = 0;
3377                 $prevlevel = 0;
3378                 $toclevel = 0;
3379                 $prevtoclevel = 0;
3380                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-{$this->mMarkerSuffix}";
3381                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3382                 $tocraw = array();
3383
3384                 foreach( $matches[3] as $headline ) {
3385                         $isTemplate = false;
3386                         $titleText = false;
3387                         $sectionIndex = false;
3388                         $numbering = '';
3389                         $markerMatches = array();
3390                         if (preg_match("/^$markerRegex/", $headline, $markerMatches)) {
3391                                 $serial = $markerMatches[1];
3392                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3393                                 $isTemplate = ($titleText != $baseTitleText);
3394                                 $headline = preg_replace("/^$markerRegex/", "", $headline);
3395                         }
3396
3397                         if( $toclevel ) {
3398                                 $prevlevel = $level;
3399                                 $prevtoclevel = $toclevel;
3400                         }
3401                         $level = $matches[1][$headlineCount];
3402
3403                         if( $doNumberHeadings || $enoughToc ) {
3404
3405                                 if ( $level > $prevlevel ) {
3406                                         # Increase TOC level
3407                                         $toclevel++;
3408                                         $sublevelCount[$toclevel] = 0;
3409                                         if( $toclevel<$wgMaxTocLevel ) {
3410                                                 $prevtoclevel = $toclevel;
3411                                                 $toc .= $sk->tocIndent();
3412                                                 $numVisible++;
3413                                         }
3414                                 }
3415                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3416                                         # Decrease TOC level, find level to jump to
3417
3418                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
3419                                                 # Can only go down to level 1
3420                                                 $toclevel = 1;
3421                                         } else {
3422                                                 for ($i = $toclevel; $i > 0; $i--) {
3423                                                         if ( $levelCount[$i] == $level ) {
3424                                                                 # Found last matching level
3425                                                                 $toclevel = $i;
3426                                                                 break;
3427                                                         }
3428                                                         elseif ( $levelCount[$i] < $level ) {
3429                                                                 # Found first matching level below current level
3430                                                                 $toclevel = $i + 1;
3431                                                                 break;
3432                                                         }
3433                                                 }
3434                                         }
3435                                         if( $toclevel<$wgMaxTocLevel ) {
3436                                                 if($prevtoclevel < $wgMaxTocLevel) {
3437                                                         # Unindent only if the previous toc level was shown :p
3438                                                         $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3439                                                 } else {
3440                                                         $toc .= $sk->tocLineEnd();
3441                                                 }
3442                                         }
3443                                 }
3444                                 else {
3445                                         # No change in level, end TOC line
3446                                         if( $toclevel<$wgMaxTocLevel ) {
3447                                                 $toc .= $sk->tocLineEnd();
3448                                         }
3449                                 }
3450
3451                                 $levelCount[$toclevel] = $level;
3452
3453                                 # count number of headlines for each level
3454                                 @$sublevelCount[$toclevel]++;
3455                                 $dot = 0;
3456                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3457                                         if( !empty( $sublevelCount[$i] ) ) {
3458                                                 if( $dot ) {
3459                                                         $numbering .= '.';
3460                                                 }
3461                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3462                                                 $dot = 1;
3463                                         }
3464                                 }
3465                         }
3466
3467                         # The safe header is a version of the header text safe to use for links
3468                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3469                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
3470
3471                         # Remove link placeholders by the link text.
3472                         #     <!--LINK number-->
3473                         # turns into
3474                         #     link text with suffix
3475                         $safeHeadline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
3476                                                             "\$this->mLinkHolders['texts'][\$1]",
3477                                                             $safeHeadline );
3478                         $safeHeadline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
3479                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
3480                                                             $safeHeadline );
3481
3482                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3483                         $tocline = preg_replace(
3484                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3485                                 array( '',                          '<$1>'),
3486                                 $safeHeadline
3487                         );
3488                         $tocline = trim( $tocline );
3489
3490                         # For the anchor, strip out HTML-y stuff period
3491                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
3492                         $safeHeadline = trim( $safeHeadline );
3493
3494                         # Save headline for section edit hint before it's escaped
3495                         $headlineHint = $safeHeadline;
3496                         $safeHeadline = Sanitizer::escapeId( $safeHeadline );
3497                         # HTML names must be case-insensitively unique (bug 10721)
3498                         $arrayKey = strtolower( $safeHeadline );
3499
3500                         # XXX : Is $refers[$headlineCount] ever accessed, actually ?
3501                         $refers[$headlineCount] = $safeHeadline;
3502
3503                         # count how many in assoc. array so we can track dupes in anchors
3504                         isset( $refers[$arrayKey] ) ? $refers[$arrayKey]++ : $refers[$arrayKey] = 1;
3505                         $refcount[$headlineCount] = $refers[$arrayKey];
3506
3507                         # Don't number the heading if it is the only one (looks silly)
3508                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3509                                 # the two are different if the line contains a link
3510                                 $headline=$numbering . ' ' . $headline;
3511                         }
3512
3513                         # Create the anchor for linking from the TOC to the section
3514                         $anchor = $safeHeadline;
3515                         if($refcount[$headlineCount] > 1 ) {
3516                                 $anchor .= '_' . $refcount[$headlineCount];
3517                         }
3518                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3519                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3520                                 $tocraw[] = array( 'toclevel' => $toclevel, 'level' => $level, 'line' => $tocline, 'number' => $numbering );
3521                         }
3522                         # give headline the correct <h#> tag
3523                         if( $showEditLink && $sectionIndex !== false ) {
3524                                 if( $isTemplate ) {
3525                                         # Put a T flag in the section identifier, to indicate to extractSections()
3526                                         # that sections inside <includeonly> should be counted.
3527                                         $editlink = $sk->editSectionLinkForOther($titleText, "T-$sectionIndex");
3528                                 } else {
3529                                         $editlink = $sk->editSectionLink($this->mTitle, $sectionIndex, $headlineHint);
3530                                 }
3531                         } else {
3532                                 $editlink = '';
3533                         }
3534                         $head[$headlineCount] = $sk->makeHeadline( $level, $matches['attrib'][$headlineCount], $anchor, $headline, $editlink );
3535
3536                         $headlineCount++;
3537                 }
3538
3539                 $this->mOutput->setSections( $tocraw );
3540
3541                 # Never ever show TOC if no headers
3542                 if( $numVisible < 1 ) {
3543                         $enoughToc = false;
3544                 }
3545
3546                 if( $enoughToc ) {
3547                         if( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
3548                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
3549                         }
3550                         $toc = $sk->tocList( $toc );
3551                 }
3552
3553                 # split up and insert constructed headlines
3554
3555                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3556                 $i = 0;
3557
3558                 foreach( $blocks as $block ) {
3559                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
3560                                 # This is the [edit] link that appears for the top block of text when
3561                                 # section editing is enabled
3562
3563                                 # Disabled because it broke block formatting
3564                                 # For example, a bullet point in the top line
3565                                 # $full .= $sk->editSectionLink(0);
3566                         }
3567                         $full .= $block;
3568                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3569                                 # Top anchor now in skin
3570                                 $full = $full.$toc;
3571                         }
3572
3573                         if( !empty( $head[$i] ) ) {
3574                                 $full .= $head[$i];
3575                         }
3576                         $i++;
3577                 }
3578                 if( $this->mForceTocPosition ) {
3579                         return str_replace( '<!--MWTOC-->', $toc, $full );
3580                 } else {
3581                         return $full;
3582                 }
3583         }
3584
3585         /**
3586          * Transform wiki markup when saving a page by doing \r\n -> \n
3587          * conversion, substitting signatures, {{subst:}} templates, etc.
