includes/parser/Parser.php

   1 <?php
   2 /**
   3  * @defgroup Parser Parser
   4  *
   5  * @file
   6  * @ingroup Parser
   7  * File for Parser and related classes
   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  *  $wgMaxArticleSize*
  40  *
  41  *  * only within ParserOptions
  42  * </pre>
  43  *
  44  * @ingroup Parser
  45  */
  46 class Parser
  47 {
  48         /**
  49          * Update this version number when the ParserOutput format
  50          * changes in an incompatible way, so the parser cache
  51          * can automatically discard old data.
  52          */
  53         const VERSION = '1.6.4';
  54
  55         # Flags for Parser::setFunctionHook
  56         # Also available as global constants from Defines.php
  57         const SFH_NO_HASH = 1;
  58         const SFH_OBJECT_ARGS = 2;
  59
  60         # Constants needed for external link processing
  61         # Everything except bracket, space, or control characters
  62         const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F]';
  63         const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F]+)
  64                 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/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         // Marker Suffix needs to be accessible staticly.
  87         const MARKER_SUFFIX = "-QINU\x7f";
  88
  89         /**#@+
  90          * @private
  91          */
  92         # Persistent:
  93         var $mTagHooks, $mTransparentTagHooks, $mFunctionHooks, $mFunctionSynonyms, $mVariables,
  94                 $mSubsts, $mImageParams, $mImageParamsMagicArray, $mStripList, $mMarkerIndex,
  95                 $mPreprocessor, $mExtLinkBracketedRegex, $mUrlProtocols, $mDefaultStripList,
  96                 $mVarCache, $mConf, $mFunctionTagHooks;
  97
  98
  99         # Cleared with clearState():
 100         var $mOutput, $mAutonumber, $mDTopen, $mStripState;
 101         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 102         var $mLinkHolders, $mLinkID;
 103         var $mIncludeSizes, $mPPNodeCount, $mDefaultSort;
 104         var $mTplExpandCache; // empty-frame expansion cache
 105         var $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
 106         var $mExpensiveFunctionCount; // number of expensive parser function calls
 107
 108         # Temporary
 109         # These are variables reset at least once per parse regardless of $clearState
 110         var $mOptions,      // ParserOptions object
 111                 $mTitle,        // Title context, used for self-link rendering and similar things
 112                 $mOutputType,   // Output type, one of the OT_xxx constants
 113                 $ot,            // Shortcut alias, see setOutputType()
 114                 $mRevisionId,   // ID to display in {{REVISIONID}} tags
 115                 $mRevisionTimestamp, // The timestamp of the specified revision ID
 116                 $mRevIdForTs;   // The revision ID which was used to fetch the timestamp
 117
 118         /**#@-*/
 119
 120         /**
 121          * Constructor
 122          *
 123          * @public
 124          */
 125         function __construct( $conf = array() ) {
 126                 $this->mConf = $conf;
 127                 $this->mTagHooks = array();
 128                 $this->mTransparentTagHooks = array();
 129                 $this->mFunctionHooks = array();
 130                 $this->mFunctionTagHooks = array();
 131                 $this->mFunctionSynonyms = array( 0 => array(), 1 => array() );
 132                 $this->mDefaultStripList = $this->mStripList = array();
 133                 $this->mUrlProtocols = wfUrlProtocols();
 134                 $this->mExtLinkBracketedRegex = '/\[(\b(' . wfUrlProtocols() . ')'.
 135                         '[^][<>"\\x00-\\x20\\x7F]+) *([^\]\\x0a\\x0d]*?)\]/S';
 136                 $this->mVarCache = array();
 137                 if ( isset( $conf['preprocessorClass'] ) ) {
 138                         $this->mPreprocessorClass = $conf['preprocessorClass'];
 139                 } elseif ( extension_loaded( 'domxml' ) ) {
 140                         // PECL extension that conflicts with the core DOM extension (bug 13770)
 141                         wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
 142                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 143                 } elseif ( extension_loaded( 'dom' ) ) {
 144                         $this->mPreprocessorClass = 'Preprocessor_DOM';
 145                 } else {
 146                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 147                 }
 148                 $this->mMarkerIndex = 0;
 149                 $this->mFirstCall = true;
 150         }
 151
 152         /**
 153          * Reduce memory usage to reduce the impact of circular references
 154          */
 155         function __destruct() {
 156                 if ( isset( $this->mLinkHolders ) ) {
 157                         $this->mLinkHolders->__destruct();
 158                 }
 159                 foreach ( $this as $name => $value ) {
 160                         unset( $this->$name );
 161                 }
 162         }
 163
 164         /**
 165          * Do various kinds of initialisation on the first call of the parser
 166          */
 167         function firstCallInit() {
 168                 if ( !$this->mFirstCall ) {
 169                         return;
 170                 }
 171                 $this->mFirstCall = false;
 172
 173                 wfProfileIn( __METHOD__ );
 174
 175                 CoreParserFunctions::register( $this );
 176                 CoreTagHooks::register( $this );
 177                 $this->initialiseVariables();
 178
 179                 wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
 180                 wfProfileOut( __METHOD__ );
 181         }
 182
 183         /**
 184          * Clear Parser state
 185          *
 186          * @private
 187          */
 188         function clearState() {
 189                 wfProfileIn( __METHOD__ );
 190                 if ( $this->mFirstCall ) {
 191                         $this->firstCallInit();
 192                 }
 193                 $this->mOutput = new ParserOutput;
 194                 $this->mAutonumber = 0;
 195                 $this->mLastSection = '';
 196                 $this->mDTopen = false;
 197                 $this->mIncludeCount = array();
 198                 $this->mStripState = new StripState;
 199                 $this->mArgStack = false;
 200                 $this->mInPre = false;
 201                 $this->mLinkHolders = new LinkHolderArray( $this );
 202                 $this->mLinkID = 0;
 203                 $this->mRevisionTimestamp = $this->mRevisionId = null;
 204                 $this->mVarCache = array();
 205
 206                 /**
 207                  * Prefix for temporary replacement strings for the multipass parser.
 208                  * \x07 should never appear in input as it's disallowed in XML.
 209                  * Using it at the front also gives us a little extra robustness
 210                  * since it shouldn't match when butted up against identifier-like
 211                  * string constructs.
 212                  *
 213                  * Must not consist of all title characters, or else it will change
 214                  * the behaviour of <nowiki> in a link.
 215                  */
 216                 #$this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 217                 # Changed to \x7f to allow XML double-parsing -- TS
 218                 $this->mUniqPrefix = "\x7fUNIQ" . self::getRandomString();
 219
 220
 221                 # Clear these on every parse, bug 4549
 222                 $this->mTplExpandCache = $this->mTplRedirCache = $this->mTplDomCache = array();
 223
 224                 $this->mShowToc = true;
 225                 $this->mForceTocPosition = false;
 226                 $this->mIncludeSizes = array(
 227                         'post-expand' => 0,
 228                         'arg' => 0,
 229                 );
 230                 $this->mPPNodeCount = 0;
 231                 $this->mDefaultSort = false;
 232                 $this->mHeadings = array();
 233                 $this->mDoubleUnderscores = array();
 234                 $this->mExpensiveFunctionCount = 0;
 235
 236                 # Fix cloning
 237                 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
 238                         $this->mPreprocessor = null;
 239                 }
 240
 241                 wfRunHooks( 'ParserClearState', array( &$this ) );
 242                 wfProfileOut( __METHOD__ );
 243         }
 244
 245         function setOutputType( $ot ) {
 246                 $this->mOutputType = $ot;
 247                 // Shortcut alias
 248                 $this->ot = array(
 249                         'html' => $ot == self::OT_HTML,
 250                         'wiki' => $ot == self::OT_WIKI,
 251                         'pre' => $ot == self::OT_PREPROCESS,
 252                 );
 253         }
 254
 255         /**
 256          * Set the context title
 257          */
 258         function setTitle( $t ) {
 259                 if ( !$t || $t instanceof FakeTitle ) {
 260                         $t = Title::newFromText( 'NO TITLE' );
 261                 }
 262
 263                 if ( strval( $t->getFragment() ) !== '' ) {
 264                         # Strip the fragment to avoid various odd effects
 265                         $this->mTitle = clone $t;
 266                         $this->mTitle->setFragment( '' );
 267                 } else {
 268                         $this->mTitle = $t;
 269                 }
 270         }
 271
 272         /**
 273          * Accessor for mUniqPrefix.
 274          *
 275          * @public
 276          */
 277         function uniqPrefix() {
 278                 if( !isset( $this->mUniqPrefix ) ) {
 279                         // @todo Fixme: this is probably *horribly wrong*
 280                         // LanguageConverter seems to want $wgParser's uniqPrefix, however
 281                         // if this is called for a parser cache hit, the parser may not
 282                         // have ever been initialized in the first place.
 283                         // Not really sure what the heck is supposed to be going on here.
 284                         return '';
 285                         //throw new MWException( "Accessing uninitialized mUniqPrefix" );
 286                 }
 287                 return $this->mUniqPrefix;
 288         }
 289
 290         /**
 291          * Convert wikitext to HTML
 292          * Do not call this function recursively.
 293          *
 294          * @param $text String: text we want to parse
 295          * @param $title A title object
 296          * @param $options ParserOptions
 297          * @param $linestart boolean
 298          * @param $clearState boolean
 299          * @param $revid Int: number to pass in {{REVISIONID}}
 300          * @return ParserOutput a ParserOutput
 301          */
 302         public function parse( $text, Title $title, ParserOptions $options, $linestart = true, $clearState = true, $revid = null ) {
 303                 /**
 304                  * First pass--just handle <nowiki> sections, pass the rest off
 305                  * to internalParse() which does all the real work.
 306                  */
 307
 308                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang, $wgDisableLangConversion;
 309                 $fname = __METHOD__.'-' . wfGetCaller();
 310                 wfProfileIn( __METHOD__ );
 311                 wfProfileIn( $fname );
 312
 313                 if ( $clearState ) {
 314                         $this->clearState();
 315                 }
 316
 317                 $this->mOptions = $options;
 318                 $this->setTitle( $title ); // Page title has to be set for the pre-processor
 319
 320                 $oldRevisionId = $this->mRevisionId;
 321                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 322                 if( $revid !== null ) {
 323                         $this->mRevisionId = $revid;
 324                         $this->mRevisionTimestamp = null;
 325                 }
 326                 $this->setOutputType( self::OT_HTML );
 327                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 328                 # No more strip!
 329                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 330                 $text = $this->internalParse( $text );
 331
 332                 $text = $this->mStripState->unstripGeneral( $text );
 333
 334                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 335                 $fixtags = array(
 336                         # french spaces, last one Guillemet-left
 337                         # only if there is something before the space
 338                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&nbsp;\\2',
 339                         # french spaces, Guillemet-right
 340                         '/(\\302\\253) /' => '\\1&nbsp;',
 341                         '/&nbsp;(!\s*important)/' => ' \\1', #Beware of CSS magic word !important, bug #11874.
 342                 );
 343                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 344
 345                 $text = $this->doBlockLevels( $text, $linestart );
 346
 347                 $this->replaceLinkHolders( $text );
 348
 349                 // The position of the convert() call should not be changed. it
 350                 // assumes that the links are all replaced and the only thing left
 351                 // is the <nowiki> mark.
 352                 if ( !( $wgDisableLangConversion
 353                                 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
 354                                 || $this->mTitle->isTalkPage()
 355                                 || $this->mTitle->isConversionTable() ) ) {
 356                         $text = $wgContLang->convert( $text );
 357                 }
 358
 359                 // A title may have been set in a conversion rule.
 360                 // Note that if a user tries to set a title in a conversion
 361                 // rule but content conversion was not done, then the parser
 362                 // won't pick it up.  This is probably expected behavior.
 363                 if ( $wgContLang->getConvRuleTitle() ) {
 364                         $this->mOutput->setTitleText( $wgContLang->getConvRuleTitle() );
 365                 } elseif ( !( $wgDisableLangConversion
 366                                           || isset( $this->mDoubleUnderscores['notitleconvert'] ) ) ) {
 367                         $this->mOutput->setTitleText( $wgContLang->convert( $title->getPrefixedText() ) );
 368                 }
 369
 370                 $text = $this->mStripState->unstripNoWiki( $text );
 371
 372                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 373
 374 //!JF Move to its own function
 375
 376                 $uniq_prefix = $this->mUniqPrefix;
 377                 $matches = array();
 378                 $elements = array_keys( $this->mTransparentTagHooks );
 379                 $text = self::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 380
 381                 foreach( $matches as $marker => $data ) {
 382                         list( $element, $content, $params, $tag ) = $data;
 383                         $tagName = strtolower( $element );
 384                         if( isset( $this->mTransparentTagHooks[$tagName] ) ) {
 385                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName],
 386                                         array( $content, $params, $this ) );
 387                         } else {
 388                                 $output = $tag;
 389                         }
 390                         $this->mStripState->general->setPair( $marker, $output );
 391                 }
 392                 $text = $this->mStripState->unstripGeneral( $text );
 393
 394                 $text = Sanitizer::normalizeCharReferences( $text );
 395
 396                 if ( ( $wgUseTidy && $this->mOptions->mTidy ) || $wgAlwaysUseTidy ) {
 397                         $text = MWTidy::tidy( $text );
 398                 } else {
 399                         # attempt to sanitize at least some nesting problems
 400                         # (bug #2702 and quite a few others)
 401                         $tidyregs = array(
 402                                 # ''Something [http://www.cool.com cool''] -->
 403                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 404                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 405                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 406                                 # fix up an anchor inside another anchor, only
 407                                 # at least for a single single nested link (bug 3695)
 408                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 409                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 410                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 411                                 # contains a div, so fix it up here; replace
 412                                 # div with escaped text
 413                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 414                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 415                                 # remove empty italic or bold tag pairs, some
 416                                 # introduced by rules above
 417                                 '/<([bi])><\/\\1>/' => '',
 418                         );
 419
 420                         $text = preg_replace(
 421                                 array_keys( $tidyregs ),
 422                                 array_values( $tidyregs ),
 423                                 $text );
 424                 }
 425                 global $wgExpensiveParserFunctionLimit;
 426                 if ( $this->mExpensiveFunctionCount > $wgExpensiveParserFunctionLimit ) {
 427                         $this->limitationWarn( 'expensive-parserfunction', $this->mExpensiveFunctionCount, $wgExpensiveParserFunctionLimit );
 428                 }
 429
 430                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 431
 432                 # Information on include size limits, for the benefit of users who try to skirt them
 433                 if ( $this->mOptions->getEnableLimitReport() ) {
 434                         global $wgExpensiveParserFunctionLimit;
 435                         $max = $this->mOptions->getMaxIncludeSize();
 436                         $PFreport = "Expensive parser function count: {$this->mExpensiveFunctionCount}/$wgExpensiveParserFunctionLimit\n";
 437                         $limitReport =
 438                                 "NewPP limit report\n" .
 439                                 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->mMaxPPNodeCount}\n" .
 440                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 441                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n".
 442                                 $PFreport;
 443                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 444                         $text .= "\n<!-- \n$limitReport-->\n";
 445                 }
 446                 $this->mOutput->setText( $text );
 447
 448                 $this->mRevisionId = $oldRevisionId;
 449                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 450                 wfProfileOut( $fname );
 451                 wfProfileOut( __METHOD__ );
 452
 453                 return $this->mOutput;
 454         }
 455
 456         /**
 457          * Recursive parser entry point that can be called from an extension tag
 458          * hook.
 459          *
 460          * If $frame is not provided, then template variables (e.g., {{{1}}}) within $text are not expanded
 461          *
 462          * @param $text String: text extension wants to have parsed
 463          * @param PPFrame $frame: The frame to use for expanding any template variables
 464          */
 465         function recursiveTagParse( $text, $frame=false ) {
 466                 wfProfileIn( __METHOD__ );
 467                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 468                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 469                 $text = $this->internalParse( $text, false, $frame );
 470                 wfProfileOut( __METHOD__ );
 471                 return $text;
 472         }
 473
 474         /**
 475          * Expand templates and variables in the text, producing valid, static wikitext.
