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