3588          *
3589          * @param string $text the text to transform
3590          * @param Title &$title the Title object for the current article
3591          * @param User &$user the User object describing the current user
3592          * @param ParserOptions $options parsing options
3593          * @param bool $clearState whether to clear the parser state first
3594          * @return string the altered wiki markup
3595          * @public
3596          */
3597         function preSaveTransform( $text, &$title, $user, $options, $clearState = true ) {
3598                 $this->mOptions = $options;
3599                 $this->setTitle( $title );
3600                 $this->setOutputType( self::OT_WIKI );
3601
3602                 if ( $clearState ) {
3603                         $this->clearState();
3604                 }
3605
3606                 $pairs = array(
3607                         "\r\n" => "\n",
3608                 );
3609                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3610                 $text = $this->pstPass2( $text, $user );
3611                 $text = $this->mStripState->unstripBoth( $text );
3612                 return $text;
3613         }
3614
3615         /**
3616          * Pre-save transform helper function
3617          * @private
3618          */
3619         function pstPass2( $text, $user ) {
3620                 global $wgContLang, $wgLocaltimezone;
3621
3622                 /* Note: This is the timestamp saved as hardcoded wikitext to
3623                  * the database, we use $wgContLang here in order to give
3624                  * everyone the same signature and use the default one rather
3625                  * than the one selected in each user's preferences.
3626                  *
3627                  * (see also bug 12815)
3628                  */
3629                 $ts = $this->mOptions->getTimestamp();
3630                 $tz = 'UTC';
3631                 if ( isset( $wgLocaltimezone ) ) {
3632                         $unixts = wfTimestamp( TS_UNIX, $ts );
3633                         $oldtz = getenv( 'TZ' );
3634                         putenv( 'TZ='.$wgLocaltimezone );
3635                         $ts = date( 'YmdHis', $unixts );
3636                         $tz = date( 'T', $unixts );  # might vary on DST changeover!
3637                         putenv( 'TZ='.$oldtz );
3638                 }
3639                 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tz)";
3640
3641                 # Variable replacement
3642                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3643                 $text = $this->replaceVariables( $text );
3644
3645                 # Signatures
3646                 $sigText = $this->getUserSig( $user );
3647                 $text = strtr( $text, array(
3648                         '~~~~~' => $d,
3649                         '~~~~' => "$sigText $d",
3650                         '~~~' => $sigText
3651                 ) );
3652
3653                 # Context links: [[|name]] and [[name (context)|]]
3654                 #
3655                 global $wgLegalTitleChars;
3656                 $tc = "[$wgLegalTitleChars]";
3657                 $nc = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
3658
3659                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
3660                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
3661                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
3662
3663                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
3664                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
3665                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
3666
3667                 $t = $this->mTitle->getText();
3668                 $m = array();
3669                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
3670                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3671                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && '' != "$m[1]$m[2]" ) {
3672                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3673                 } else {
3674                         # if there's no context, don't bother duplicating the title
3675                         $text = preg_replace( $p2, '[[\\1]]', $text );
3676                 }
3677
3678                 # Trim trailing whitespace
3679                 $text = rtrim( $text );
3680
3681                 return $text;
3682         }
3683
3684         /**
3685          * Fetch the user's signature text, if any, and normalize to
3686          * validated, ready-to-insert wikitext.
3687          *
3688          * @param User $user
3689          * @return string
3690          * @private
3691          */
3692         function getUserSig( &$user ) {
3693                 global $wgMaxSigChars;
3694
3695                 $username = $user->getName();
3696                 $nickname = $user->getOption( 'nickname' );
3697                 $nickname = $nickname === '' ? $username : $nickname;
3698
3699                 if( mb_strlen( $nickname ) > $wgMaxSigChars ) {
3700                         $nickname = $username;
3701                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
3702                 } elseif( $user->getBoolOption( 'fancysig' ) !== false ) {
3703                         # Sig. might contain markup; validate this
3704                         if( $this->validateSig( $nickname ) !== false ) {
3705                                 # Validated; clean up (if needed) and return it
3706                                 return $this->cleanSig( $nickname, true );
3707                         } else {
3708                                 # Failed to validate; fall back to the default
3709                                 $nickname = $username;
3710                                 wfDebug( "Parser::getUserSig: $username has bad XML tags in signature.\n" );
3711                         }
3712                 }
3713
3714                 // Make sure nickname doesnt get a sig in a sig
3715                 $nickname = $this->cleanSigInSig( $nickname );
3716
3717                 # If we're still here, make it a link to the user page
3718                 $userText = wfEscapeWikiText( $username );
3719                 $nickText = wfEscapeWikiText( $nickname );
3720                 if ( $user->isAnon() )  {
3721                         return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
3722                 } else {
3723                         return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
3724                 }
3725         }
3726
3727         /**
3728          * Check that the user's signature contains no bad XML
3729          *
3730          * @param string $text
3731          * @return mixed An expanded string, or false if invalid.