 476          * Also removes comments.
 477          */
 478         function preprocess( $text, $title, $options, $revid = null ) {
 479                 wfProfileIn( __METHOD__ );
 480                 $this->clearState();
 481                 $this->setOutputType( self::OT_PREPROCESS );
 482                 $this->mOptions = $options;
 483                 $this->setTitle( $title );
 484                 if( $revid !== null ) {
 485                         $this->mRevisionId = $revid;
 486                 }
 487                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 488                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 489                 $text = $this->replaceVariables( $text );
 490                 $text = $this->mStripState->unstripBoth( $text );
 491                 wfProfileOut( __METHOD__ );
 492                 return $text;
 493         }
 494
 495         /**
 496          * Get the wikitext of a page as though it was transcluded.
 497          *
 498          * Specifically <includeonly> etc. are parsed, redirects are followed, comments
 499          * are removed, but templates arguments and parser functions are untouched.
 500          *
 501          * This is not called by the parser itself, see braceSubstitution for its transclusion.
 502          */
 503         public function getTransclusionText( $title ) {
 504                 list( $text, $title ) = $this->getTemplateDom( $title );
 505                 $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
 506                 return $this->getPreprocessor()->newFrame()->expand( $text, $flags );
 507         }
 508
 509         /**
 510          * Get a random string
 511          *
 512          * @private
 513          * @static
 514          */
 515         function getRandomString() {
 516                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 517         }
 518
 519         function &getTitle() { return $this->mTitle; }
 520         function getOptions() { return $this->mOptions; }
 521         function getRevisionId() { return $this->mRevisionId; }
 522         function getOutput() { return $this->mOutput; }
 523         function nextLinkID() { return $this->mLinkID++; }
 524
 525         function getFunctionLang() {
 526                 global $wgLang, $wgContLang;
 527
 528                 $target = $this->mOptions->getTargetLanguage();
 529                 if ( $target !== null ) {
 530                         return $target;
 531                 } else {
 532                         return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 533                 }
 534         }
 535
 536         /**
 537          * Get a preprocessor object
 538          */
 539         function getPreprocessor() {
 540                 if ( !isset( $this->mPreprocessor ) ) {
 541                         $class = $this->mPreprocessorClass;
 542                         $this->mPreprocessor = new $class( $this );
 543                 }
 544                 return $this->mPreprocessor;
 545         }
 546
 547         /**
 548          * Replaces all occurrences of HTML-style comments and the given tags
 549          * in the text with a random marker and returns the next text. The output
 550          * parameter $matches will be an associative array filled with data in
 551          * the form:
 552          *   'UNIQ-xxxxx' => array(
 553          *     'element',
 554          *     'tag content',
 555          *     array( 'param' => 'x' ),
 556          *     '<element param="x">tag content</element>' ) )
 557          *
 558          * @param $elements list of element names. Comments are always extracted.
 559          * @param $text Source text string.
 560          * @param $uniq_prefix
 561          *
 562          * @public
 563          * @static
 564          */
 565         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 566                 static $n = 1;
 567                 $stripped = '';
 568                 $matches = array();
 569
 570                 $taglist = implode( '|', $elements );
 571                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
 572
 573                 while ( $text != '' ) {
 574                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 575                         $stripped .= $p[0];
 576                         if( count( $p ) < 5 ) {
 577                                 break;
 578                         }
 579                         if( count( $p ) > 5 ) {
 580                                 // comment
 581                                 $element    = $p[4];
 582                                 $attributes = '';
 583                                 $close      = '';
 584                                 $inside     = $p[5];
 585                         } else {
 586                                 // tag
 587                                 $element    = $p[1];
 588                                 $attributes = $p[2];
 589                                 $close      = $p[3];
 590                                 $inside     = $p[4];
 591                         }
 592
 593                         $marker = "$uniq_prefix-$element-" . sprintf('%08X', $n++) . self::MARKER_SUFFIX;
 594                         $stripped .= $marker;
 595
 596                         if ( $close === '/>' ) {
 597                                 // Empty element tag, <tag />
 598                                 $content = null;
 599                                 $text = $inside;
 600                                 $tail = null;
 601                         } else {
 602                                 if( $element === '!--' ) {
 603                                         $end = '/(-->)/';
 604                                 } else {
 605                                         $end = "/(<\\/$element\\s*>)/i";
 606                                 }
 607                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 608                                 $content = $q[0];
 609                                 if( count( $q ) < 3 ) {
 610                                         # No end tag -- let it run out to the end of the text.
 611                                         $tail = '';
 612                                         $text = '';
 613                                 } else {
 614                                         $tail = $q[1];
 615                                         $text = $q[2];
 616                                 }
 617                         }
 618
 619                         $matches[$marker] = array( $element,
 620                                 $content,
 621                                 Sanitizer::decodeTagAttributes( $attributes ),
 622                                 "<$element$attributes$close$content$tail" );
 623                 }
 624                 return $stripped;
 625         }
 626
 627         /**
 628          * Get a list of strippable XML-like elements
 629          */
 630         function getStripList() {
 631                 return $this->mStripList;
 632         }
 633
 634         /**
 635          * @deprecated use replaceVariables
 636          */
 637         function strip( $text, $state, $stripcomments = false , $dontstrip = array () ) {
 638                 return $text;
 639         }
 640
 641         /**
 642          * Restores pre, math, and other extensions removed by strip()
 643          *
 644          * always call unstripNoWiki() after this one
 645          * @private
 646          * @deprecated use $this->mStripState->unstrip()
 647          */
 648         function unstrip( $text, $state ) {
 649                 return $state->unstripGeneral( $text );
 650         }
 651
 652         /**
 653          * Always call this after unstrip() to preserve the order
 654          *
 655          * @private
 656          * @deprecated use $this->mStripState->unstrip()
 657          */
 658         function unstripNoWiki( $text, $state ) {
 659                 return $state->unstripNoWiki( $text );
 660         }
 661
 662         /**
 663          * @deprecated use $this->mStripState->unstripBoth()
 664          */
 665         function unstripForHTML( $text ) {
 666                 return $this->mStripState->unstripBoth( $text );
 667         }
 668
 669         /**
 670          * Add an item to the strip state
 671          * Returns the unique tag which must be inserted into the stripped text
 672          * The tag will be replaced with the original text in unstrip()
 673          *
 674          * @private
 675          */
 676         function insertStripItem( $text ) {
 677                 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
 678                 $this->mMarkerIndex++;
 679                 $this->mStripState->general->setPair( $rnd, $text );
 680                 return $rnd;
 681         }
 682
 683         /**
 684          * Interface with html tidy
 685          * @deprecated Use MWTidy::tidy()
 686          */
 687         public static function tidy( $text ) {
 688                 wfDeprecated( __METHOD__ );
 689                 return MWTidy::tidy( $text );
 690         }
 691
 692         /**
 693          * parse the wiki syntax used to render tables
 694          *
 695          * @private
 696          */
 697         function doTableStuff ( $text ) {
 698                 wfProfileIn( __METHOD__ );
 699
 700                 $lines = StringUtils::explode( "\n", $text );
 701                 $out = '';
 702                 $td_history = array (); // Is currently a td tag open?
 703                 $last_tag_history = array (); // Save history of last lag activated (td, th or caption)
 704                 $tr_history = array (); // Is currently a tr tag open?
 705                 $tr_attributes = array (); // history of tr attributes
 706                 $has_opened_tr = array(); // Did this table open a <tr> element?
 707                 $indent_level = 0; // indent level of the table
 708
 709                 foreach ( $lines as $outLine ) {
 710                         $line = trim( $outLine );
 711
 712                         if( $line == '' ) { // empty line, go to next line
 713                                 $out .= $outLine."\n";
 714                                 continue;
 715                         }
 716                         $first_character = $line[0];
 717                         $matches = array();
 718
 719                         if ( preg_match( '/^(:*)\{\|(.*)$/', $line , $matches ) ) {
 720                                 // First check if we are starting a new table
 721                                 $indent_level = strlen( $matches[1] );
 722
 723                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 724                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'table' );
 725
 726                                 $outLine = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 727                                 array_push( $td_history , false );
 728                                 array_push( $last_tag_history , '' );
 729                                 array_push( $tr_history , false );
 730                                 array_push( $tr_attributes , '' );
 731                                 array_push( $has_opened_tr , false );
 732                         } else if ( count ( $td_history ) == 0 ) {
 733                                 // Don't do any of the following
 734                                 $out .= $outLine."\n";
 735                                 continue;
 736                         } else if ( substr ( $line , 0 , 2 ) === '|}' ) {
 737                                 // We are ending a table
 738                                 $line = '</table>' . substr ( $line , 2 );
 739                                 $last_tag = array_pop ( $last_tag_history );
 740
 741                                 if ( !array_pop ( $has_opened_tr ) ) {
 742                                         $line = "<tr><td></td></tr>{$line}";
 743                                 }
 744
 745                                 if ( array_pop ( $tr_history ) ) {
 746                                         $line = "</tr>{$line}";
 747                                 }
 748
 749                                 if ( array_pop ( $td_history ) ) {
 750                                         $line = "</{$last_tag}>{$line}";
 751                                 }
 752                                 array_pop ( $tr_attributes );
 753                                 $outLine = $line . str_repeat( '</dd></dl>' , $indent_level );
 754                         } else if ( substr ( $line , 0 , 2 ) === '|-' ) {
 755                                 // Now we have a table row
 756                                 $line = preg_replace( '#^\|-+#', '', $line );
 757
 758                                 // Whats after the tag is now only attributes
 759                                 $attributes = $this->mStripState->unstripBoth( $line );
 760                                 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
 761                                 array_pop( $tr_attributes );
 762                                 array_push( $tr_attributes, $attributes );
 763
 764                                 $line = '';
 765                                 $last_tag = array_pop ( $last_tag_history );
 766                                 array_pop ( $has_opened_tr );
 767                                 array_push ( $has_opened_tr , true );
 768
 769                                 if ( array_pop ( $tr_history ) ) {
 770                                         $line = '</tr>';
 771                                 }
 772
 773                                 if ( array_pop ( $td_history ) ) {
 774                                         $line = "</{$last_tag}>{$line}";
 775                                 }
 776
 777                                 $outLine = $line;
 778                                 array_push ( $tr_history , false );
 779                                 array_push ( $td_history , false );
 780                                 array_push ( $last_tag_history , '' );
 781                         }
 782                         else if ( $first_character === '|' || $first_character === '!' || substr ( $line , 0 , 2 )  === '|+' ) {
 783                                 // This might be cell elements, td, th or captions
 784                                 if ( substr ( $line , 0 , 2 ) === '|+' ) {
 785                                         $first_character = '+';
 786                                         $line = substr ( $line , 1 );
 787                                 }
 788
 789                                 $line = substr ( $line , 1 );
 790
 791                                 if ( $first_character === '!' ) {
 792                                         $line = str_replace ( '!!' , '||' , $line );
 793                                 }
 794
 795                                 // Split up multiple cells on the same line.
 796                                 // FIXME : This can result in improper nesting of tags processed
 797                                 // by earlier parser steps, but should avoid splitting up eg
 798                                 // attribute values containing literal "||".
 799                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 800
 801                                 $outLine = '';
 802
 803                                 // Loop through each table cell
 804                                 foreach ( $cells as $cell )
 805                                 {
 806                                         $previous = '';
 807                                         if ( $first_character !== '+' )
 808                                         {
 809                                                 $tr_after = array_pop ( $tr_attributes );
 810                                                 if ( !array_pop ( $tr_history ) ) {
 811                                                         $previous = "<tr{$tr_after}>\n";
 812                                                 }
 813                                                 array_push ( $tr_history , true );
 814                                                 array_push ( $tr_attributes , '' );
 815                                                 array_pop ( $has_opened_tr );
 816                                                 array_push ( $has_opened_tr , true );
 817                                         }
 818
 819                                         $last_tag = array_pop ( $last_tag_history );
 820
 821                                         if ( array_pop ( $td_history ) ) {
 822                                                 $previous = "</{$last_tag}>{$previous}";
 823                                         }
 824
 825                                         if ( $first_character === '|' ) {
 826                                                 $last_tag = 'td';
 827                                         } else if ( $first_character === '!' ) {
 828                                                 $last_tag = 'th';
 829                                         } else if ( $first_character === '+' ) {
 830                                                 $last_tag = 'caption';
 831                                         } else {
 832                                                 $last_tag = '';
 833                                         }
 834
 835                                         array_push ( $last_tag_history , $last_tag );
 836
 837                                         // A cell could contain both parameters and data
 838                                         $cell_data = explode ( '|' , $cell , 2 );
 839
 840                                         // Bug 553: Note that a '|' inside an invalid link should not
 841                                         // be mistaken as delimiting cell parameters
 842                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 843                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 844                                         } else if ( count ( $cell_data ) == 1 )
 845                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 846                                         else {
 847                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 848                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 849                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 850                                         }
 851
 852                                         $outLine .= $cell;
 853                                         array_push ( $td_history , true );
 854                                 }
 855                         }
 856                         $out .= $outLine . "\n";
 857                 }
 858
 859                 // Closing open td, tr && table
 860                 while ( count ( $td_history ) > 0 )
 861                 {
 862                         if ( array_pop ( $td_history ) ) {
 863                                 $out .= "</td>\n";
 864                         }
 865                         if ( array_pop ( $tr_history ) ) {
 866                                 $out .= "</tr>\n";
 867                         }
 868                         if ( !array_pop ( $has_opened_tr ) ) {
 869                                 $out .= "<tr><td></td></tr>\n" ;
 870                         }
 871
 872                         $out .= "</table>\n";
 873                 }
 874
 875                 // Remove trailing line-ending (b/c)
 876                 if ( substr( $out, -1 ) === "\n" ) {
 877                         $out = substr( $out, 0, -1 );
 878                 }
 879
 880                 // special case: don't return empty table
 881                 if( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
 882                         $out = '';
 883                 }
 884
 885                 wfProfileOut( __METHOD__ );
 886
 887                 return $out;
 888         }
 889
 890         /**
 891          * Helper function for parse() that transforms wiki markup into
 892          * HTML. Only called for $mOutputType == self::OT_HTML.
 893          *
 894          * @private
 895          */
 896         function internalParse( $text, $isMain = true, $frame=false ) {
 897                 wfProfileIn( __METHOD__ );
 898
 899                 $origText = $text;
 900
 901                 # Hook to suspend the parser in this state
 902                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
 903                         wfProfileOut( __METHOD__ );
 904                         return $text ;
 905                 }
 906
 907                 // if $frame is provided, then use $frame for replacing any variables
 908                 if ($frame) {
 909                         // use frame depth to infer how include/noinclude tags should be handled
 910                         // depth=0 means this is the top-level document; otherwise it's an included document
 911                         if( !$frame->depth )
 912                                 $flag = 0;
 913                         else
 914                                 $flag = Parser::PTD_FOR_INCLUSION;
 915                         $dom = $this->preprocessToDom( $text, $flag );
 916                         $text = $frame->expand( $dom );
 917                 }
 918                 // if $frame is not provided, then use old-style replaceVariables
 919                 else {
 920                         $text = $this->replaceVariables( $text );
 921                 }
 922
 923                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
 924                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
 925
 926                 // Tables need to come after variable replacement for things to work
 927                 // properly; putting them before other transformations should keep
 928                 // exciting things like link expansions from showing up in surprising
 929                 // places.