3732          */
3733         function validateSig( $text ) {
3734                 return( wfIsWellFormedXmlFragment( $text ) ? $text : false );
3735         }
3736
3737         /**
3738          * Clean up signature text
3739          *
3740          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
3741          * 2) Substitute all transclusions
3742          *
3743          * @param string $text
3744          * @param $parsing Whether we're cleaning (preferences save) or parsing
3745          * @return string Signature text
3746          */
3747         function cleanSig( $text, $parsing = false ) {
3748                 if ( !$parsing ) {
3749                         global $wgTitle;
3750                         $this->clearState();
3751                         $this->setTitle( $wgTitle );
3752                         $this->mOptions = new ParserOptions;
3753                         $this->setOutputType = self::OT_PREPROCESS;
3754                 }
3755
3756                 # FIXME: regex doesn't respect extension tags or nowiki
3757                 #  => Move this logic to braceSubstitution()
3758                 $substWord = MagicWord::get( 'subst' );
3759                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3760                 $substText = '{{' . $substWord->getSynonym( 0 );
3761
3762                 $text = preg_replace( $substRegex, $substText, $text );
3763                 $text = $this->cleanSigInSig( $text );
3764                 $dom = $this->preprocessToDom( $text );
3765                 $frame = $this->getPreprocessor()->newFrame();
3766                 $text = $frame->expand( $dom );
3767
3768                 if ( !$parsing ) {
3769                         $text = $this->mStripState->unstripBoth( $text );
3770                 }
3771
3772                 return $text;
3773         }
3774
3775         /**
3776          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
3777          * @param string $text
3778          * @return string Signature text with /~{3,5}/ removed
3779          */
3780         function cleanSigInSig( $text ) {
3781                 $text = preg_replace( '/~{3,5}/', '', $text );
3782                 return $text;
3783         }
3784
3785         /**
3786          * Set up some variables which are usually set up in parse()
3787          * so that an external function can call some class members with confidence
3788          * @public
3789          */
3790         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
3791                 $this->setTitle( $title );
3792                 $this->mOptions = $options;
3793                 $this->setOutputType( $outputType );
3794                 if ( $clearState ) {
3795                         $this->clearState();
3796                 }
3797         }
3798
3799         /**
3800          * Wrapper for preprocess()
3801          *
3802          * @param string $text the text to preprocess
3803          * @param ParserOptions $options  options
3804          * @return string
3805          * @public
3806          */
3807         function transformMsg( $text, $options ) {
3808                 global $wgTitle;
3809                 static $executing = false;
3810
3811                 $fname = "Parser::transformMsg";
3812
3813                 # Guard against infinite recursion
3814                 if ( $executing ) {
3815                         return $text;
3816                 }
3817                 $executing = true;
3818
3819                 wfProfileIn($fname);
3820                 $text = $this->preprocess( $text, $wgTitle, $options );
3821
3822                 $executing = false;
3823                 wfProfileOut($fname);
3824                 return $text;
3825         }
3826
3827         /**
3828          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
3829          * The callback should have the following form:
3830          *    function myParserHook( $text, $params, &$parser ) { ... }
3831          *
3832          * Transform and return $text. Use $parser for any required context, e.g. use
3833          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
3834          *
3835          * @public
3836          *
3837          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
3838          * @param mixed $callback The callback function (and object) to use for the tag
3839          *
3840          * @return The old value of the mTagHooks array associated with the hook
3841          */
3842         function setHook( $tag, $callback ) {
3843                 $tag = strtolower( $tag );
3844                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
3845                 $this->mTagHooks[$tag] = $callback;
3846                 if( !in_array( $tag, $this->mStripList ) ) {
3847                         $this->mStripList[] = $tag;
3848                 }
3849
3850                 return $oldVal;
3851         }
3852
3853         function setTransparentTagHook( $tag, $callback ) {
3854                 $tag = strtolower( $tag );
3855                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
3856                 $this->mTransparentTagHooks[$tag] = $callback;
3857
3858                 return $oldVal;
3859         }
3860
3861         /**
3862          * Remove all tag hooks
3863          */
3864         function clearTagHooks() {
3865                 $this->mTagHooks = array();
3866                 $this->mStripList = $this->mDefaultStripList;
3867         }
3868
3869         /**
3870          * Create a function, e.g. {{sum:1|2|3}}
3871          * The callback function should have the form:
3872          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
3873          *
3874          * The callback may either return the text result of the function, or an array with the text
3875          * in element 0, and a number of flags in the other elements. The names of the flags are
3876          * specified in the keys. Valid flags are:
3877          *   found                     The text returned is valid, stop processing the template. This
3878          *                             is on by default.
3879          *   nowiki                    Wiki markup in the return value should be escaped
3880          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
3881          *
3882          * @public
3883          *
3884          * @param string $id The magic word ID
3885          * @param mixed $callback The callback function (and object) to use
3886          * @param integer $flags a combination of the following flags:
3887          *                SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
3888          *
3889          * @return The old callback function for this name, if any
3890          */
3891         function setFunctionHook( $id, $callback, $flags = 0 ) {
3892                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
3893                 $this->mFunctionHooks[$id] = array( $callback, $flags );
3894
3895                 # Add to function cache
3896                 $mw = MagicWord::get( $id );
3897                 if( !$mw )
3898                         throw new MWException( 'Parser::setFunctionHook() expecting a magic word identifier.' );
3899
3900                 $synonyms = $mw->getSynonyms();
3901                 $sensitive = intval( $mw->isCaseSensitive() );
3902
3903                 foreach ( $synonyms as $syn ) {
3904                         # Case
3905                         if ( !$sensitive ) {
3906                                 $syn = strtolower( $syn );
3907                         }
3908                         # Add leading hash
3909                         if ( !( $flags & SFH_NO_HASH ) ) {
3910                                 $syn = '#' . $syn;
3911                         }
3912                         # Remove trailing colon
3913                         if ( substr( $syn, -1, 1 ) == ':' ) {
3914                                 $syn = substr( $syn, 0, -1 );
3915                         }
3916                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
3917                 }
3918                 return $oldVal;
3919         }
3920
3921         /**
3922          * Get all registered function hook identifiers
3923          *
3924          * @return array
3925          */
3926         function getFunctionHooks() {
3927                 return array_keys( $this->mFunctionHooks );
3928         }
3929
3930         /**
3931          * Replace <!--LINK--> link placeholders with actual links, in the buffer
3932          * Placeholders created in Skin::makeLinkObj()
3933          * Returns an array of link CSS classes, indexed by PDBK.
3934          * $options is a bit field, RLH_FOR_UPDATE to select for update
3935          */
3936         function replaceLinkHolders( &$text, $options = 0 ) {
3937                 global $wgUser;
3938                 global $wgContLang;
3939
3940                 $fname = 'Parser::replaceLinkHolders';
3941                 wfProfileIn( $fname );
3942
3943                 $pdbks = array();
3944                 $colours = array();
3945                 $linkcolour_ids = array();
3946                 $sk = $this->mOptions->getSkin();
3947                 $linkCache =& LinkCache::singleton();
3948
3949                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
3950                         wfProfileIn( $fname.'-check' );
3951                         $dbr = wfGetDB( DB_SLAVE );
3952                         $page = $dbr->tableName( 'page' );
3953                         $threshold = $wgUser->getOption('stubthreshold');
3954
3955                         # Sort by namespace
3956                         asort( $this->mLinkHolders['namespaces'] );
3957
3958                         # Generate query
3959                         $query = false;
3960                         $current = null;
3961                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3962                                 # Make title object
3963                                 $title = $this->mLinkHolders['titles'][$key];
3964
3965                                 # Skip invalid entries.
3966                                 # Result will be ugly, but prevents crash.