 930                 $text = $this->doTableStuff( $text );
 931
 932                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 933
 934                 $text = $this->doDoubleUnderscore( $text );
 935
 936                 $text = $this->doHeadings( $text );
 937                 if( $this->mOptions->getUseDynamicDates() ) {
 938                         $df = DateFormatter::getInstance();
 939                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 940                 }
 941                 $text = $this->doAllQuotes( $text );
 942                 $text = $this->replaceInternalLinks( $text );
 943                 $text = $this->replaceExternalLinks( $text );
 944
 945                 # replaceInternalLinks may sometimes leave behind
 946                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 947                 $text = str_replace($this->mUniqPrefix.'NOPARSE', '', $text);
 948
 949                 $text = $this->doMagicLinks( $text );
 950                 $text = $this->formatHeadings( $text, $origText, $isMain );
 951
 952                 wfProfileOut( __METHOD__ );
 953                 return $text;
 954         }
 955
 956         /**
 957          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 958          * magic external links.
 959          *
 960          * DML
 961          * @private
 962          */
 963         function doMagicLinks( $text ) {
 964                 wfProfileIn( __METHOD__ );
 965                 $prots = $this->mUrlProtocols;
 966                 $urlChar = self::EXT_LINK_URL_CLASS;
 967                 $text = preg_replace_callback(
 968                         '!(?:                           # Start cases
 969                                 (<a.*?</a>) |               # m[1]: Skip link text
 970                                 (<.*?>) |                   # m[2]: Skip stuff inside HTML elements' . "
 971                                 (\\b(?:$prots)$urlChar+) |  # m[3]: Free external links" . '
 972                                 (?:RFC|PMID)\s+([0-9]+) |   # m[4]: RFC or PMID, capture number
 973                                 ISBN\s+(\b                  # m[5]: ISBN, capture number
 974                                     (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
 975                                     (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
 976                                     [0-9Xx]                 # check digit
 977                                     \b)
 978                         )!x', array( &$this, 'magicLinkCallback' ), $text );
 979                 wfProfileOut( __METHOD__ );
 980                 return $text;
 981         }
 982
 983         function magicLinkCallback( $m ) {
 984                 if ( isset( $m[1] ) && $m[1] !== '' ) {
 985                         # Skip anchor
 986                         return $m[0];
 987                 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
 988                         # Skip HTML element
 989                         return $m[0];
 990                 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
 991                         # Free external link
 992                         return $this->makeFreeExternalLink( $m[0] );
 993                 } elseif ( isset( $m[4] ) && $m[4] !== '' ) {
 994                         # RFC or PMID
 995                         $CssClass = '';
 996                         if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
 997                                 $keyword = 'RFC';
 998                                 $urlmsg = 'rfcurl';
 999                                 $CssClass = 'mw-magiclink-rfc';
1000                                 $id = $m[4];
1001                         } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1002                                 $keyword = 'PMID';
1003                                 $urlmsg = 'pubmedurl';
1004                                 $CssClass = 'mw-magiclink-pmid';
1005                                 $id = $m[4];
1006                         } else {
1007                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1008                                         substr($m[0], 0, 20 ) . '"' );
1009                         }
1010                         $url = wfMsg( $urlmsg, $id);
1011                         $sk = $this->mOptions->getSkin();
1012                         $la = $sk->getExternalLinkAttributes( "external $CssClass" );
1013                         return "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
1014                 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1015                         # ISBN
1016                         $isbn = $m[5];
1017                         $num = strtr( $isbn, array(
1018                                 '-' => '',
1019                                 ' ' => '',
1020                                 'x' => 'X',
1021                         ));
1022                         $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
1023                         return'<a href="' .
1024                                 $titleObj->escapeLocalUrl() .
1025                                 "\" class=\"internal mw-magiclink-isbn\">ISBN $isbn</a>";
1026                 } else {
1027                         return $m[0];
1028                 }
1029         }
1030
1031         /**
1032          * Make a free external link, given a user-supplied URL
1033          * @return HTML
1034          * @private
1035          */
1036         function makeFreeExternalLink( $url ) {
1037                 global $wgContLang;
1038                 wfProfileIn( __METHOD__ );
1039
1040                 $sk = $this->mOptions->getSkin();
1041                 $trail = '';
1042
1043                 # The characters '<' and '>' (which were escaped by
1044                 # removeHTMLtags()) should not be included in
1045                 # URLs, per RFC 2396.
1046                 $m2 = array();
1047                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1048                         $trail = substr($url, $m2[0][1]) . $trail;
1049                         $url = substr($url, 0, $m2[0][1]);
1050                 }
1051
1052                 # Move trailing punctuation to $trail
1053                 $sep = ',;\.:!?';
1054                 # If there is no left bracket, then consider right brackets fair game too
1055                 if ( strpos( $url, '(' ) === false ) {
1056                         $sep .= ')';
1057                 }
1058
1059                 $numSepChars = strspn( strrev( $url ), $sep );
1060                 if ( $numSepChars ) {
1061                         $trail = substr( $url, -$numSepChars ) . $trail;
1062                         $url = substr( $url, 0, -$numSepChars );
1063                 }
1064
1065                 $url = Sanitizer::cleanUrl( $url );
1066
1067                 # Is this an external image?
1068                 $text = $this->maybeMakeExternalImage( $url );
1069                 if ( $text === false ) {
1070                         # Not an image, make a link
1071                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free',
1072                                 $this->getExternalLinkAttribs( $url ) );
1073                         # Register it in the output object...
1074                         # Replace unnecessary URL escape codes with their equivalent characters
1075                         $pasteurized = self::replaceUnusualEscapes( $url );
1076                         $this->mOutput->addExternalLink( $pasteurized );
1077                 }
1078                 wfProfileOut( __METHOD__ );
1079                 return $text . $trail;
1080         }
1081
1082
1083         /**
1084          * Parse headers and return html
1085          *
1086          * @private
1087          */
1088         function doHeadings( $text ) {
1089                 wfProfileIn( __METHOD__ );
1090                 for ( $i = 6; $i >= 1; --$i ) {
1091                         $h = str_repeat( '=', $i );
1092                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1093                           "<h$i>\\1</h$i>", $text );
1094                 }
1095                 wfProfileOut( __METHOD__ );
1096                 return $text;
1097         }
1098
1099         /**
1100          * Replace single quotes with HTML markup
1101          * @private
1102          * @return string the altered text
1103          */
1104         function doAllQuotes( $text ) {
1105                 wfProfileIn( __METHOD__ );
1106                 $outtext = '';
1107                 $lines = StringUtils::explode( "\n", $text );
1108                 foreach ( $lines as $line ) {
1109                         $outtext .= $this->doQuotes( $line ) . "\n";
1110                 }
1111                 $outtext = substr($outtext, 0,-1);
1112                 wfProfileOut( __METHOD__ );
1113                 return $outtext;
1114         }
1115
1116         /**
1117          * Helper function for doAllQuotes()
1118          */
1119         public function doQuotes( $text ) {
1120                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1121                 if ( count( $arr ) == 1 )
1122                         return $text;
1123                 else
1124                 {
1125                         # First, do some preliminary work. This may shift some apostrophes from
1126                         # being mark-up to being text. It also counts the number of occurrences
1127                         # of bold and italics mark-ups.
1128                         $i = 0;
1129                         $numbold = 0;
1130                         $numitalics = 0;
1131                         foreach ( $arr as $r )
1132                         {
1133                                 if ( ( $i % 2 ) == 1 )
1134                                 {
1135                                         # If there are ever four apostrophes, assume the first is supposed to
1136                                         # be text, and the remaining three constitute mark-up for bold text.
1137                                         if ( strlen( $arr[$i] ) == 4 )
1138                                         {
1139                                                 $arr[$i-1] .= "'";
1140                                                 $arr[$i] = "'''";
1141                                         }
1142                                         # If there are more than 5 apostrophes in a row, assume they're all
1143                                         # text except for the last 5.
1144                                         else if ( strlen( $arr[$i] ) > 5 )
1145                                         {
1146                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1147                                                 $arr[$i] = "'''''";
1148                                         }
1149                                         # Count the number of occurrences of bold and italics mark-ups.
1150                                         # We are not counting sequences of five apostrophes.
1151                                         if ( strlen( $arr[$i] ) == 2 )      { $numitalics++;             }
1152                                         else if ( strlen( $arr[$i] ) == 3 ) { $numbold++;                }
1153                                         else if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
1154                                 }
1155                                 $i++;
1156                         }
1157
1158                         # If there is an odd number of both bold and italics, it is likely
1159                         # that one of the bold ones was meant to be an apostrophe followed
1160                         # by italics. Which one we cannot know for certain, but it is more
1161                         # likely to be one that has a single-letter word before it.
1162                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1163                         {
1164                                 $i = 0;
1165                                 $firstsingleletterword = -1;
1166                                 $firstmultiletterword = -1;
1167                                 $firstspace = -1;
1168                                 foreach ( $arr as $r )
1169                                 {
1170                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1171                                         {
1172                                                 $x1 = substr ($arr[$i-1], -1);
1173                                                 $x2 = substr ($arr[$i-1], -2, 1);
1174                                                 if ($x1 === ' ') {
1175                                                         if ($firstspace == -1) $firstspace = $i;
1176                                                 } else if ($x2 === ' ') {
1177                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1178                                                 } else {
1179                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1180                                                 }
1181                                         }
1182                                         $i++;
1183                                 }
1184
1185                                 # If there is a single-letter word, use it!
1186                                 if ($firstsingleletterword > -1)
1187                                 {
1188                                         $arr [ $firstsingleletterword ] = "''";
1189                                         $arr [ $firstsingleletterword-1 ] .= "'";
1190                                 }
1191                                 # If not, but there's a multi-letter word, use that one.
1192                                 else if ($firstmultiletterword > -1)
1193                                 {
1194                                         $arr [ $firstmultiletterword ] = "''";
1195                                         $arr [ $firstmultiletterword-1 ] .= "'";
1196                                 }
1197                                 # ... otherwise use the first one that has neither.
1198                                 # (notice that it is possible for all three to be -1 if, for example,
1199                                 # there is only one pentuple-apostrophe in the line)
1200                                 else if ($firstspace > -1)
1201                                 {
1202                                         $arr [ $firstspace ] = "''";
1203                                         $arr [ $firstspace-1 ] .= "'";
1204                                 }
1205                         }
1206
1207                         # Now let's actually convert our apostrophic mush to HTML!
1208                         $output = '';
1209                         $buffer = '';
1210                         $state = '';
1211                         $i = 0;
1212                         foreach ($arr as $r)
1213                         {
1214                                 if (($i % 2) == 0)
1215                                 {
1216                                         if ($state === 'both')
1217                                                 $buffer .= $r;
1218                                         else
1219                                                 $output .= $r;
1220                                 }
1221                                 else
1222                                 {
1223                                         if (strlen ($r) == 2)
1224                                         {
1225                                                 if ($state === 'i')
1226                                                 { $output .= '</i>'; $state = ''; }
1227                                                 else if ($state === 'bi')
1228                                                 { $output .= '</i>'; $state = 'b'; }
1229                                                 else if ($state === 'ib')
1230                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1231                                                 else if ($state === 'both')
1232                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1233                                                 else # $state can be 'b' or ''
1234                                                 { $output .= '<i>'; $state .= 'i'; }
1235                                         }
1236                                         else if (strlen ($r) == 3)
1237                                         {
1238                                                 if ($state === 'b')
1239                                                 { $output .= '</b>'; $state = ''; }
1240                                                 else if ($state === 'bi')
1241                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1242                                                 else if ($state === 'ib')
1243                                                 { $output .= '</b>'; $state = 'i'; }
1244                                                 else if ($state === 'both')
1245                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1246                                                 else # $state can be 'i' or ''
1247                                                 { $output .= '<b>'; $state .= 'b'; }
1248                                         }
1249                                         else if (strlen ($r) == 5)
1250                                         {
1251                                                 if ($state === 'b')
1252                                                 { $output .= '</b><i>'; $state = 'i'; }
1253                                                 else if ($state === 'i')
1254                                                 { $output .= '</i><b>'; $state = 'b'; }
1255                                                 else if ($state === 'bi')
1256                                                 { $output .= '</i></b>'; $state = ''; }
1257                                                 else if ($state === 'ib')
1258                                                 { $output .= '</b></i>'; $state = ''; }
1259                                                 else if ($state === 'both')
1260                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1261                                                 else # ($state == '')
1262                                                 { $buffer = ''; $state = 'both'; }
1263                                         }
1264                                 }
1265                                 $i++;
1266                         }
1267                         # Now close all remaining tags.  Notice that the order is important.
1268                         if ($state === 'b' || $state === 'ib')
1269                                 $output .= '</b>';
1270                         if ($state === 'i' || $state === 'bi' || $state === 'ib')
1271                                 $output .= '</i>';
1272                         if ($state === 'bi')
1273                                 $output .= '</b>';
1274                         # There might be lonely ''''', so make sure we have a buffer
1275                         if ($state === 'both' && $buffer)
1276                                 $output .= '<b><i>'.$buffer.'</i></b>';
1277                         return $output;
1278                 }
1279         }
1280
1281         /**
1282          * Replace external links (REL)
1283          *
1284          * Note: this is all very hackish and the order of execution matters a lot.
1285          * Make sure to run maintenance/parserTests.php if you change this code.
1286          *
1287          * @private
1288          */
1289         function replaceExternalLinks( $text ) {
1290                 global $wgContLang;
1291                 wfProfileIn( __METHOD__ );
1292
1293                 $sk = $this->mOptions->getSkin();
1294
1295                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1296                 $s = array_shift( $bits );
1297
1298                 $i = 0;
1299                 while ( $i<count( $bits ) ) {
1300                         $url = $bits[$i++];
1301                         $protocol = $bits[$i++];
1302                         $text = $bits[$i++];
1303                         $trail = $bits[$i++];
1304
1305                         # The characters '<' and '>' (which were escaped by
1306                         # removeHTMLtags()) should not be included in
1307                         # URLs, per RFC 2396.
1308                         $m2 = array();
1309                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1310                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1311                                 $url = substr($url, 0, $m2[0][1]);
1312                         }
1313
1314                         # If the link text is an image URL, replace it with an <img> tag
1315                         # This happened by accident in the original parser, but some people used it extensively
1316                         $img = $this->maybeMakeExternalImage( $text );
1317                         if ( $img !== false ) {
1318                                 $text = $img;
1319                         }
1320
1321                         $dtrail = '';
1322
1323                         # Set linktype for CSS - if URL==text, link is essentially free
1324                         $linktype = ($text === $url) ? 'free' : 'text';
1325
1326                         # No link text, e.g. [http://domain.tld/some.link]
1327                         if ( $text == '' ) {
1328                                 # Autonumber if allowed. See bug #5918
1329                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1330                                         $langObj = $this->getFunctionLang();
1331                                         $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1332                                         $linktype = 'autonumber';
1333                                 } else {
1334                                         # Otherwise just use the URL
1335                                         $text = htmlspecialchars( $url );
1336                                         $linktype = 'free';
1337                                 }
1338                         } else {
1339                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1340                                 # Check for trail
1341                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1342                         }
1343
1344                         $text = $wgContLang->markNoConversion($text);
1345
1346                         $url = Sanitizer::cleanUrl( $url );
1347
1348                         # Use the encoded URL
1349                         # This means that users can paste URLs directly into the text
1350                         # Funny characters like &ouml; aren't valid in URLs anyway
1351                         # This was changed in August 2004
1352                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype,
1353                                 $this->getExternalLinkAttribs( $url ) ) . $dtrail . $trail;
1354
1355                         # Register link in the output object.
1356                         # Replace unnecessary URL escape codes with the referenced character
1357                         # This prevents spammers from hiding links from the filters
1358                         $pasteurized = self::replaceUnusualEscapes( $url );
1359                         $this->mOutput->addExternalLink( $pasteurized );
1360                 }
1361
1362                 wfProfileOut( __METHOD__ );
1363                 return $s;
1364         }
1365
1366         /**
1367          * Get an associative array of additional HTML attributes appropriate for a
1368          * particular external link.  This currently may include rel => nofollow
1369          * (depending on configuration, namespace, and the URL's domain) and/or a
1370          * target attribute (depending on configuration).
1371          *
1372          * @param string $url Optional URL, to extract the domain from for rel =>
1373          *   nofollow if appropriate
1374          * @return array Associative array of HTML attributes
1375          */
1376         function getExternalLinkAttribs( $url = false ) {
1377                 $attribs = array();
1378                 global $wgNoFollowLinks, $wgNoFollowNsExceptions;
1379                 $ns = $this->mTitle->getNamespace();
1380                 if( $wgNoFollowLinks && !in_array($ns, $wgNoFollowNsExceptions) ) {
1381                         $attribs['rel'] = 'nofollow';
1382
1383                         global $wgNoFollowDomainExceptions;
1384                         if ( $wgNoFollowDomainExceptions ) {
1385                                 $bits = wfParseUrl( $url );
1386                                 if ( is_array( $bits ) && isset( $bits['host'] ) ) {
1387                                         foreach ( $wgNoFollowDomainExceptions as $domain ) {
1388                                                 if( substr( $bits['host'], -strlen( $domain ) )
1389                                                 == $domain ) {
1390                                                         unset( $attribs['rel'] );
1391                                                         break;
1392                                                 }
1393                                         }
1394                                 }
1395                         }
1396                 }
1397                 if ( $this->mOptions->getExternalLinkTarget() ) {
1398                         $attribs['target'] = $this->mOptions->getExternalLinkTarget();
1399                 }
1400                 return $attribs;
1401         }
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( __CLASS__, '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, through the exception, or through the on-wiki whitelist
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                 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1449                 if( $imagesexception && is_array( $imagesfrom ) ) {
1450                         $imagematch = false;
1451                         foreach( $imagesfrom as $match ) {
1452                                 if( strpos( $url, $match ) === 0 ) {
1453                                         $imagematch = true;
1454                                         break;
1455                                 }
1456                         }
1457                 } elseif( $imagesexception ) {
1458                         $imagematch = (strpos( $url, $imagesfrom ) === 0);
1459                 } else {
1460                         $imagematch = false;
1461                 }
1462                 if ( $this->mOptions->getAllowExternalImages()
1463                      || ( $imagesexception && $imagematch ) ) {
1464                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1465                                 # Image found
1466                                 $text = $sk->makeExternalImage( $url );
1467                         }
1468                 }
1469                 if( !$text && $this->mOptions->getEnableImageWhitelist()
1470                          && preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1471                         $whitelist = explode( "\n", wfMsgForContent( 'external_image_whitelist' ) );
1472                         foreach( $whitelist as $entry ) {
1473                                 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1474                                 if( strpos( $entry, '#' ) === 0 || $entry === '' )
1475                                         continue;
1476                                 if( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1477                                         # Image matches a whitelist entry
1478                                         $text = $sk->makeExternalImage( $url );
1479                                         break;
1480                                 }
1481                         }
1482                 }
1483                 return $text;
1484         }
1485
1486         /**
1487          * Process [[ ]] wikilinks
1488          * @return processed text
1489          *
1490          * @private
1491          */
1492         function replaceInternalLinks( $s ) {
1493                 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
1494                 return $s;
1495         }
1496
1497         /**
1498          * Process [[ ]] wikilinks (RIL)
1499          * @return LinkHolderArray
1500          *
1501          * @private
1502          */
1503         function replaceInternalLinks2( &$s ) {
1504                 global $wgContLang;
1505
1506                 wfProfileIn( __METHOD__ );
1507
1508                 wfProfileIn( __METHOD__.'-setup' );
1509                 static $tc = FALSE, $e1, $e1_img;
1510                 # the % is needed to support urlencoded titles as well
1511                 if ( !$tc ) {
1512                         $tc = Title::legalChars() . '#%';
1513                         # Match a link having the form [[namespace:link|alternate]]trail
1514                         $e1 = "/^([{$tc}]*)(\\|.*?)?]](.*)\$/sD";
1515                         # Match cases where there is no "]]", which might still be images
1516                         $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
1517                 }
1518
1519                 $sk = $this->mOptions->getSkin();
1520                 $holders = new LinkHolderArray( $this );
1521
1522                 #split the entire text string on occurences of [[
1523                 $a = StringUtils::explode( '[[', ' ' . $s );
1524                 #get the first element (all text up to first [[), and remove the space we added
1525                 $s = $a->current();
1526                 $a->next();
1527                 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
1528                 $s = substr( $s, 1 );
1529
1530                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1531                 $e2 = null;
1532                 if ( $useLinkPrefixExtension ) {
1533                         # Match the end of a line for a word that's not followed by whitespace,
1534                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1535                         $e2 = wfMsgForContent( 'linkprefix' );
1536                 }
1537
1538                 if( is_null( $this->mTitle ) ) {
1539                         wfProfileOut( __METHOD__.'-setup' );
1540                         wfProfileOut( __METHOD__ );
1541                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1542                 }
1543                 $nottalk = !$this->mTitle->isTalkPage();
1544
1545                 if ( $useLinkPrefixExtension ) {
1546                         $m = array();
1547                         if ( preg_match( $e2, $s, $m ) ) {
1548                                 $first_prefix = $m[2];
1549                         } else {
1550                                 $first_prefix = false;
1551                         }
1552                 } else {
1553                         $prefix = '';
1554                 }
1555
1556                 if($wgContLang->hasVariants()) {
1557                         $selflink = $wgContLang->convertLinkToAllVariants($this->mTitle->getPrefixedText());
1558                 } else {
1559                         $selflink = array($this->mTitle->getPrefixedText());
1560                 }
1561                 $useSubpages = $this->areSubpagesAllowed();
1562                 wfProfileOut( __METHOD__.'-setup' );
1563
1564                 # Loop for each link
1565                 for ( ; $line !== false && $line !== null ; $a->next(), $line = $a->current() ) {
1566                         # Check for excessive memory usage
1567                         if ( $holders->isBig() ) {
1568                                 # Too big
1569                                 # Do the existence check, replace the link holders and clear the array
1570                                 $holders->replace( $s );
1571                                 $holders->clear();
1572                         }
1573
1574                         if ( $useLinkPrefixExtension ) {
1575                                 wfProfileIn( __METHOD__.'-prefixhandling' );
1576                                 if ( preg_match( $e2, $s, $m ) ) {
1577                                         $prefix = $m[2];
1578                                         $s = $m[1];
1579                                 } else {
1580                                         $prefix='';
1581                                 }
1582                                 # first link
1583                                 if($first_prefix) {
1584                                         $prefix = $first_prefix;
1585                                         $first_prefix = false;
1586                                 }
1587                                 wfProfileOut( __METHOD__.'-prefixhandling' );
1588                         }
1589
1590                         $might_be_img = false;
1591
1592                         wfProfileIn( __METHOD__."-e1" );
1593                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1594
1595                                 if( $m[2] === '' ) {
1596                                         $text = '';
1597                                 } elseif( $m[2] === '|' ) {
1598                                         $text = $this->getPipeTrickText( $m[1] );
1599                                 } else {
1600                                         $text = substr( $m[2], 1 );
1601                                 }
1602
1603                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1604                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1605                                 # the real problem is with the $e1 regex
1606                                 # See bug 1300.
1607                                 #
1608                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1609                                 # and no image is in sight. See bug 2095.
1610                                 #
1611                                 if( $text !== '' &&
1612                                         substr( $m[3], 0, 1 ) === ']' &&
1613                                         strpos($text, '[') !== false
1614                                 )
1615                                 {
1616                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1617                                         $m[3] = substr( $m[3], 1 );
1618                                 }
1619
1620                                 # Handle pipe-trick for [[|<blah>]]
1621                                 $lnk = $m[1] === '' ? $this->getPipeTrickLink( $text ) : $m[1];
1622                                 # fix up urlencoded title texts
1623                                 if( strpos( $lnk, '%' ) !== false ) {
1624                                         # Should anchors '#' also be rejected?
1625                                         $lnk = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($lnk) );
1626                                 }
1627
1628                                 $trail = $m[3];
1629                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1630                                 $might_be_img = true;
1631                                 $text = $m[2];
1632                                 $lnk = strpos( $m[1], '%' ) === false ? $m[1] : urldecode( $m[1] );
1633                                 $trail = "";
1634                         } else { # Invalid form; output directly
1635                                 $s .= $prefix . '[[' . $line ;
1636                                 wfProfileOut( __METHOD__."-e1" );
1637                                 continue;
1638                         }
1639                         wfProfileOut( __METHOD__."-e1" );
1640                         wfProfileIn( __METHOD__."-misc" );
1641
1642                         # Don't allow internal links to pages containing
1643                         # PROTO: where PROTO is a valid URL protocol; these
1644                         # should be external links.
1645                         if ( preg_match( '/^\b(?:' . wfUrlProtocols() . ')/', $lnk ) ) {
1646                                 $s .= $prefix . '[[' . $line ;
1647                                 wfProfileOut( __METHOD__."-misc" );
1648                                 continue;
1649                         }
1650
1651                         # Make subpage if necessary
1652                         if ( $useSubpages ) {
1653                                 $link = $this->maybeDoSubpageLink( $lnk, $text );
1654                         } else {
1655                                 $link = $lnk;
1656                         }
1657
1658                         $noforce = (substr( $lnk, 0, 1 ) !== ':');
1659                         if (!$noforce) {
1660                                 # Strip off leading ':'
1661                                 $link = substr( $link, 1 );
1662                         }
1663
1664                         wfProfileOut( __METHOD__."-misc" );
1665                         wfProfileIn( __METHOD__."-title" );
1666                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki( $link ) );
1667                         if ( $nt === null ) {
1668                                 $s .= $prefix . '[[' . $line;
1669                                 wfProfileOut( __METHOD__."-title" );
1670                                 continue;
1671                         }
1672
1673                         $ns = $nt->getNamespace();
1674                         $iw = $nt->getInterWiki();
1675                         wfProfileOut( __METHOD__."-title" );
1676
1677                         if ( $might_be_img ) { # if this is actually an invalid link
1678                                 wfProfileIn( __METHOD__."-might_be_img" );
1679                                 if ( $ns == NS_FILE && $noforce ) { #but might be an image
1680                                         $found = false;
1681                                         while ( true ) {
1682                                                 #look at the next 'line' to see if we can close it there
1683                                                 $a->next();
1684                                                 $next_line = $a->current();
1685                                                 if ( $next_line === false || $next_line === null ) {
1686                                                         break;
1687                                                 }
1688                                                 $m = explode( ']]', $next_line, 3 );
1689                                                 if ( count( $m ) == 3 ) {
1690                                                         # the first ]] closes the inner link, the second the image
1691                                                         $found = true;
1692                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1693                                                         $trail = $m[2];
1694                                                         break;
1695                                                 } elseif ( count( $m ) == 2 ) {
1696                                                         #if there's exactly one ]] that's fine, we'll keep looking
1697                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1698                                                 } else {
1699                                                         #if $next_line is invalid too, we need look no further
1700                                                         $text .= '[[' . $next_line;
1701                                                         break;
1702                                                 }
1703                                         }
1704                                         if ( !$found ) {
1705                                                 # we couldn't find the end of this imageLink, so output it raw
1706                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1707                                                 $holders->merge( $this->replaceInternalLinks2( $text ) );
1708                                                 $s .= "{$prefix}[[$link|$text";
1709                                                 # note: no $trail, because without an end, there *is* no trail
1710                                                 wfProfileOut( __METHOD__."-might_be_img" );
1711                                                 continue;
1712                                         }
1713                                 } else { #it's not an image, so output it raw
1714                                         $s .= "{$prefix}[[$link|$text";
1715                                         # note: no $trail, because without an end, there *is* no trail
1716                                         wfProfileOut( __METHOD__."-might_be_img" );
1717                                         continue;
1718                                 }
1719                                 wfProfileOut( __METHOD__."-might_be_img" );
1720                         }
1721
1722                         $wasblank = ( $text  == '' );
1723                         if ( $wasblank ) $text = $link;
1724
1725                         # Link not escaped by : , create the various objects
1726                         if ( $noforce ) {
1727
1728                                 # Interwikis
1729                                 wfProfileIn( __METHOD__."-interwiki" );
1730                                 if ( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1731                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1732                                         $s = rtrim($s . $prefix);
1733                                         $s .= trim($trail, "\n") == '' ? '': $prefix . $trail;
1734                                         wfProfileOut( __METHOD__."-interwiki" );
1735                                         continue;
1736                                 }
1737                                 wfProfileOut( __METHOD__."-interwiki" );
1738
1739                                 if ( $ns == NS_FILE ) {
1740                                         wfProfileIn( __METHOD__."-image" );
1741                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1742                                                 if ( $wasblank ) {
1743                                                         # if no parameters were passed, $text
1744                                                         # becomes something like "File:Foo.png",
1745                                                         # which we don't want to pass on to the
1746                                                         # image generator
1747                                                         $text = '';
1748                                                 } else {
1749                                                         # recursively parse links inside the image caption
1750                                                         # actually, this will parse them in any other parameters, too,
1751                                                         # but it might be hard to fix that, and it doesn't matter ATM
1752                                                         $text = $this->replaceExternalLinks($text);
1753                                                         $holders->merge( $this->replaceInternalLinks2( $text ) );
1754                                                 }
1755                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1756                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text, $holders ) ) . $trail;
1757                                         } else {
1758                                                 $s .= $prefix . $trail;
1759                                         }
1760                                         $this->mOutput->addImage( $nt->getDBkey() );
1761                                         wfProfileOut( __METHOD__."-image" );
1762                                         continue;
1763
1764                                 }
1765
1766                                 if ( $ns == NS_CATEGORY ) {
1767                                         wfProfileIn( __METHOD__."-category" );
1768                                         $s = rtrim($s . "\n"); # bug 87
1769
1770                                         if ( $wasblank ) {
1771                                                 $sortkey = $this->getDefaultSort();
1772                                         } else {
1773                                                 $sortkey = $text;
1774                                         }
1775                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1776                                         $sortkey = str_replace( "\n", '', $sortkey );
1777                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1778                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1779
1780                                         /**
1781                                          * Strip the whitespace Category links produce, see bug 87
1782                                          * @todo We might want to use trim($tmp, "\n") here.
1783                                          */
1784                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1785
1786                                         wfProfileOut( __METHOD__."-category" );
1787                                         continue;
1788                                 }
1789                         }
1790
1791                         # Self-link checking
1792                         if( $nt->getFragment() === '' && $ns != NS_SPECIAL ) {
1793                                 if( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1794                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1795                                         continue;
1796                                 }
1797                         }
1798
1799                         # NS_MEDIA is a pseudo-namespace for linking directly to a file
1800                         # FIXME: Should do batch file existence checks, see comment below
1801                         if( $ns == NS_MEDIA ) {
1802                                 wfProfileIn( __METHOD__."-media" );
1803                                 # Give extensions a chance to select the file revision for us
1804                                 $skip = $time = false;
1805                                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$nt, &$skip, &$time ) );
1806                                 if ( $skip ) {
1807                                         $link = $sk->link( $nt );
1808                                 } else {
1809                                         $link = $sk->makeMediaLinkObj( $nt, $text, $time );
1810                                 }
1811                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1812                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1813                                 $this->mOutput->addImage( $nt->getDBkey() );
1814                                 wfProfileOut( __METHOD__."-media" );
1815                                 continue;
1816                         }
1817
1818                         wfProfileIn( __METHOD__."-always_known" );
1819                         # Some titles, such as valid special pages or files in foreign repos, should
1820                         # be shown as bluelinks even though they're not included in the page table
1821                         #
1822                         # FIXME: isAlwaysKnown() can be expensive for file links; we should really do
1823                         # batch file existence checks for NS_FILE and NS_MEDIA
1824                         if( $iw == '' && $nt->isAlwaysKnown() ) {
1825                                 $this->mOutput->addLink( $nt );
1826                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1827                         } else {
1828                                 # Links will be added to the output link list after checking
1829                                 $s .= $holders->makeHolder( $nt, $text, '', $trail, $prefix );
1830                         }
1831                         wfProfileOut( __METHOD__."-always_known" );
1832                 }
1833                 wfProfileOut( __METHOD__ );
1834                 return $holders;
1835         }
1836
1837         /**
1838          * Make a link placeholder. The text returned can be later resolved to a real link with
1839          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1840          * parsing of interwiki links, and secondly to allow all existence checks and
1841          * article length checks (for stub links) to be bundled into a single query.