3967                                 if ( is_null( $title ) ) {
3968                                         continue;
3969                                 }
3970                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
3971
3972                                 # Check if it's a static known link, e.g. interwiki
3973                                 if ( $title->isAlwaysKnown() ) {
3974                                         $colours[$pdbk] = '';
3975                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
3976                                         $colours[$pdbk] = '';
3977                                         $this->mOutput->addLink( $title, $id );
3978                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
3979                                         $colours[$pdbk] = 'new';
3980                                 } elseif ( $title->getNamespace() == NS_SPECIAL && !SpecialPage::exists( $pdbk ) ) {
3981                                         $colours[$pdbk] = 'new';
3982                                 } else {
3983                                         # Not in the link cache, add it to the query
3984                                         if ( !isset( $current ) ) {
3985                                                 $current = $ns;
3986                                                 $query =  "SELECT page_id, page_namespace, page_title, page_is_redirect";
3987                                                 if ( $threshold > 0 ) {
3988                                                         $query .= ', page_len';
3989                                                 }
3990                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
3991                                         } elseif ( $current != $ns ) {
3992                                                 $current = $ns;
3993                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
3994                                         } else {
3995                                                 $query .= ', ';
3996                                         }
3997
3998                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
3999                                 }
4000                         }
4001                         if ( $query ) {
4002                                 $query .= '))';
4003                                 if ( $options & RLH_FOR_UPDATE ) {
4004                                         $query .= ' FOR UPDATE';
4005                                 }
4006
4007                                 $res = $dbr->query( $query, $fname );
4008
4009                                 # Fetch data and form into an associative array
4010                                 # non-existent = broken
4011                                 while ( $s = $dbr->fetchObject($res) ) {
4012                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
4013                                         $pdbk = $title->getPrefixedDBkey();
4014                                         $linkCache->addGoodLinkObj( $s->page_id, $title );
4015                                         $this->mOutput->addLink( $title, $s->page_id );
4016                                         $colours[$pdbk] = $sk->getLinkColour( $s, $threshold );
4017                                         //add id to the extension todolist
4018                                         $linkcolour_ids[$s->page_id] = $pdbk;
4019                                 }
4020                                 //pass an array of page_ids to an extension
4021                                 wfRunHooks( 'GetLinkColours', array( $linkcolour_ids, &$colours ) );
4022                         }
4023                         wfProfileOut( $fname.'-check' );
4024
4025                         # Do a second query for different language variants of links and categories
4026                         if($wgContLang->hasVariants()){
4027                                 $linkBatch = new LinkBatch();
4028                                 $variantMap = array(); // maps $pdbkey_Variant => $keys (of link holders)
4029                                 $categoryMap = array(); // maps $category_variant => $category (dbkeys)
4030                                 $varCategories = array(); // category replacements oldDBkey => newDBkey
4031
4032                                 $categories = $this->mOutput->getCategoryLinks();
4033
4034                                 // Add variants of links to link batch
4035                                 foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4036                                         $title = $this->mLinkHolders['titles'][$key];
4037                                         if ( is_null( $title ) )
4038                                                 continue;
4039
4040                                         $pdbk = $title->getPrefixedDBkey();
4041                                         $titleText = $title->getText();
4042
4043                                         // generate all variants of the link title text
4044                                         $allTextVariants = $wgContLang->convertLinkToAllVariants($titleText);
4045
4046                                         // if link was not found (in first query), add all variants to query
4047                                         if ( !isset($colours[$pdbk]) ){
4048                                                 foreach($allTextVariants as $textVariant){
4049                                                         if($textVariant != $titleText){
4050                                                                 $variantTitle = Title::makeTitle( $ns, $textVariant );
4051                                                                 if(is_null($variantTitle)) continue;
4052                                                                 $linkBatch->addObj( $variantTitle );
4053                                                                 $variantMap[$variantTitle->getPrefixedDBkey()][] = $key;
4054                                                         }
4055                                                 }
4056                                         }
4057                                 }
4058
4059                                 // process categories, check if a category exists in some variant
4060                                 foreach( $categories as $category ){
4061                                         $variants = $wgContLang->convertLinkToAllVariants($category);
4062                                         foreach($variants as $variant){
4063                                                 if($variant != $category){
4064                                                         $variantTitle = Title::newFromDBkey( Title::makeName(NS_CATEGORY,$variant) );
4065                                                         if(is_null($variantTitle)) continue;
4066                                                         $linkBatch->addObj( $variantTitle );
4067                                                         $categoryMap[$variant] = $category;
4068                                                 }
4069                                         }
4070                                 }
4071
4072
4073                                 if(!$linkBatch->isEmpty()){
4074                                         // construct query
4075                                         $titleClause = $linkBatch->constructSet('page', $dbr);
4076
4077                                         $variantQuery =  "SELECT page_id, page_namespace, page_title, page_is_redirect";
4078                                         if ( $threshold > 0 ) {
4079                                                 $variantQuery .= ', page_len';
4080                                         }
4081
4082                                         $variantQuery .= " FROM $page WHERE $titleClause";
4083                                         if ( $options & RLH_FOR_UPDATE ) {
4084                                                 $variantQuery .= ' FOR UPDATE';
4085                                         }
4086
4087                                         $varRes = $dbr->query( $variantQuery, $fname );
4088
4089                                         // for each found variants, figure out link holders and replace
4090                                         while ( $s = $dbr->fetchObject($varRes) ) {
4091
4092                                                 $variantTitle = Title::makeTitle( $s->page_namespace, $s->page_title );
4093                                                 $varPdbk = $variantTitle->getPrefixedDBkey();
4094                                                 $vardbk = $variantTitle->getDBkey();
4095
4096                                                 $holderKeys = array();
4097                                                 if(isset($variantMap[$varPdbk])){
4098                                                         $holderKeys = $variantMap[$varPdbk];
4099                                                         $linkCache->addGoodLinkObj( $s->page_id, $variantTitle );
4100                                                         $this->mOutput->addLink( $variantTitle, $s->page_id );
4101                                                 }
4102
4103                                                 // loop over link holders
4104                                                 foreach($holderKeys as $key){
4105                                                         $title = $this->mLinkHolders['titles'][$key];
4106                                                         if ( is_null( $title ) ) continue;
4107
4108                                                         $pdbk = $title->getPrefixedDBkey();
4109
4110                                                         if(!isset($colours[$pdbk])){
4111                                                                 // found link in some of the variants, replace the link holder data
4112                                                                 $this->mLinkHolders['titles'][$key] = $variantTitle;
4113                                                                 $this->mLinkHolders['dbkeys'][$key] = $variantTitle->getDBkey();
4114
4115                                                                 // set pdbk and colour