1842          *
1843          * @deprecated
1844          */
1845         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1846                 return $this->mLinkHolders->makeHolder( $nt, $text, $query, $trail, $prefix );
1847         }
1848
1849         /**
1850          * Render a forced-blue link inline; protect against double expansion of
1851          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1852          * Since this little disaster has to split off the trail text to avoid
1853          * breaking URLs in the following text without breaking trails on the
1854          * wiki links, it's been made into a horrible function.
1855          *
1856          * @param Title $nt
1857          * @param string $text
1858          * @param string $query
1859          * @param string $trail
1860          * @param string $prefix
1861          * @return string HTML-wikitext mix oh yuck
1862          */
1863         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1864                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1865                 $sk = $this->mOptions->getSkin();
1866                 // FIXME: use link() instead of deprecated makeKnownLinkObj()
1867                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1868                 return $this->armorLinks( $link ) . $trail;
1869         }
1870
1871         /**
1872          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1873          * going to go through further parsing steps before inline URL expansion.
1874          *
1875          * Not needed quite as much as it used to be since free links are a bit
1876          * more sensible these days. But bracketed links are still an issue.
1877          *
1878          * @param string more-or-less HTML
1879          * @return string less-or-more HTML with NOPARSE bits
1880          */
1881         function armorLinks( $text ) {
1882                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1883                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1884         }
1885
1886         /**
1887          * Return true if subpage links should be expanded on this page.
1888          * @return bool
1889          */
1890         function areSubpagesAllowed() {
1891                 # Some namespaces don't allow subpages
1892                 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
1893         }
1894
1895         /**
1896          * Handle link to subpage if necessary
1897          * @param string $target the source of the link
1898          * @param string &$text the link text, modified as necessary
1899          * @return string the full name of the link
1900          * @private
1901          */
1902         function maybeDoSubpageLink($target, &$text) {
1903                 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
1904         }
1905
1906         /**
1907          * Returns valid title characters and namespace characters for pipe trick.
1908          *
1909          * FIXME: the namespace characters should not be specified like this...
1910          */
1911         static function getPipeTrickCharacterClasses() {
1912                 global $wgLegalTitleChars;
1913                 return  array( "[$wgLegalTitleChars]", '[ _0-9A-Za-z\x80-\xff-]' );
1914         }
1915
1916         /**
1917          * From the [[title|]] return link-text as though the used typed [[title|link-text]]
1918          *
1919          * For most links this be as though the user typed [[ns:title|title]]
1920          * However [[ns:title (context)|]], [[ns:title, context|]] and [[ns:title (context), context|]]
1921          * [[#title (context)|]] [[../context/title (context), context|]]
1922          * all return the |title]] with no context or indicative punctuation.
1923          */
1924         function getPipeTrickText( $link ) {
1925                 static $rexps = FALSE;
1926                 if( !$rexps ) {
1927                         list( $tc, $nc ) = Parser::getPipeTrickCharacterClasses();
1928                         $rexps = array (
1929                                 # try this first, to turn "[[A, B (C)|]]" into "A, B"
1930                                 "/^(:?$nc+:|[:#\/]|$tc+\\/|)($tc+?)( \\($tc+\\)| （$tc+）)$/", # [[ns:page (context)|]]
1931                                 "/^(:?$nc+:|[:#\/]|$tc+\\/|)($tc+?)( \\($tc+\\)| （$tc+）|)((?:,|，) $tc+|)$/",  # [[ns:page (context), context|]]
1932                         );
1933                 }
1934                 $text = urldecode( $link );
1935
1936                 for( $i = 0; $i < count( $rexps ); $i++) {
1937                         if( preg_match( $rexps[$i], $text, $m ) )
1938                                 return $m[2];
1939                 }
1940                 return $text;
1941         }
1942
1943         /**
1944          * From the [[|link-text]] return the title as though the user typed [[title|link-text]]
1945          *
1946          * On most pages this will return link-text or "" if the link-text is not a valid title
1947          * On pages like [[ns:title (context)]] and [[ns:title, context]] it will act like
1948          * [[ns:link-text (context)|link-text]] and [[ns:link-text, context|link-text]]
1949          */
1950         function getPipeTrickLink( $text ) {
1951                 static $rexps = FALSE, $tc;
1952                 if( !$rexps ) {
1953                         list( $tc, $nc ) = Parser::getPipeTrickCharacterClasses();
1954                         $rexps = array (
1955                                 "/^($nc+:|)$tc+?( \\($tc+\\)| （$tc+）)$/", # [[ns:page (context)|]]
1956                                 "/^($nc+:|)$tc+?(?:(?: \\($tc+\\)| （$tc+）|)((?:,|，) $tc+|))$/"  # [[ns:page (context), context|]]
1957                         );
1958                 }
1959
1960                 if( !preg_match( "/^$tc+$/", $text ) )
1961                         return '';
1962
1963                 $t = $this->mTitle->getText();
1964
1965                 for( $i = 0; $i < count( $rexps ); $i++) {
1966                         if( preg_match( $rexps[$i], $t, $m ) )
1967                                 return "$m[1]$text$m[2]";
1968                 }
1969                 return $text;
1970         }
1971
1972         /**#@+
1973          * Used by doBlockLevels()
1974          * @private
1975          */
1976         /* private */ function closeParagraph() {
1977                 $result = '';
1978                 if ( $this->mLastSection != '' ) {
1979                         $result = '</' . $this->mLastSection  . ">\n";
1980                 }
1981                 $this->mInPre = false;
1982                 $this->mLastSection = '';
1983                 return $result;
1984         }
1985         # getCommon() returns the length of the longest common substring
1986         # of both arguments, starting at the beginning of both.
1987         #
1988         /* private */ function getCommon( $st1, $st2 ) {
1989                 $fl = strlen( $st1 );
1990                 $shorter = strlen( $st2 );
1991                 if ( $fl < $shorter ) { $shorter = $fl; }
1992
1993                 for ( $i = 0; $i < $shorter; ++$i ) {
1994                         if ( $st1{$i} != $st2{$i} ) { break; }
1995                 }
1996                 return $i;
1997         }
1998         # These next three functions open, continue, and close the list
1999         # element appropriate to the prefix character passed into them.
2000         #
2001         /* private */ function openList( $char ) {
2002                 $result = $this->closeParagraph();
2003
2004                 if ( '*' === $char ) { $result .= '<ul><li>'; }
2005                 elseif ( '#' === $char ) { $result .= '<ol><li>'; }
2006                 elseif ( ':' === $char ) { $result .= '<dl><dd>'; }
2007                 elseif ( ';' === $char ) {
2008                         $result .= '<dl><dt>';
2009                         $this->mDTopen = true;
2010                 }
2011                 else { $result = '<!-- ERR 1 -->'; }
2012
2013                 return $result;
2014         }
2015
2016         /* private */ function nextItem( $char ) {
2017                 if ( '*' === $char || '#' === $char ) { return '</li><li>'; }
2018                 elseif ( ':' === $char || ';' === $char ) {
2019                         $close = '</dd>';
2020                         if ( $this->mDTopen ) { $close = '</dt>'; }
2021                         if ( ';' === $char ) {
2022                                 $this->mDTopen = true;
2023                                 return $close . '<dt>';
2024                         } else {
2025                                 $this->mDTopen = false;
2026                                 return $close . '<dd>';
2027                         }
2028                 }
2029                 return '<!-- ERR 2 -->';
2030         }
2031
2032         /* private */ function closeList( $char ) {
2033                 if ( '*' === $char ) { $text = '</li></ul>'; }
2034                 elseif ( '#' === $char ) { $text = '</li></ol>'; }
2035                 elseif ( ':' === $char ) {
2036                         if ( $this->mDTopen ) {
2037                                 $this->mDTopen = false;
2038                                 $text = '</dt></dl>';
2039                         } else {
2040                                 $text = '</dd></dl>';
2041                         }
2042                 }
2043                 else {  return '<!-- ERR 3 -->'; }
2044                 return $text."\n";
2045         }
2046         /**#@-*/
2047
2048         /**
2049          * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2050          *
2051          * @param $linestart bool whether or not this is at the start of a line.
2052          * @private
2053          * @return string the lists rendered as HTML
2054          */
2055         function doBlockLevels( $text, $linestart ) {
2056                 wfProfileIn( __METHOD__ );
2057
2058                 # Parsing through the text line by line.  The main thing
2059                 # happening here is handling of block-level elements p, pre,
2060                 # and making lists from lines starting with * # : etc.
2061                 #
2062                 $textLines = StringUtils::explode( "\n", $text );
2063
2064                 $lastPrefix = $output = '';
2065                 $this->mDTopen = $inBlockElem = false;
2066                 $prefixLength = 0;
2067                 $paragraphStack = false;
2068
2069                 foreach ( $textLines as $oLine ) {
2070                         # Fix up $linestart
2071                         if ( !$linestart ) {
2072                                 $output .= $oLine;
2073                                 $linestart = true;
2074                                 continue;
2075                         }
2076                         // * = ul
2077                         // # = ol
2078                         // ; = dt
2079                         // : = dd
2080
2081                         $lastPrefixLength = strlen( $lastPrefix );
2082                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
2083                         $preOpenMatch = preg_match('/<pre/i', $oLine );
2084                         // If not in a <pre> element, scan for and figure out what prefixes are there.
2085                         if ( !$this->mInPre ) {
2086                                 # Multiple prefixes may abut each other for nested lists.
2087                                 $prefixLength = strspn( $oLine, '*#:;' );
2088                                 $prefix = substr( $oLine, 0, $prefixLength );
2089
2090                                 # eh?
2091                                 // ; and : are both from definition-lists, so they're equivalent
2092                                 //  for the purposes of determining whether or not we need to open/close
2093                                 //  elements.
2094                                 $prefix2 = str_replace( ';', ':', $prefix );
2095                                 $t = substr( $oLine, $prefixLength );
2096                                 $this->mInPre = (bool)$preOpenMatch;
2097                         } else {
2098                                 # Don't interpret any other prefixes in preformatted text
2099                                 $prefixLength = 0;
2100                                 $prefix = $prefix2 = '';
2101                                 $t = $oLine;
2102                         }
2103
2104                         # List generation
2105                         if( $prefixLength && $lastPrefix === $prefix2 ) {
2106                                 # Same as the last item, so no need to deal with nesting or opening stuff
2107                                 $output .= $this->nextItem( substr( $prefix, -1 ) );
2108                                 $paragraphStack = false;
2109
2110                                 if ( substr( $prefix, -1 ) === ';') {
2111                                         # The one nasty exception: definition lists work like this:
2112                                         # ; title : definition text
2113                                         # So we check for : in the remainder text to split up the
2114                                         # title and definition, without b0rking links.
2115                                         $term = $t2 = '';
2116                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2117                                                 $t = $t2;
2118                                                 $output .= $term . $this->nextItem( ':' );
2119                                         }
2120                                 }
2121                         } elseif( $prefixLength || $lastPrefixLength ) {
2122                                 // We need to open or close prefixes, or both.
2123
2124                                 # Either open or close a level...
2125                                 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2126                                 $paragraphStack = false;
2127
2128                                 // Close all the prefixes which aren't shared.
2129                                 while( $commonPrefixLength < $lastPrefixLength ) {
2130                                         $output .= $this->closeList( $lastPrefix[$lastPrefixLength-1] );
2131                                         --$lastPrefixLength;
2132                                 }
2133
2134                                 // Continue the current prefix if appropriate.
2135                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2136                                         $output .= $this->nextItem( $prefix[$commonPrefixLength-1] );
2137                                 }
2138
2139                                 // Open prefixes where appropriate.
2140                                 while ( $prefixLength > $commonPrefixLength ) {
2141                                         $char = substr( $prefix, $commonPrefixLength, 1 );
2142                                         $output .= $this->openList( $char );
2143
2144                                         if ( ';' === $char ) {
2145                                                 # FIXME: This is dupe of code above
2146                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2147                                                         $t = $t2;
2148                                                         $output .= $term . $this->nextItem( ':' );
2149                                                 }
2150                                         }
2151                                         ++$commonPrefixLength;
2152                                 }
2153                                 $lastPrefix = $prefix2;
2154                         }
2155
2156                         // If we have no prefixes, go to paragraph mode.
2157                         if( 0 == $prefixLength ) {
2158                                 wfProfileIn( __METHOD__."-paragraph" );
2159                                 # No prefix (not in list)--go to paragraph mode
2160                                 // XXX: use a stack for nestable elements like span, table and div
2161                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2162                                 $closematch = preg_match(
2163                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2164                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2165                                 if ( $openmatch or $closematch ) {
2166                                         $paragraphStack = false;
2167                                         # TODO bug 5718: paragraph closed
2168                                         $output .= $this->closeParagraph();
2169                                         if ( $preOpenMatch and !$preCloseMatch ) {
2170                                                 $this->mInPre = true;
2171                                         }
2172                                         if ( $closematch ) {
2173                                                 $inBlockElem = false;
2174                                         } else {
2175                                                 $inBlockElem = true;
2176                                         }
2177                                 } else if ( !$inBlockElem && !$this->mInPre ) {
2178                                         if ( ' ' == substr( $t, 0, 1 ) and ( $this->mLastSection === 'pre' or trim($t) != '' ) ) {
2179                                                 // pre
2180                                                 if ($this->mLastSection !== 'pre') {
2181                                                         $paragraphStack = false;
2182                                                         $output .= $this->closeParagraph().'<pre>';
2183                                                         $this->mLastSection = 'pre';
2184                                                 }
2185                                                 $t = substr( $t, 1 );
2186                                         } else {
2187                                                 // paragraph
2188                                                 if ( trim($t) == '' ) {
2189                                                         if ( $paragraphStack ) {
2190                                                                 $output .= $paragraphStack.'<br />';
2191                                                                 $paragraphStack = false;
2192                                                                 $this->mLastSection = 'p';
2193                                                         } else {
2194                                                                 if ($this->mLastSection !== 'p' ) {
2195                                                                         $output .= $this->closeParagraph();
2196                                                                         $this->mLastSection = '';
2197                                                                         $paragraphStack = '<p>';
2198                                                                 } else {
2199                                                                         $paragraphStack = '</p><p>';
2200                                                                 }
2201                                                         }
2202                                                 } else {
2203                                                         if ( $paragraphStack ) {
2204                                                                 $output .= $paragraphStack;
2205                                                                 $paragraphStack = false;
2206                                                                 $this->mLastSection = 'p';
2207                                                         } else if ($this->mLastSection !== 'p') {
2208                                                                 $output .= $this->closeParagraph().'<p>';
2209                                                                 $this->mLastSection = 'p';
2210                                                         }
2211                                                 }
2212                                         }
2213                                 }
2214                                 wfProfileOut( __METHOD__."-paragraph" );
2215                         }
2216                         // somewhere above we forget to get out of pre block (bug 785)
2217                         if($preCloseMatch && $this->mInPre) {
2218                                 $this->mInPre = false;
2219                         }
2220                         if ($paragraphStack === false) {
2221                                 $output .= $t."\n";
2222                         }
2223                 }
2224                 while ( $prefixLength ) {
2225                         $output .= $this->closeList( $prefix2[$prefixLength-1] );
2226                         --$prefixLength;
2227                 }
2228                 if ( $this->mLastSection != '' ) {
2229                         $output .= '</' . $this->mLastSection . '>';
2230                         $this->mLastSection = '';
2231                 }
2232
2233                 wfProfileOut( __METHOD__ );
2234                 return $output;
2235         }
2236
2237         /**
2238          * Split up a string on ':', ignoring any occurences inside tags
2239          * to prevent illegal overlapping.