4116                                                                 $pdbks[$key] = $varPdbk;
4117                                                                 $colours[$varPdbk] = $sk->getLinkColour( $s, $threshold );
4118                                                                 $linkcolour_ids[$s->page_id] = $pdbk;
4119                                                         }
4120                                                         wfRunHooks( 'GetLinkColours', array( $linkcolour_ids, &$colours ) );
4121                                                 }
4122
4123                                                 // check if the object is a variant of a category
4124                                                 if(isset($categoryMap[$vardbk])){
4125                                                         $oldkey = $categoryMap[$vardbk];
4126                                                         if($oldkey != $vardbk)
4127                                                                 $varCategories[$oldkey]=$vardbk;
4128                                                 }
4129                                         }
4130
4131                                         // rebuild the categories in original order (if there are replacements)
4132                                         if(count($varCategories)>0){
4133                                                 $newCats = array();
4134                                                 $originalCats = $this->mOutput->getCategories();
4135                                                 foreach($originalCats as $cat => $sortkey){
4136                                                         // make the replacement
4137                                                         if( array_key_exists($cat,$varCategories) )
4138                                                                 $newCats[$varCategories[$cat]] = $sortkey;
4139                                                         else $newCats[$cat] = $sortkey;
4140                                                 }
4141                                                 $this->mOutput->setCategoryLinks($newCats);
4142                                         }
4143                                 }
4144                         }
4145
4146                         # Construct search and replace arrays
4147                         wfProfileIn( $fname.'-construct' );
4148                         $replacePairs = array();
4149                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4150                                 $pdbk = $pdbks[$key];
4151                                 $searchkey = "<!--LINK $key-->";
4152                                 $title = $this->mLinkHolders['titles'][$key];
4153                                 if ( !isset( $colours[$pdbk] ) || $colours[$pdbk] == 'new' ) {
4154                                         $linkCache->addBadLinkObj( $title );
4155                                         $colours[$pdbk] = 'new';
4156                                         $this->mOutput->addLink( $title, 0 );
4157                                         $replacePairs[$searchkey] = $sk->makeBrokenLinkObj( $title,
4158                                                                         $this->mLinkHolders['texts'][$key],
4159                                                                         $this->mLinkHolders['queries'][$key] );
4160                                 } else {
4161                                         $replacePairs[$searchkey] = $sk->makeColouredLinkObj( $title, $colours[$pdbk],
4162                                                                         $this->mLinkHolders['texts'][$key],
4163                                                                         $this->mLinkHolders['queries'][$key] );
4164                                 }
4165                         }
4166                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4167                         wfProfileOut( $fname.'-construct' );
4168
4169                         # Do the thing
4170                         wfProfileIn( $fname.'-replace' );
4171                         $text = preg_replace_callback(
4172                                 '/(<!--LINK .*?-->)/',
4173                                 $replacer->cb(),
4174                                 $text);
4175
4176                         wfProfileOut( $fname.'-replace' );
4177                 }
4178
4179                 # Now process interwiki link holders
4180                 # This is quite a bit simpler than internal links
4181                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
4182                         wfProfileIn( $fname.'-interwiki' );
4183                         # Make interwiki link HTML
4184                         $replacePairs = array();
4185                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
4186                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
4187                                 $replacePairs[$key] = $sk->makeLinkObj( $title, $link );
4188                         }
4189                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4190
4191                         $text = preg_replace_callback(
4192                                 '/<!--IWLINK (.*?)-->/',
4193                                 $replacer->cb(),
4194                                 $text );
4195                         wfProfileOut( $fname.'-interwiki' );
4196                 }
4197
4198                 wfProfileOut( $fname );
4199                 return $colours;
4200         }
4201
4202         /**
4203          * Replace <!--LINK--> link placeholders with plain text of links
4204          * (not HTML-formatted).
4205          * @param string $text
4206          * @return string
4207          */
4208         function replaceLinkHoldersText( $text ) {
4209                 $fname = 'Parser::replaceLinkHoldersText';
4210                 wfProfileIn( $fname );
4211
4212                 $text = preg_replace_callback(
4213                         '/<!--(LINK|IWLINK) (.*?)-->/',
4214                         array( &$this, 'replaceLinkHoldersTextCallback' ),
4215                         $text );
4216
4217                 wfProfileOut( $fname );
4218                 return $text;
4219         }
4220
4221         /**
4222          * @param array $matches
4223          * @return string
4224          * @private
4225          */
4226         function replaceLinkHoldersTextCallback( $matches ) {
4227                 $type = $matches[1];
4228                 $key  = $matches[2];
4229                 if( $type == 'LINK' ) {
4230                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
4231                                 return $this->mLinkHolders['texts'][$key];
4232                         }
4233                 } elseif( $type == 'IWLINK' ) {
4234                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
4235                                 return $this->mInterwikiLinkHolders['texts'][$key];
4236                         }
4237                 }
4238                 return $matches[0];
4239         }
4240
4241         /**
4242          * Tag hook handler for 'pre'.
4243          */
4244         function renderPreTag( $text, $attribs ) {
4245                 // Backwards-compatibility hack
4246                 $content = StringUtils::delimiterReplace( '<nowiki>', '</nowiki>', '$1', $text, 'i' );
4247
4248                 $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
4249                 return wfOpenElement( 'pre', $attribs ) .
4250                         Xml::escapeTagsOnly( $content ) .
4251                         '</pre>';
4252         }
4253
4254         /**
4255          * Renders an image gallery from a text with one line per image.
4256          * text labels may be given by using |-style alternative text. E.g.
4257          *   Image:one.jpg|The number "1"
4258          *   Image:tree.jpg|A tree
4259          * given as text will return the HTML of a gallery with two images,
4260          * labeled 'The number "1"' and
4261          * 'A tree'.
4262          */
4263         function renderImageGallery( $text, $params ) {
4264                 $ig = new ImageGallery();
4265                 $ig->setContextTitle( $this->mTitle );
4266                 $ig->setShowBytes( false );
4267                 $ig->setShowFilename( false );
4268                 $ig->setParser( $this );
4269                 $ig->setHideBadImages();
4270                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4271                 $ig->useSkin( $this->mOptions->getSkin() );
4272                 $ig->mRevisionId = $this->mRevisionId;
4273
4274                 if( isset( $params['caption'] ) ) {
4275                         $caption = $params['caption'];
4276                         $caption = htmlspecialchars( $caption );
4277                         $caption = $this->replaceInternalLinks( $caption );
4278                         $ig->setCaptionHtml( $caption );
4279                 }
4280                 if( isset( $params['perrow'] ) ) {
4281                         $ig->setPerRow( $params['perrow'] );
4282                 }
4283                 if( isset( $params['widths'] ) ) {
4284                         $ig->setWidths( $params['widths'] );
4285                 }
4286                 if( isset( $params['heights'] ) ) {
4287                         $ig->setHeights( $params['heights'] );
4288                 }
4289
4290                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4291
4292                 $lines = explode( "\n", $text );
4293                 foreach ( $lines as $line ) {
4294                         # match lines like these:
4295                         # Image:someimage.jpg|This is some image
4296                         $matches = array();
4297                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4298                         # Skip empty lines
4299                         if ( count( $matches ) == 0 ) {
4300                                 continue;
4301                         }
4302                         $tp = Title::newFromText( $matches[1] );
4303                         $nt =& $tp;
4304                         if( is_null( $nt ) ) {
4305                                 # Bogus title. Ignore these so we don't bomb out later.