2240          * @param string $str the string to split
2241          * @param string &$before set to everything before the ':'
2242          * @param string &$after set to everything after the ':'
2243          * return string the position of the ':', or false if none found
2244          */
2245         function findColonNoLinks($str, &$before, &$after) {
2246                 wfProfileIn( __METHOD__ );
2247
2248                 $pos = strpos( $str, ':' );
2249                 if( $pos === false ) {
2250                         // Nothing to find!
2251                         wfProfileOut( __METHOD__ );
2252                         return false;
2253                 }
2254
2255                 $lt = strpos( $str, '<' );
2256                 if( $lt === false || $lt > $pos ) {
2257                         // Easy; no tag nesting to worry about
2258                         $before = substr( $str, 0, $pos );
2259                         $after = substr( $str, $pos+1 );
2260                         wfProfileOut( __METHOD__ );
2261                         return $pos;
2262                 }
2263
2264                 // Ugly state machine to walk through avoiding tags.
2265                 $state = self::COLON_STATE_TEXT;
2266                 $stack = 0;
2267                 $len = strlen( $str );
2268                 for( $i = 0; $i < $len; $i++ ) {
2269                         $c = $str{$i};
2270
2271                         switch( $state ) {
2272                         // (Using the number is a performance hack for common cases)
2273                         case 0: // self::COLON_STATE_TEXT:
2274                                 switch( $c ) {
2275                                 case "<":
2276                                         // Could be either a <start> tag or an </end> tag
2277                                         $state = self::COLON_STATE_TAGSTART;
2278                                         break;
2279                                 case ":":
2280                                         if( $stack == 0 ) {
2281                                                 // We found it!
2282                                                 $before = substr( $str, 0, $i );
2283                                                 $after = substr( $str, $i + 1 );
2284                                                 wfProfileOut( __METHOD__ );
2285                                                 return $i;
2286                                         }
2287                                         // Embedded in a tag; don't break it.
2288                                         break;
2289                                 default:
2290                                         // Skip ahead looking for something interesting
2291                                         $colon = strpos( $str, ':', $i );
2292                                         if( $colon === false ) {
2293                                                 // Nothing else interesting
2294                                                 wfProfileOut( __METHOD__ );
2295                                                 return false;
2296                                         }
2297                                         $lt = strpos( $str, '<', $i );
2298                                         if( $stack === 0 ) {
2299                                                 if( $lt === false || $colon < $lt ) {
2300                                                         // We found it!
2301                                                         $before = substr( $str, 0, $colon );
2302                                                         $after = substr( $str, $colon + 1 );
2303                                                         wfProfileOut( __METHOD__ );
2304                                                         return $i;
2305                                                 }
2306                                         }
2307                                         if( $lt === false ) {
2308                                                 // Nothing else interesting to find; abort!
2309                                                 // We're nested, but there's no close tags left. Abort!
2310                                                 break 2;
2311                                         }
2312                                         // Skip ahead to next tag start
2313                                         $i = $lt;
2314                                         $state = self::COLON_STATE_TAGSTART;
2315                                 }
2316                                 break;
2317                         case 1: // self::COLON_STATE_TAG:
2318                                 // In a <tag>
2319                                 switch( $c ) {
2320                                 case ">":
2321                                         $stack++;
2322                                         $state = self::COLON_STATE_TEXT;
2323                                         break;
2324                                 case "/":
2325                                         // Slash may be followed by >?
2326                                         $state = self::COLON_STATE_TAGSLASH;
2327                                         break;
2328                                 default:
2329                                         // ignore
2330                                 }
2331                                 break;
2332                         case 2: // self::COLON_STATE_TAGSTART:
2333                                 switch( $c ) {
2334                                 case "/":
2335                                         $state = self::COLON_STATE_CLOSETAG;
2336                                         break;
2337                                 case "!":
2338                                         $state = self::COLON_STATE_COMMENT;
2339                                         break;
2340                                 case ">":
2341                                         // Illegal early close? This shouldn't happen D:
2342                                         $state = self::COLON_STATE_TEXT;
2343                                         break;
2344                                 default:
2345                                         $state = self::COLON_STATE_TAG;
2346                                 }
2347                                 break;
2348                         case 3: // self::COLON_STATE_CLOSETAG:
2349                                 // In a </tag>
2350                                 if( $c === ">" ) {
2351                                         $stack--;
2352                                         if( $stack < 0 ) {
2353                                                 wfDebug( __METHOD__.": Invalid input; too many close tags\n" );
2354                                                 wfProfileOut( __METHOD__ );
2355                                                 return false;
2356                                         }
2357                                         $state = self::COLON_STATE_TEXT;
2358                                 }
2359                                 break;
2360                         case self::COLON_STATE_TAGSLASH:
2361                                 if( $c === ">" ) {
2362                                         // Yes, a self-closed tag <blah/>
2363                                         $state = self::COLON_STATE_TEXT;
2364                                 } else {
2365                                         // Probably we're jumping the gun, and this is an attribute
2366                                         $state = self::COLON_STATE_TAG;
2367                                 }
2368                                 break;
2369                         case 5: // self::COLON_STATE_COMMENT:
2370                                 if( $c === "-" ) {
2371                                         $state = self::COLON_STATE_COMMENTDASH;
2372                                 }
2373                                 break;
2374                         case self::COLON_STATE_COMMENTDASH:
2375                                 if( $c === "-" ) {
2376                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2377                                 } else {
2378                                         $state = self::COLON_STATE_COMMENT;
2379                                 }
2380                                 break;
2381                         case self::COLON_STATE_COMMENTDASHDASH:
2382                                 if( $c === ">" ) {
2383                                         $state = self::COLON_STATE_TEXT;
2384                                 } else {
2385                                         $state = self::COLON_STATE_COMMENT;
2386                                 }
2387                                 break;
2388                         default:
2389                                 throw new MWException( "State machine error in " . __METHOD__ );
2390                         }
2391                 }
2392                 if( $stack > 0 ) {
2393                         wfDebug( __METHOD__.": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2394                         return false;
2395                 }
2396                 wfProfileOut( __METHOD__ );
2397                 return false;
2398         }
2399
2400         /**
2401          * Return value of a magic variable (like PAGENAME)
2402          *
2403          * @private
2404          */
2405         function getVariableValue( $index, $frame=false ) {
2406                 global $wgContLang, $wgSitename, $wgServer, $wgServerName;
2407                 global $wgScriptPath, $wgStylePath;
2408
2409                 /**
2410                  * Some of these require message or data lookups and can be
2411                  * expensive to check many times.
2412                  */
2413                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2414                         if ( isset( $this->mVarCache[$index] ) ) {
2415                                 return $this->mVarCache[$index];
2416                         }
2417                 }
2418
2419                 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2420                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2421
2422                 # Use the time zone
2423                 global $wgLocaltimezone;
2424                 if ( isset( $wgLocaltimezone ) ) {
2425                         $oldtz = date_default_timezone_get();
2426                         date_default_timezone_set( $wgLocaltimezone );
2427                 }
2428
2429                 $localTimestamp = date( 'YmdHis', $ts );
2430                 $localMonth = date( 'm', $ts );
2431                 $localMonth1 = date( 'n', $ts );
2432                 $localMonthName = date( 'n', $ts );
2433                 $localDay = date( 'j', $ts );
2434                 $localDay2 = date( 'd', $ts );
2435                 $localDayOfWeek = date( 'w', $ts );
2436                 $localWeek = date( 'W', $ts );
2437                 $localYear = date( 'Y', $ts );
2438                 $localHour = date( 'H', $ts );
2439                 if ( isset( $wgLocaltimezone ) ) {
2440                         date_default_timezone_set( $oldtz );
2441                 }
2442
2443                 switch ( $index ) {
2444                         case 'currentmonth':
2445                                 $value = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2446                                 break;
2447                         case 'currentmonth1':
2448                                 $value = $wgContLang->formatNum( gmdate( 'n', $ts ) );
2449                                 break;
2450                         case 'currentmonthname':
2451                                 $value = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2452                                 break;
2453                         case 'currentmonthnamegen':
2454                                 $value = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2455                                 break;
2456                         case 'currentmonthabbrev':
2457                                 $value = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2458                                 break;
2459                         case 'currentday':
2460                                 $value = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2461                                 break;
2462                         case 'currentday2':
2463                                 $value = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2464                                 break;
2465                         case 'localmonth':
2466                                 $value = $wgContLang->formatNum( $localMonth );
2467                                 break;
2468                         case 'localmonth1':
2469                                 $value = $wgContLang->formatNum( $localMonth1 );
2470                                 break;
2471                         case 'localmonthname':
2472                                 $value = $wgContLang->getMonthName( $localMonthName );
2473                                 break;
2474                         case 'localmonthnamegen':
2475                                 $value = $wgContLang->getMonthNameGen( $localMonthName );
2476                                 break;
2477                         case 'localmonthabbrev':
2478                                 $value = $wgContLang->getMonthAbbreviation( $localMonthName );
2479                                 break;
2480                         case 'localday':
2481                                 $value = $wgContLang->formatNum( $localDay );
2482                                 break;
2483                         case 'localday2':
2484                                 $value = $wgContLang->formatNum( $localDay2 );
2485                                 break;
2486                         case 'pagename':
2487                                 $value = wfEscapeWikiText( $this->mTitle->getText() );
2488                                 break;
2489                         case 'pagenamee':
2490                                 $value = $this->mTitle->getPartialURL();
2491                                 break;
2492                         case 'fullpagename':
2493                                 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2494                                 break;
2495                         case 'fullpagenamee':
2496                                 $value = $this->mTitle->getPrefixedURL();
2497                                 break;
2498                         case 'subpagename':
2499                                 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2500                                 break;
2501                         case 'subpagenamee':
2502                                 $value = $this->mTitle->getSubpageUrlForm();
2503                                 break;
2504                         case 'basepagename':
2505                                 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2506                                 break;
2507                         case 'basepagenamee':
2508                                 $value = wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2509                                 break;
2510                         case 'talkpagename':
2511                                 if( $this->mTitle->canTalk() ) {
2512                                         $talkPage = $this->mTitle->getTalkPage();
2513                                         $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2514                                 } else {
2515                                         $value = '';
2516                                 }
2517                                 break;
2518                         case 'talkpagenamee':
2519                                 if( $this->mTitle->canTalk() ) {
2520                                         $talkPage = $this->mTitle->getTalkPage();
2521                                         $value = $talkPage->getPrefixedUrl();
2522                                 } else {
2523                                         $value = '';
2524                                 }
2525                                 break;
2526                         case 'subjectpagename':
2527                                 $subjPage = $this->mTitle->getSubjectPage();
2528                                 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2529                                 break;
2530                         case 'subjectpagenamee':
2531                                 $subjPage = $this->mTitle->getSubjectPage();
2532                                 $value = $subjPage->getPrefixedUrl();
2533                                 break;
2534                         case 'revisionid':
2535                                 // Let the edit saving system know we should parse the page
2536                                 // *after* a revision ID has been assigned.
2537                                 $this->mOutput->setFlag( 'vary-revision' );
2538                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2539                                 $value = $this->mRevisionId;
2540                                 break;
2541                         case 'revisionday':
2542                                 // Let the edit saving system know we should parse the page
2543                                 // *after* a revision ID has been assigned. This is for null edits.
2544                                 $this->mOutput->setFlag( 'vary-revision' );
2545                                 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2546                                 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2547                                 break;
2548                         case 'revisionday2':
2549                                 // Let the edit saving system know we should parse the page
2550                                 // *after* a revision ID has been assigned. This is for null edits.
2551                                 $this->mOutput->setFlag( 'vary-revision' );
2552                                 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2553                                 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2554                                 break;
2555                         case 'revisionmonth':
2556                                 // Let the edit saving system know we should parse the page
2557                                 // *after* a revision ID has been assigned. This is for null edits.
2558                                 $this->mOutput->setFlag( 'vary-revision' );
2559                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2560                                 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2561                                 break;
2562                         case 'revisionyear':
2563                                 // Let the edit saving system know we should parse the page
2564                                 // *after* a revision ID has been assigned. This is for null edits.
2565                                 $this->mOutput->setFlag( 'vary-revision' );
2566                                 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2567                                 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2568                                 break;
2569                         case 'revisiontimestamp':
2570                                 // Let the edit saving system know we should parse the page
2571                                 // *after* a revision ID has been assigned. This is for null edits.
2572                                 $this->mOutput->setFlag( 'vary-revision' );
2573                                 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2574                                 $value = $this->getRevisionTimestamp();
2575                                 break;
2576                         case 'revisionuser':
2577                                 // Let the edit saving system know we should parse the page
2578                                 // *after* a revision ID has been assigned. This is for null edits.
2579                                 $this->mOutput->setFlag( 'vary-revision' );
2580                                 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-revision...\n" );
2581                                 $value = $this->getRevisionUser();
2582                                 break;
2583                         case 'namespace':
2584                                 $value = str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2585                                 break;
2586                         case 'namespacee':
2587                                 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2588                                 break;
2589                         case 'talkspace':
2590                                 $value = $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2591                                 break;
2592                         case 'talkspacee':
2593                                 $value = $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2594                                 break;
2595                         case 'subjectspace':
2596                                 $value = $this->mTitle->getSubjectNsText();
2597                                 break;
2598                         case 'subjectspacee':
2599                                 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2600                                 break;
2601                         case 'currentdayname':
2602                                 $value = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2603                                 break;
2604                         case 'currentyear':
2605                                 $value = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2606                                 break;
2607                         case 'currenttime':
2608                                 $value = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2609                                 break;
2610                         case 'currenthour':
2611                                 $value = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2612                                 break;
2613                         case 'currentweek':
2614                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2615                                 // int to remove the padding
2616                                 $value = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2617                                 break;
2618                         case 'currentdow':
2619                                 $value = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2620                                 break;
2621                         case 'localdayname':
2622                                 $value = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2623                                 break;
2624                         case 'localyear':
2625                                 $value = $wgContLang->formatNum( $localYear, true );
2626                                 break;
2627                         case 'localtime':
2628                                 $value = $wgContLang->time( $localTimestamp, false, false );
2629                                 break;
2630                         case 'localhour':
2631                                 $value = $wgContLang->formatNum( $localHour, true );
2632                                 break;
2633                         case 'localweek':
2634                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2635                                 // int to remove the padding
2636                                 $value = $wgContLang->formatNum( (int)$localWeek );
2637                                 break;
2638                         case 'localdow':
2639                                 $value = $wgContLang->formatNum( $localDayOfWeek );
2640                                 break;
2641                         case 'numberofarticles':
2642                                 $value = $wgContLang->formatNum( SiteStats::articles() );
2643                                 break;
2644                         case 'numberoffiles':
2645                                 $value = $wgContLang->formatNum( SiteStats::images() );
2646                                 break;
2647                         case 'numberofusers':
2648                                 $value = $wgContLang->formatNum( SiteStats::users() );
2649                                 break;
2650                         case 'numberofactiveusers':
2651                                 $value = $wgContLang->formatNum( SiteStats::activeUsers() );
2652                                 break;
2653                         case 'numberofpages':
2654                                 $value = $wgContLang->formatNum( SiteStats::pages() );
2655                                 break;
2656                         case 'numberofadmins':
2657                                 $value = $wgContLang->formatNum( SiteStats::numberingroup('sysop') );
2658                                 break;
2659                         case 'numberofedits':
2660                                 $value = $wgContLang->formatNum( SiteStats::edits() );
2661                                 break;
2662                         case 'numberofviews':
2663                                 $value = $wgContLang->formatNum( SiteStats::views() );
2664                                 break;
2665                         case 'currenttimestamp':
2666                                 $value = wfTimestamp( TS_MW, $ts );
2667                                 break;
2668                         case 'localtimestamp':
2669                                 $value = $localTimestamp;
2670                                 break;
2671                         case 'currentversion':
2672                                 $value = SpecialVersion::getVersion();
2673                                 break;
2674                         case 'sitename':
2675                                 return $wgSitename;
2676                         case 'server':
2677                                 return $wgServer;
2678                         case 'servername':
2679                                 return $wgServerName;
2680                         case 'scriptpath':
2681                                 return $wgScriptPath;
2682                         case 'stylepath':
2683                                 return $wgStylePath;
2684                         case 'directionmark':
2685                                 return $wgContLang->getDirMark();
2686                         case 'contentlanguage':
2687                                 global $wgContLanguageCode;
2688                                 return $wgContLanguageCode;
2689                         default:
2690                                 $ret = null;
2691                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret, &$frame ) ) )
2692                                         return $ret;
2693                                 else
2694                                         return null;
2695                 }
2696
2697                 if ( $index )
2698                         $this->mVarCache[$index] = $value;
2699
2700                 return $value;
2701         }
2702
2703         /**
2704          * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2705          *
2706          * @private
2707          */
2708         function initialiseVariables() {
2709                 wfProfileIn( __METHOD__ );
2710                 $variableIDs = MagicWord::getVariableIDs();
2711                 $substIDs = MagicWord::getSubstIDs();
2712
2713                 $this->mVariables = new MagicWordArray( $variableIDs );
2714                 $this->mSubsts = new MagicWordArray( $substIDs );
2715                 wfProfileOut( __METHOD__ );
2716         }
2717
2718         /**
2719          * Preprocess some wikitext and return the document tree.