4306                                 continue;
4307                         }
4308                         if ( isset( $matches[3] ) ) {
4309                                 $label = $matches[3];
4310                         } else {
4311                                 $label = '';
4312                         }
4313
4314                         $html = $this->recursiveTagParse( trim( $label ) );
4315
4316                         $ig->add( $nt, $html );
4317
4318                         # Only add real images (bug #5586)
4319                         if ( $nt->getNamespace() == NS_IMAGE ) {
4320                                 $this->mOutput->addImage( $nt->getDBkey() );
4321                         }
4322                 }
4323                 return $ig->toHTML();
4324         }
4325
4326         function getImageParams( $handler ) {
4327                 if ( $handler ) {
4328                         $handlerClass = get_class( $handler );
4329                 } else {
4330                         $handlerClass = '';
4331                 }
4332                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4333                         // Initialise static lists
4334                         static $internalParamNames = array(
4335                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4336                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4337                                         'bottom', 'text-bottom' ),
4338                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4339                                         'upright', 'border' ),
4340                         );
4341                         static $internalParamMap;
4342                         if ( !$internalParamMap ) {
4343                                 $internalParamMap = array();
4344                                 foreach ( $internalParamNames as $type => $names ) {
4345                                         foreach ( $names as $name ) {
4346                                                 $magicName = str_replace( '-', '_', "img_$name" );
4347                                                 $internalParamMap[$magicName] = array( $type, $name );
4348                                         }
4349                                 }
4350                         }
4351
4352                         // Add handler params
4353                         $paramMap = $internalParamMap;
4354                         if ( $handler ) {
4355                                 $handlerParamMap = $handler->getParamMap();
4356                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4357                                         $paramMap[$magic] = array( 'handler', $paramName );
4358                                 }
4359                         }
4360                         $this->mImageParams[$handlerClass] = $paramMap;
4361                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4362                 }
4363                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4364         }
4365
4366         /**
4367          * Parse image options text and use it to make an image
4368          */
4369         function makeImage( $title, $options ) {
4370                 # @TODO: let the MediaHandler specify its transform parameters
4371                 #
4372                 # Check if the options text is of the form "options|alt text"
4373                 # Options are:
4374                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
4375                 #  * left               no resizing, just left align. label is used for alt= only
4376                 #  * right              same, but right aligned
4377                 #  * none               same, but not aligned
4378                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
4379                 #  * center             center the image
4380                 #  * framed             Keep original image size, no magnify-button.
4381                 #  * frameless          like 'thumb' but without a frame. Keeps user preferences for width
4382                 #  * upright            reduce width for upright images, rounded to full __0 px
4383                 #  * border             draw a 1px border around the image
4384                 # vertical-align values (no % or length right now):
4385                 #  * baseline
4386                 #  * sub
4387                 #  * super
4388                 #  * top
4389                 #  * text-top
4390                 #  * middle
4391                 #  * bottom
4392                 #  * text-bottom
4393
4394                 $parts = array_map( 'trim', explode( '|', $options) );
4395                 $sk = $this->mOptions->getSkin();
4396
4397                 # Give extensions a chance to select the file revision for us
4398                 $skip = $time = false;
4399                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time ) );
4400
4401                 if ( $skip ) {
4402                         return $sk->makeLinkObj( $title );
4403                 }
4404
4405                 # Get parameter map
4406                 $file = wfFindFile( $title, $time );
4407                 $handler = $file ? $file->getHandler() : false;
4408
4409                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4410
4411                 # Process the input parameters
4412                 $caption = '';
4413                 $params = array( 'frame' => array(), 'handler' => array(),
4414                         'horizAlign' => array(), 'vertAlign' => array() );
4415                 foreach( $parts as $part ) {
4416                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4417                         if ( isset( $paramMap[$magicName] ) ) {
4418                                 list( $type, $paramName ) = $paramMap[$magicName];
4419                                 $params[$type][$paramName] = $value;
4420
4421                                 // Special case; width and height come in one variable together
4422                                 if( $type == 'handler' && $paramName == 'width' ) {
4423                                         $m = array();
4424                                         if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $value, $m ) ) {
4425                                                 $params[$type]['width'] = intval( $m[1] );
4426                                                 $params[$type]['height'] = intval( $m[2] );
4427                                         } else {
4428                                                 $params[$type]['width'] = intval( $value );
4429                                         }
4430                                 }
4431                         } else {
4432                                 $caption = $part;
4433                         }
4434                 }
4435
4436                 # Process alignment parameters
4437                 if ( $params['horizAlign'] ) {
4438                         $params['frame']['align'] = key( $params['horizAlign'] );
4439                 }
4440                 if ( $params['vertAlign'] ) {
4441                         $params['frame']['valign'] = key( $params['vertAlign'] );
4442                 }
4443
4444                 # Validate the handler parameters
4445                 if ( $handler ) {
4446                         foreach ( $params['handler'] as $name => $value ) {
4447                                 if ( !$handler->validateParam( $name, $value ) ) {
4448                                         unset( $params['handler'][$name] );
4449                                 }
4450                         }
4451                 }
4452
4453                 # Strip bad stuff out of the alt text
4454                 $alt = $this->replaceLinkHoldersText( $caption );
4455
4456                 # make sure there are no placeholders in thumbnail attributes
4457                 # that are later expanded to html- so expand them now and
4458                 # remove the tags
4459                 $alt = $this->mStripState->unstripBoth( $alt );
4460                 $alt = Sanitizer::stripAllTags( $alt );
4461
4462                 $params['frame']['alt'] = $alt;
4463                 $params['frame']['caption'] = $caption;
4464
4465                 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params ) );
4466
4467                 # Linker does the rest
4468                 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'], $time );
4469
4470                 # Give the handler a chance to modify the parser object
4471                 if ( $handler ) {
4472                         $handler->parserTransformHook( $this, $file );
4473                 }
4474
4475                 return $ret;
4476         }
4477
4478         /**
4479          * Set a flag in the output object indicating that the content is dynamic and
4480          * shouldn't be cached.
4481          */
4482         function disableCache() {
4483                 wfDebug( "Parser output marked as uncacheable.\n" );
4484                 $this->mOutput->mCacheTime = -1;
4485         }
4486
4487         /**#@+
4488          * Callback from the Sanitizer for expanding items found in HTML attribute
4489          * values, so they can be safely tested and escaped.