2720          * This is the ghost of replace_variables().
2721          *
2722          * @param string $text The text to parse
2723          * @param integer flags Bitwise combination of:
2724          *          self::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
2725          *                                     included. Default is to assume a direct page view.
2726          *
2727          * The generated DOM tree must depend only on the input text and the flags.
2728          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2729          *
2730          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2731          * change in the DOM tree for a given text, must be passed through the section identifier
2732          * in the section edit link and thus back to extractSections().
2733          *
2734          * The output of this function is currently only cached in process memory, but a persistent
2735          * cache may be implemented at a later date which takes further advantage of these strict
2736          * dependency requirements.
2737          *
2738          * @private
2739          */
2740         function preprocessToDom ( $text, $flags = 0 ) {
2741                 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2742                 return $dom;
2743         }
2744
2745         /*
2746          * Return a three-element array: leading whitespace, string contents, trailing whitespace
2747          */
2748         public static function splitWhitespace( $s ) {
2749                 $ltrimmed = ltrim( $s );
2750                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2751                 $trimmed = rtrim( $ltrimmed );
2752                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2753                 if ( $diff > 0 ) {
2754                         $w2 = substr( $ltrimmed, -$diff );
2755                 } else {
2756                         $w2 = '';
2757                 }
2758                 return array( $w1, $trimmed, $w2 );
2759         }
2760
2761         /**
2762          * Replace magic variables, templates, and template arguments
2763          * with the appropriate text. Templates are substituted recursively,
2764          * taking care to avoid infinite loops.
2765          *
2766          * Note that the substitution depends on value of $mOutputType:
2767          *  self::OT_WIKI: only {{subst:}} templates
2768          *  self::OT_PREPROCESS: templates but not extension tags
2769          *  self::OT_HTML: all templates and extension tags
2770          *
2771          * @param string $tex The text to transform
2772          * @param PPFrame $frame Object describing the arguments passed to the template.
2773          *        Arguments may also be provided as an associative array, as was the usual case before MW1.12.
2774          *        Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
2775          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2776          * @private
2777          */
2778         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2779                 # Is there any text? Also, Prevent too big inclusions!
2780                 if ( strlen( $text ) < 1 || strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2781                         return $text;
2782                 }
2783                 wfProfileIn( __METHOD__ );
2784
2785                 if ( $frame === false ) {
2786                         $frame = $this->getPreprocessor()->newFrame();
2787                 } elseif ( !( $frame instanceof PPFrame ) ) {
2788                         wfDebug( __METHOD__." called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
2789                         $frame = $this->getPreprocessor()->newCustomFrame($frame);
2790                 }
2791
2792                 $dom = $this->preprocessToDom( $text );
2793                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2794                 $text = $frame->expand( $dom, $flags );
2795
2796                 wfProfileOut( __METHOD__ );
2797                 return $text;
2798         }
2799
2800         /// Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2801         static function createAssocArgs( $args ) {
2802                 $assocArgs = array();
2803                 $index = 1;
2804                 foreach( $args as $arg ) {
2805                         $eqpos = strpos( $arg, '=' );
2806                         if ( $eqpos === false ) {
2807                                 $assocArgs[$index++] = $arg;
2808                         } else {
2809                                 $name = trim( substr( $arg, 0, $eqpos ) );
2810                                 $value = trim( substr( $arg, $eqpos+1 ) );
2811                                 if ( $value === false ) {
2812                                         $value = '';
2813                                 }
2814                                 if ( $name !== false ) {
2815                                         $assocArgs[$name] = $value;
2816                                 }
2817                         }
2818                 }
2819
2820                 return $assocArgs;
2821         }
2822
2823         /**
2824          * Warn the user when a parser limitation is reached
2825          * Will warn at most once the user per limitation type
2826          *
2827          * @param string $limitationType, should be one of:
2828          *   'expensive-parserfunction' (corresponding messages: 'expensive-parserfunction-warning', 'expensive-parserfunction-category')
2829          *   'post-expand-template-argument' (corresponding messages: 'post-expand-template-argument-warning', 'post-expand-template-argument-category')
2830          *   'post-expand-template-inclusion' (corresponding messages: 'post-expand-template-inclusion-warning', 'post-expand-template-inclusion-category')
2831          * @params int $current, $max When an explicit limit has been
2832          *       exceeded, provide the values (optional)
2833          */
2834         function limitationWarn( $limitationType, $current=null, $max=null) {
2835                 //does no harm if $current and $max are present but are unnecessary for the message
2836                 $warning = wfMsgExt( "$limitationType-warning", array( 'parsemag', 'escape' ), $current, $max );
2837                 $this->mOutput->addWarning( $warning );
2838                 $this->addTrackingCategory( "$limitationType-category" );
2839         }
2840
2841         /**
2842          * Return the text of a template, after recursively
2843          * replacing any variables or templates within the template.
2844          *
2845          * @param array $piece The parts of the template
2846          *  $piece['title']: the title, i.e. the part before the |
2847          *  $piece['parts']: the parameter array
2848          *  $piece['lineStart']: whether the brace was at the start of a line
2849          * @param PPFrame The current frame, contains template arguments
2850          * @return string the text of the template
2851          * @private
2852          */
2853         function braceSubstitution( $piece, $frame ) {
2854                 global $wgContLang, $wgNonincludableNamespaces;
2855                 wfProfileIn( __METHOD__ );
2856                 wfProfileIn( __METHOD__.'-setup' );
2857
2858                 # Flags
2859                 $found = false;             # $text has been filled
2860                 $nowiki = false;            # wiki markup in $text should be escaped
2861                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2862                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2863                 $isChildObj = false;        # $text is a DOM node needing expansion in a child frame
2864                 $isLocalObj = false;        # $text is a DOM node needing expansion in the current frame
2865
2866                 # Title object, where $text came from
2867                 $title = null;
2868
2869                 # $part1 is the bit before the first |, and must contain only title characters.
2870                 # Various prefixes will be stripped from it later.
2871                 $titleWithSpaces = $frame->expand( $piece['title'] );
2872                 $part1 = trim( $titleWithSpaces );
2873                 $titleText = false;
2874
2875                 # Original title text preserved for various purposes
2876                 $originalTitle = $part1;
2877
2878                 # $args is a list of argument nodes, starting from index 0, not including $part1
2879                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2880                 wfProfileOut( __METHOD__.'-setup' );
2881
2882                 # SUBST
2883                 wfProfileIn( __METHOD__.'-modifiers' );
2884                 if ( !$found ) {
2885
2886                         $substMatch = $this->mSubsts->matchStartAndRemove( $part1 );
2887
2888                         # Possibilities for substMatch: "subst", "safesubst" or FALSE
2889                         # Whether to include depends also on whether we are in the pre-save-transform
2890                         #
2891                         # safesubst || (subst && PST) || (false && !PST) => transclude (skip the if)
2892                         # (false && PST) || (subst && !PST)  => return input (handled by if)
2893                         if ( $substMatch != 'safesubst' && ($substMatch == 'subst' xor $this->ot['wiki']) ) {
2894                                 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2895                                 $isLocalObj = true;
2896                                 $found = true;
2897                         }
2898                 }
2899
2900                 # Variables
2901                 if ( !$found && $args->getLength() == 0 ) {
2902                         $id = $this->mVariables->matchStartToEnd( $part1 );
2903                         if ( $id !== false ) {
2904                                 $text = $this->getVariableValue( $id, $frame );
2905                                 if (MagicWord::getCacheTTL($id)>-1)
2906                                         $this->mOutput->mContainsOldMagic = true;
2907                                 $found = true;
2908                         }
2909                 }
2910
2911                 # MSG, MSGNW and RAW
2912                 if ( !$found ) {
2913                         # Check for MSGNW:
2914                         $mwMsgnw = MagicWord::get( 'msgnw' );
2915                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2916                                 $nowiki = true;
2917                         } else {
2918                                 # Remove obsolete MSG:
2919                                 $mwMsg = MagicWord::get( 'msg' );
2920                                 $mwMsg->matchStartAndRemove( $part1 );
2921                         }
2922
2923                         # Check for RAW:
2924                         $mwRaw = MagicWord::get( 'raw' );
2925                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2926                                 $forceRawInterwiki = true;
2927                         }
2928                 }
2929                 wfProfileOut( __METHOD__.'-modifiers' );
2930
2931                 # Parser functions
2932                 if ( !$found ) {
2933                         wfProfileIn( __METHOD__ . '-pfunc' );
2934
2935                         $colonPos = strpos( $part1, ':' );
2936                         if ( $colonPos !== false ) {
2937                                 # Case sensitive functions
2938                                 $function = substr( $part1, 0, $colonPos );
2939                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
2940                                         $function = $this->mFunctionSynonyms[1][$function];
2941                                 } else {
2942                                         # Case insensitive functions
2943                                         $function = $wgContLang->lc( $function );
2944                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
2945                                                 $function = $this->mFunctionSynonyms[0][$function];
2946                                         } else {
2947                                                 $function = false;
2948                                         }
2949                                 }
2950                                 if ( $function ) {
2951                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
2952                                         $initialArgs = array( &$this );
2953                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
2954                                         if ( $flags & SFH_OBJECT_ARGS ) {
2955                                                 # Add a frame parameter, and pass the arguments as an array
2956                                                 $allArgs = $initialArgs;
2957                                                 $allArgs[] = $frame;
2958                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2959                                                         $funcArgs[] = $args->item( $i );
2960                                                 }
2961                                                 $allArgs[] = $funcArgs;
2962                                         } else {
2963                                                 # Convert arguments to plain text
2964                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2965                                                         $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
2966                                                 }
2967                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
2968                                         }
2969
2970                                         # Workaround for PHP bug 35229 and similar
2971                                         if ( !is_callable( $callback ) ) {
2972                                                 wfProfileOut( __METHOD__ . '-pfunc' );
2973                                                 wfProfileOut( __METHOD__ );
2974                                                 throw new MWException( "Tag hook for $function is not callable\n" );
2975                                         }
2976                                         $result = call_user_func_array( $callback, $allArgs );
2977                                         $found = true;
2978                                         $noparse = true;
2979                                         $preprocessFlags = 0;
2980
2981                                         if ( is_array( $result ) ) {
2982                                                 if ( isset( $result[0] ) ) {
2983                                                         $text = $result[0];
2984                                                         unset( $result[0] );
2985                                                 }
2986
2987                                                 // Extract flags into the local scope
2988                                                 // This allows callers to set flags such as nowiki, found, etc.
2989                                                 extract( $result );
2990                                         } else {
2991                                                 $text = $result;
2992                                         }
2993                                         if ( !$noparse ) {
2994                                                 $text = $this->preprocessToDom( $text, $preprocessFlags );
2995                                                 $isChildObj = true;
2996                                         }
2997                                 }
2998                         }
2999                         wfProfileOut( __METHOD__ . '-pfunc' );
3000                 }
3001
3002                 # Finish mangling title and then check for loops.
3003                 # Set $title to a Title object and $titleText to the PDBK
3004                 if ( !$found ) {
3005                         $ns = NS_TEMPLATE;
3006                         # Split the title into page and subpage
3007                         $subpage = '';
3008                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3009                         if ($subpage !== '') {
3010                                 $ns = $this->mTitle->getNamespace();
3011                         }
3012                         $title = Title::newFromText( $part1, $ns );
3013                         if ( $title ) {
3014                                 $titleText = $title->getPrefixedText();
3015                                 # Check for language variants if the template is not found
3016                                 if($wgContLang->hasVariants() && $title->getArticleID() == 0){
3017                                         $wgContLang->findVariantLink( $part1, $title, true );
3018                                 }
3019                                 # Do recursion depth check
3020                                 $limit = $this->mOptions->getMaxTemplateDepth();
3021                                 if ( $frame->depth >= $limit ) {
3022                                         $found = true;
3023                                         $text = '<span class="error">' . wfMsgForContent( 'parser-template-recursion-depth-warning', $limit ) . '</span>';
3024                                 }
3025                         }
3026                 }
3027
3028                 # Load from database
3029                 if ( !$found && $title ) {
3030                         wfProfileIn( __METHOD__ . '-loadtpl' );
3031                         if ( !$title->isExternal() ) {
3032                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
3033                                         $text = SpecialPage::capturePath( $title );
3034                                         if ( is_string( $text ) ) {
3035                                                 $found = true;
3036                                                 $isHTML = true;
3037                                                 $this->disableCache();
3038                                         }
3039                                 } else if ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
3040                                         $found = false; //access denied
3041                                         wfDebug( __METHOD__.": template inclusion denied for " . $title->getPrefixedDBkey() );
3042                                 } else {
3043                                         list( $text, $title ) = $this->getTemplateDom( $title );
3044                                         if ( $text !== false ) {
3045                                                 $found = true;
3046                                                 $isChildObj = true;
3047                                         }
3048                                 }
3049
3050                                 # If the title is valid but undisplayable, make a link to it
3051                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3052                                         $text = "[[:$titleText]]";
3053                                         $found = true;
3054                                 }
3055                         } elseif ( $title->isTrans() ) {
3056                                 // Interwiki transclusion
3057                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3058                                         $text = $this->interwikiTransclude( $title, 'render' );
3059                                         $isHTML = true;
3060                                 } else {
3061                                         $text = $this->interwikiTransclude( $title, 'raw' );
3062                                         // Preprocess it like a template
3063                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3064                                         $isChildObj = true;
3065                                 }
3066                                 $found = true;
3067                         }
3068
3069                         # Do infinite loop check
3070                         # This has to be done after redirect resolution to avoid infinite loops via redirects
3071                         if ( !$frame->loopCheck( $title ) ) {
3072                                 $found = true;
3073                                 $text = '<span class="error">' . wfMsgForContent( 'parser-template-loop-warning', $titleText ) . '</span>';
3074                                 wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
3075                         }
3076                         wfProfileOut( __METHOD__ . '-loadtpl' );
3077                 }
3078
3079                 # If we haven't found text to substitute by now, we're done
3080                 # Recover the source wikitext and return it
3081                 if ( !$found ) {
3082                         $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3083                         wfProfileOut( __METHOD__ );
3084                         return array( 'object' => $text );
3085                 }
3086
3087                 # Expand DOM-style return values in a child frame
3088                 if ( $isChildObj ) {
3089                         # Clean up argument array
3090                         $newFrame = $frame->newChild( $args, $title );
3091
3092                         if ( $nowiki ) {
3093                                 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3094                         } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3095                                 # Expansion is eligible for the empty-frame cache
3096                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
3097                                         $text = $this->mTplExpandCache[$titleText];
3098                                 } else {
3099                                         $text = $newFrame->expand( $text );
3100                                         $this->mTplExpandCache[$titleText] = $text;
3101                                 }
3102                         } else {
3103                                 # Uncached expansion
3104                                 $text = $newFrame->expand( $text );
3105                         }
3106                 }
3107                 if ( $isLocalObj && $nowiki ) {
3108                         $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3109                         $isLocalObj = false;
3110                 }
3111
3112                 # Replace raw HTML by a placeholder
3113                 # Add a blank line preceding, to prevent it from mucking up
3114                 # immediately preceding headings
3115                 if ( $isHTML ) {
3116                         $text = "\n\n" . $this->insertStripItem( $text );
3117                 }
3118                 # Escape nowiki-style return values
3119                 elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3120                         $text = wfEscapeWikiText( $text );
3121                 }
3122                 # Bug 529: if the template begins with a table or block-level
3123                 # element, it should be treated as beginning a new line.