4490          * @param string $text
4491          * @param PPFrame $frame
4492          * @return string
4493          * @private
4494          */
4495         function attributeStripCallback( &$text, $frame = false ) {
4496                 $text = $this->replaceVariables( $text, $frame );
4497                 $text = $this->mStripState->unstripBoth( $text );
4498                 return $text;
4499         }
4500
4501         /**#@-*/
4502
4503         /**#@+
4504          * Accessor/mutator
4505          */
4506         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4507         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4508         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4509         /**#@-*/
4510
4511         /**#@+
4512          * Accessor
4513          */
4514         function getTags() { return array_merge( array_keys($this->mTransparentTagHooks), array_keys( $this->mTagHooks ) ); }
4515         /**#@-*/
4516
4517
4518         /**
4519          * Break wikitext input into sections, and either pull or replace
4520          * some particular section's text.
4521          *
4522          * External callers should use the getSection and replaceSection methods.
4523          *
4524          * @param string $text Page wikitext
4525          * @param string $section A section identifier string of the form:
4526          *   <flag1> - <flag2> - ... - <section number>
4527          *
4528          * Currently the only recognised flag is "T", which means the target section number
4529          * was derived during a template inclusion parse, in other words this is a template
4530          * section edit link. If no flags are given, it was an ordinary section edit link.
4531          * This flag is required to avoid a section numbering mismatch when a section is
4532          * enclosed by <includeonly> (bug 6563).
4533          *
4534          * The section number 0 pulls the text before the first heading; other numbers will
4535          * pull the given section along with its lower-level subsections. If the section is
4536          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
4537          *
4538          * @param string $mode One of "get" or "replace"
4539          * @param string $newText Replacement text for section data.
4540          * @return string for "get", the extracted section text.
4541          *                for "replace", the whole page with the section replaced.
4542          */
4543         private function extractSections( $text, $section, $mode, $newText='' ) {
4544                 global $wgTitle;
4545                 $this->clearState();
4546                 $this->setTitle( $wgTitle ); // not generally used but removes an ugly failure mode
4547                 $this->mOptions = new ParserOptions;
4548                 $this->setOutputType( self::OT_WIKI );
4549                 $outText = '';
4550                 $frame = $this->getPreprocessor()->newFrame();
4551
4552                 // Process section extraction flags
4553                 $flags = 0;
4554                 $sectionParts = explode( '-', $section );
4555                 $sectionIndex = array_pop( $sectionParts );
4556                 foreach ( $sectionParts as $part ) {
4557                         if ( $part == 'T' ) {
4558                                 $flags |= self::PTD_FOR_INCLUSION;
4559                         }
4560                 }
4561                 // Preprocess the text
4562                 $root = $this->preprocessToDom( $text, $flags );
4563
4564                 // <h> nodes indicate section breaks
4565                 // They can only occur at the top level, so we can find them by iterating the root's children
4566                 $node = $root->getFirstChild();
4567
4568                 // Find the target section
4569                 if ( $sectionIndex == 0 ) {
4570                         // Section zero doesn't nest, level=big
4571                         $targetLevel = 1000;
4572                 } else {
4573             while ( $node ) {
4574                 if ( $node->getName() == 'h' ) {
4575                     $bits = $node->splitHeading();
4576                                         if ( $bits['i'] == $sectionIndex ) {
4577                                         $targetLevel = $bits['level'];
4578                                                 break;
4579                                         }
4580                                 }
4581                                 if ( $mode == 'replace' ) {
4582                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4583                                 }
4584                                 $node = $node->getNextSibling();
4585                         }
4586                 }
4587
4588                 if ( !$node ) {
4589                         // Not found
4590                         if ( $mode == 'get' ) {
4591                                 return $newText;
4592                         } else {
4593                                 return $text;
4594                         }
4595                 }
4596
4597                 // Find the end of the section, including nested sections
4598                 do {
4599                         if ( $node->getName() == 'h' ) {
4600                                 $bits = $node->splitHeading();
4601                                 $curLevel = $bits['level'];
4602                                 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
4603                                         break;
4604                                 }
4605                         }
4606                         if ( $mode == 'get' ) {
4607                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4608                         }
4609                         $node = $node->getNextSibling();
4610                 } while ( $node );
4611
4612                 // Write out the remainder (in replace mode only)
4613                 if ( $mode == 'replace' ) {
4614                         // Output the replacement text
4615                         // Add two newlines on -- trailing whitespace in $newText is conventionally
4616                         // stripped by the editor, so we need both newlines to restore the paragraph gap
4617                         $outText .= $newText . "\n\n";
4618                         while ( $node ) {
4619                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4620                                 $node = $node->getNextSibling();
4621                         }
4622                 }
4623
4624                 if ( is_string( $outText ) ) {
4625                         // Re-insert stripped tags
4626                         $outText = trim( $this->mStripState->unstripBoth( $outText ) );
4627                 }
4628
4629                 return $outText;
4630         }
4631
4632         /**
4633          * This function returns the text of a section, specified by a number ($section).
4634          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4635          * the first section before any such heading (section 0).
4636          *
4637          * If a section contains subsections, these are also returned.
4638          *
4639          * @param string $text text to look in
4640          * @param string $section section identifier
4641          * @param string $deftext default to return if section is not found
4642          * @return string text of the requested section
4643          */
4644         public function getSection( $text, $section, $deftext='' ) {
4645                 return $this->extractSections( $text, $section, "get", $deftext );
4646         }
4647
4648         public function replaceSection( $oldtext, $section, $text ) {
4649                 return $this->extractSections( $oldtext, $section, "replace", $text );
4650         }
4651
4652         /**
4653          * Get the timestamp associated with the current revision, adjusted for
4654          * the default server-local timestamp
4655          */
4656         function getRevisionTimestamp() {
4657                 if ( is_null( $this->mRevisionTimestamp ) ) {
4658                         wfProfileIn( __METHOD__ );
4659                         global $wgContLang;
4660                         $dbr = wfGetDB( DB_SLAVE );
4661                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
4662                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
4663
4664                         // Normalize timestamp to internal MW format for timezone processing.
4665                         // This has the added side-effect of replacing a null value with
4666                         // the current time, which gives us more sensible behavior for
4667                         // previews.
4668                         $timestamp = wfTimestamp( TS_MW, $timestamp );
4669
4670                         // The cryptic '' timezone parameter tells to use the site-default
4671                         // timezone offset instead of the user settings.
4672                         //
4673                         // Since this value will be saved into the parser cache, served
4674                         // to other users, and potentially even used inside links and such,
4675                         // it needs to be consistent for all visitors.
4676                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
4677
4678                         wfProfileOut( __METHOD__ );
4679                 }
4680                 return $this->mRevisionTimestamp;
4681         }
4682
4683         /**
4684          * Mutator for $mDefaultSort
4685          *
4686          * @param $sort New value
4687          */
4688         public function setDefaultSort( $sort ) {
4689                 $this->mDefaultSort = $sort;
4690         }
4691
4692         /**
4693          * Accessor for $mDefaultSort
4694          * Will use the title/prefixed title if none is set
4695          *
4696          * @return string
4697          */
4698         public function getDefaultSort() {
4699                 if( $this->mDefaultSort !== false ) {
4700                         return $this->mDefaultSort;
4701                 } else {
4702                         return $this->mTitle->getNamespace() == NS_CATEGORY
4703                                         ? $this->mTitle->getText()
4704                                         : $this->mTitle->getPrefixedText();
4705                 }
4706         }
4707
4708         /**
4709          * Try to guess the section anchor name based on a wikitext fragment
4710          * presumably extracted from a heading, for example "Header" from
4711          * "== Header ==".
4712          */
4713         public function guessSectionNameFromWikiText( $text ) {
4714                 # Strip out wikitext links(they break the anchor)
4715                 $text = $this->stripSectionName( $text );
4716                 $headline = Sanitizer::decodeCharReferences( $text );
4717                 # strip out HTML
4718                 $headline = StringUtils::delimiterReplace( '<', '>', '', $headline );
4719                 $headline = trim( $headline );
4720                 $sectionanchor = '#' . urlencode( str_replace( ' ', '_', $headline ) );
4721                 $replacearray = array(
4722                         '%3A' => ':',
4723                         '%' => '.'
4724                 );
4725                 return str_replace(
4726                         array_keys( $replacearray ),
4727                         array_values( $replacearray ),
4728                         $sectionanchor );
4729         }
4730
4731         /**
4732          * Strips a text string of wikitext for use in a section anchor
4733          *
4734          * Accepts a text string and then removes all wikitext from the
4735          * string and leaves only the resultant text (i.e. the result of
4736          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
4737          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
4738          * to create valid section anchors by mimicing the output of the
4739          * parser when headings are parsed.
4740          *
4741          * @param $text string Text string to be stripped of wikitext
4742          * for use in a Section anchor
4743          * @return Filtered text string
4744          */
4745         public function stripSectionName( $text ) {
4746                 # Strip internal link markup
4747                 $text = preg_replace('/\[\[:?([^[|]+)\|([^[]+)\]\]/','$2',$text);
4748                 $text = preg_replace('/\[\[:?([^[]+)\|?\]\]/','$1',$text);
4749
4750                 # Strip external link markup (FIXME: Not Tolerant to blank link text
4751                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
4752                 # on how many empty links there are on the page - need to figure that out.
4753                 $text = preg_replace('/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/','$2',$text);
4754
4755                 # Parse wikitext quotes (italics & bold)
4756                 $text = $this->doQuotes($text);
4757
4758                 # Strip HTML tags
4759                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
4760                 return $text;
4761         }
4762
4763         function srvus( $text ) {
4764                 return $this->testSrvus( $text, $this->mOutputType );
4765         }
4766
4767         /**
4768          * strip/replaceVariables/unstrip for preprocessor regression testing
4769          */
4770         function testSrvus( $text, $title, $options, $outputType = self::OT_HTML ) {
4771                 $this->clearState();
4772                 if ( ! ( $title instanceof Title ) ) {
4773                         $title = Title::newFromText( $title );
4774                 }
4775                 $this->mTitle = $title;
4776                 $this->mOptions = $options;
4777                 $this->setOutputType( $outputType );
4778                 $text = $this->replaceVariables( $text );
4779                 $text = $this->mStripState->unstripBoth( $text );
4780                 $text = Sanitizer::removeHTMLtags( $text );
4781                 return $text;
4782         }
4783
4784         function testPst( $text, $title, $options ) {
4785                 global $wgUser;
4786                 if ( ! ( $title instanceof Title ) ) {
4787                         $title = Title::newFromText( $title );
4788                 }
4789                 return $this->preSaveTransform( $text, $title, $wgUser, $options );
4790         }
4791
4792         function testPreprocess( $text, $title, $options ) {
4793                 if ( ! ( $title instanceof Title ) ) {
4794                         $title = Title::newFromText( $title );
4795                 }
4796                 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
4797         }
4798
4799         function markerSkipCallback( $s, $callback ) {
4800                 $i = 0;
4801                 $out = '';
4802                 while ( $i < strlen( $s ) ) {
4803                         $markerStart = strpos( $s, $this->mUniqPrefix, $i );
4804                         if ( $markerStart === false ) {
4805                                 $out .= call_user_func( $callback, substr( $s, $i ) );
4806                                 break;
4807                         } else {
4808                                 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
4809                                 $markerEnd = strpos( $s, $this->mMarkerSuffix, $markerStart );
4810                                 if ( $markerEnd === false ) {
4811                                         $out .= substr( $s, $markerStart );
4812                                         break;
4813                                 } else {
4814                                         $markerEnd += strlen( $this->mMarkerSuffix );
4815                                         $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
4816                                         $i = $markerEnd;
4817                                 }
4818                         }
4819                 }
4820                 return $out;
4821         }
4822 }
4823
4824 /**
4825  * @todo document, briefly.
4826  * @addtogroup Parser
4827  */
4828 class StripState {
4829         var $general, $nowiki;
4830
4831         function __construct() {
4832                 $this->general = new ReplacementArray;
4833                 $this->nowiki = new ReplacementArray;
4834         }
4835
4836         function unstripGeneral( $text ) {
4837                 wfProfileIn( __METHOD__ );
4838                 do {
4839                         $oldText = $text;
4840                         $text = $this->general->replace( $text );
4841                 } while ( $text != $oldText );
4842                 wfProfileOut( __METHOD__ );
4843                 return $text;
4844         }
4845
4846         function unstripNoWiki( $text ) {
4847                 wfProfileIn( __METHOD__ );
4848                 do {
4849                         $oldText = $text;
4850                         $text = $this->nowiki->replace( $text );
4851                 } while ( $text != $oldText );
4852                 wfProfileOut( __METHOD__ );
4853                 return $text;
4854         }
4855
4856         function unstripBoth( $text ) {
4857                 wfProfileIn( __METHOD__ );
4858                 do {
4859                         $oldText = $text;
4860                         $text = $this->general->replace( $text );
4861                         $text = $this->nowiki->replace( $text );
4862                 } while ( $text != $oldText );
4863                 wfProfileOut( __METHOD__ );
4864                 return $text;
4865         }
4866 }
4867
4868 /**
4869  * @todo document, briefly.
4870  * @addtogroup Parser
4871  */
4872 class OnlyIncludeReplacer {
4873         var $output = '';
4874
4875         function replace( $matches ) {
4876                 if ( substr( $matches[1], -1 ) == "\n" ) {
4877                         $this->output .= substr( $matches[1], 0, -1 );
4878                 } else {
4879                         $this->output .= $matches[1];
4880                 }
4881         }
4882 }
4883