3124                 # This behaviour is somewhat controversial.
3125                 elseif ( is_string( $text ) && !$piece['lineStart'] && preg_match('/^(?:{\\||:|;|#|\*)/', $text)) /*}*/{
3126                         $text = "\n" . $text;
3127                 }
3128
3129                 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3130                         # Error, oversize inclusion
3131                         $text = "[[$originalTitle]]" .
3132                                 $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3133                         $this->limitationWarn( 'post-expand-template-inclusion' );
3134                 }
3135
3136                 if ( $isLocalObj ) {
3137                         $ret = array( 'object' => $text );
3138                 } else {
3139                         $ret = array( 'text' => $text );
3140                 }
3141
3142                 wfProfileOut( __METHOD__ );
3143                 return $ret;
3144         }
3145
3146         /**
3147          * Get the semi-parsed DOM representation of a template with a given title,
3148          * and its redirect destination title. Cached.
3149          */
3150         function getTemplateDom( $title ) {
3151                 $cacheTitle = $title;
3152                 $titleText = $title->getPrefixedDBkey();
3153
3154                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3155                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3156                         $title = Title::makeTitle( $ns, $dbk );
3157                         $titleText = $title->getPrefixedDBkey();
3158                 }
3159                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3160                         return array( $this->mTplDomCache[$titleText], $title );
3161                 }
3162
3163                 // Cache miss, go to the database
3164                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3165
3166                 if ( $text === false ) {
3167                         $this->mTplDomCache[$titleText] = false;
3168                         return array( false, $title );
3169                 }
3170
3171                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3172                 $this->mTplDomCache[ $titleText ] = $dom;
3173
3174                 if (! $title->equals($cacheTitle)) {
3175                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3176                                 array( $title->getNamespace(),$cdb = $title->getDBkey() );
3177                 }
3178
3179                 return array( $dom, $title );
3180         }
3181
3182         /**
3183          * Fetch the unparsed text of a template and register a reference to it.
3184          */
3185         function fetchTemplateAndTitle( $title ) {
3186                 $templateCb = $this->mOptions->getTemplateCallback();
3187                 $stuff = call_user_func( $templateCb, $title, $this );
3188                 $text = $stuff['text'];
3189                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3190                 if ( isset( $stuff['deps'] ) ) {
3191                         foreach ( $stuff['deps'] as $dep ) {
3192                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3193                         }
3194                 }
3195                 return array($text,$finalTitle);
3196         }
3197
3198         function fetchTemplate( $title ) {
3199                 $rv = $this->fetchTemplateAndTitle($title);
3200                 return $rv[0];
3201         }
3202
3203         /**
3204          * Static function to get a template
3205          * Can be overridden via ParserOptions::setTemplateCallback().
3206          */
3207         static function statelessFetchTemplate( $title, $parser=false ) {
3208                 $text = $skip = false;
3209                 $finalTitle = $title;
3210                 $deps = array();
3211
3212                 // Loop to fetch the article, with up to 1 redirect
3213                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3214                         # Give extensions a chance to select the revision instead
3215                         $id = false; // Assume current
3216                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( $parser, &$title, &$skip, &$id ) );
3217
3218                         if( $skip ) {
3219                                 $text = false;
3220                                 $deps[] = array(
3221                                         'title' => $title,
3222                                         'page_id' => $title->getArticleID(),
3223                                         'rev_id' => null );
3224                                 break;
3225                         }
3226                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3227                         $rev_id = $rev ? $rev->getId() : 0;
3228                         // If there is no current revision, there is no page
3229                         if( $id === false && !$rev ) {
3230                                 $linkCache = LinkCache::singleton();
3231                                 $linkCache->addBadLinkObj( $title );
3232                         }
3233
3234                         $deps[] = array(
3235                                 'title' => $title,
3236                                 'page_id' => $title->getArticleID(),
3237                                 'rev_id' => $rev_id );
3238
3239                         if( $rev ) {
3240                                 $text = $rev->getText();
3241                         } elseif( $title->getNamespace() == NS_MEDIAWIKI ) {
3242                                 global $wgContLang;
3243                                 $message = $wgContLang->lcfirst( $title->getText() );
3244                                 $text = wfMsgForContentNoTrans( $message );
3245                                 if( wfEmptyMsg( $message, $text ) ) {
3246                                         $text = false;
3247                                         break;
3248                                 }
3249                         } else {
3250                                 break;
3251                         }
3252                         if ( $text === false ) {
3253                                 break;
3254                         }
3255                         // Redirect?
3256                         $finalTitle = $title;
3257                         $title = Title::newFromRedirect( $text );
3258                 }
3259                 return array(
3260                         'text' => $text,
3261                         'finalTitle' => $finalTitle,
3262                         'deps' => $deps );
3263         }
3264
3265         /**
3266          * Transclude an interwiki link.
3267          */
3268         function interwikiTransclude( $title, $action ) {
3269                 global $wgEnableScaryTranscluding;
3270
3271                 if (!$wgEnableScaryTranscluding)
3272                         return wfMsg('scarytranscludedisabled');
3273
3274                 $url = $title->getFullUrl( "action=$action" );
3275
3276                 if (strlen($url) > 255)
3277                         return wfMsg('scarytranscludetoolong');
3278                 return $this->fetchScaryTemplateMaybeFromCache($url);
3279         }
3280
3281         function fetchScaryTemplateMaybeFromCache($url) {
3282                 global $wgTranscludeCacheExpiry;
3283                 $dbr = wfGetDB(DB_SLAVE);
3284                 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
3285                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3286                                 array('tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ) );
3287                 if ($obj) {
3288                         return $obj->tc_contents;
3289                 }
3290
3291                 $text = Http::get($url);
3292                 if (!$text)
3293                         return wfMsg('scarytranscludefailed', $url);
3294
3295                 $dbw = wfGetDB(DB_MASTER);
3296                 $dbw->replace('transcache', array('tc_url'), array(
3297                         'tc_url' => $url,
3298                         'tc_time' => $dbw->timestamp( time() ),
3299                         'tc_contents' => $text));
3300                 return $text;
3301         }
3302
3303
3304         /**
3305          * Triple brace replacement -- used for template arguments
3306          * @private
3307          */
3308         function argSubstitution( $piece, $frame ) {
3309                 wfProfileIn( __METHOD__ );
3310
3311                 $error = false;
3312                 $parts = $piece['parts'];
3313                 $nameWithSpaces = $frame->expand( $piece['title'] );
3314                 $argName = trim( $nameWithSpaces );
3315                 $object = false;
3316                 $text = $frame->getArgument( $argName );
3317                 if (  $text === false && $parts->getLength() > 0
3318                   && (
3319                     $this->ot['html']
3320                     || $this->ot['pre']
3321                     || ( $this->ot['wiki'] && $frame->isTemplate() )
3322                   )
3323                 ) {
3324                         # No match in frame, use the supplied default
3325                         $object = $parts->item( 0 )->getChildren();
3326                 }
3327                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3328                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3329                         $this->limitationWarn( 'post-expand-template-argument' );
3330                 }
3331
3332                 if ( $text === false && $object === false ) {
3333                         # No match anywhere
3334                         $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3335                 }
3336                 if ( $error !== false ) {
3337                         $text .= $error;
3338                 }
3339                 if ( $object !== false ) {
3340                         $ret = array( 'object' => $object );
3341                 } else {
3342                         $ret = array( 'text' => $text );
3343                 }
3344
3345                 wfProfileOut( __METHOD__ );
3346                 return $ret;
3347         }
3348
3349         /**
3350          * Return the text to be used for a given extension tag.
3351          * This is the ghost of strip().
3352          *
3353          * @param array $params Associative array of parameters:
3354          *     name       PPNode for the tag name
3355          *     attr       PPNode for unparsed text where tag attributes are thought to be
3356          *     attributes Optional associative array of parsed attributes
3357          *     inner      Contents of extension element
3358          *     noClose    Original text did not have a close tag
3359          * @param PPFrame $frame
3360          */
3361         function extensionSubstitution( $params, $frame ) {
3362                 global $wgRawHtml, $wgContLang;
3363
3364                 $name = $frame->expand( $params['name'] );
3365                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3366                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3367                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf('%08X', $this->mMarkerIndex++) . self::MARKER_SUFFIX;
3368
3369                 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower($name)] ) &&
3370                         ( $this->ot['html'] || $this->ot['pre'] );
3371                 if ( $isFunctionTag ) {
3372                         $markerType = 'none';
3373                 } else {
3374                         $markerType = 'general';
3375                 }
3376                 if ( $this->ot['html'] || $isFunctionTag ) {
3377                         $name = strtolower( $name );
3378                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3379                         if ( isset( $params['attributes'] ) ) {
3380                                 $attributes = $attributes + $params['attributes'];
3381                         }
3382
3383                         if( isset( $this->mTagHooks[$name] ) ) {
3384                                 # Workaround for PHP bug 35229 and similar
3385                                 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3386                                         throw new MWException( "Tag hook for $name is not callable\n" );
3387                                 }
3388                                 $output = call_user_func_array( $this->mTagHooks[$name],
3389                                         array( $content, $attributes, $this, $frame ) );
3390                         } elseif( isset( $this->mFunctionTagHooks[$name] ) ) {
3391                                 list( $callback, $flags ) = $this->mFunctionTagHooks[$name];
3392                                 if( !is_callable( $callback ) )
3393                                         throw new MWException( "Tag hook for $name is not callable\n" );
3394
3395                                 $output = call_user_func_array( $callback,
3396                                         array( &$this, $frame, $content, $attributes ) );
3397                         } else {
3398                                 $output = '<span class="error">Invalid tag extension name: ' .
3399                                         htmlspecialchars( $name ) . '</span>';
3400                         }
3401
3402                         if ( is_array( $output ) ) {
3403                                 // Extract flags to local scope (to override $markerType)
3404                                 $flags = $output;
3405                                 $output = $flags[0];
3406                                 unset( $flags[0] );
3407                                 extract( $flags );
3408                         }
3409                 } else {
3410                         if ( is_null( $attrText ) ) {
3411                                 $attrText = '';
3412                         }
3413                         if ( isset( $params['attributes'] ) ) {
3414                                 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3415                                         $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3416                                                 htmlspecialchars( $attrValue ) . '"';
3417                                 }
3418                         }
3419                         if ( $content === null ) {
3420                                 $output = "<$name$attrText/>";
3421                         } else {
3422                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3423                                 $output = "<$name$attrText>$content$close";
3424                         }
3425                 }
3426
3427                 if( $markerType === 'none' ) {
3428                         return $output;
3429                 } elseif ( $markerType === 'nowiki' ) {
3430                         $this->mStripState->nowiki->setPair( $marker, $output );
3431                 } elseif ( $markerType === 'general' ) {
3432                         $this->mStripState->general->setPair( $marker, $output );
3433                 } else {
3434                         throw new MWException( __METHOD__.': invalid marker type' );
3435                 }
3436                 return $marker;
3437         }
3438
3439         /**
3440          * Increment an include size counter
3441          *
3442          * @param string $type The type of expansion
3443          * @param integer $size The size of the text
3444          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3445          */
3446         function incrementIncludeSize( $type, $size ) {
3447                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize( $type ) ) {
3448                         return false;
3449                 } else {
3450                         $this->mIncludeSizes[$type] += $size;
3451                         return true;
3452                 }
3453         }
3454
3455         /**
3456          * Increment the expensive function count
3457          *
3458          * @return boolean False if the limit has been exceeded
3459          */
3460         function incrementExpensiveFunctionCount() {
3461                 global $wgExpensiveParserFunctionLimit;
3462                 $this->mExpensiveFunctionCount++;
3463                 if($this->mExpensiveFunctionCount <= $wgExpensiveParserFunctionLimit) {
3464                         return true;
3465                 }
3466                 return false;
3467         }
3468
3469         /**
3470          * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3471          * Fills $this->mDoubleUnderscores, returns the modified text
3472          */
3473         function doDoubleUnderscore( $text ) {
3474                 wfProfileIn( __METHOD__ );
3475
3476                 // The position of __TOC__ needs to be recorded
3477                 $mw = MagicWord::get( 'toc' );
3478                 if( $mw->match( $text ) ) {
3479                         $this->mShowToc = true;
3480                         $this->mForceTocPosition = true;
3481
3482                         // Set a placeholder. At the end we'll fill it in with the TOC.
3483                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3484
3485                         // Only keep the first one.
3486                         $text = $mw->replace( '', $text );
3487                 }
3488
3489                 // Now match and remove the rest of them
3490                 $mwa = MagicWord::getDoubleUnderscoreArray();
3491                 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3492
3493                 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3494                         $this->mOutput->mNoGallery = true;
3495                 }
3496                 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3497                         $this->mShowToc = false;
3498                 }
3499                 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
3500                         $this->mOutput->setProperty( 'hiddencat', 'y' );
3501                         $this->addTrackingCategory( 'hidden-category-category' );
3502                 }
3503                 # (bug 8068) Allow control over whether robots index a page.
3504                 #
3505                 # FIXME (bug 14899): __INDEX__ always overrides __NOINDEX__ here!  This
3506                 # is not desirable, the last one on the page should win.
3507                 if( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
3508                         $this->mOutput->setIndexPolicy( 'noindex' );
3509                         $this->addTrackingCategory( 'noindex-category' );
3510                 }
3511                 if( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ){
3512                         $this->mOutput->setIndexPolicy( 'index' );
3513                         $this->addTrackingCategory( 'index-category' );
3514                 }
3515
3516                 wfProfileOut( __METHOD__ );
3517                 return $text;
3518         }
3519
3520         /**
3521          * Add a tracking category, getting the title from a system message,
3522          * or print a debug message if the title is invalid.
3523          * @param $msg String message key
3524          * @return Bool whether the addition was successful
3525          */
3526         protected function addTrackingCategory( $msg ){
3527                 $cat = wfMsgForContent( $msg );
3528
3529                 # Allow tracking categories to be disabled by setting them to "-"
3530                 if( $cat === '-' ) return false;
3531
3532                 $containerCategory = Title::makeTitleSafe( NS_CATEGORY, $cat );
3533                 if ( $containerCategory ) {
3534                         $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3535                         return true;
3536                 } else {
3537                         wfDebug( __METHOD__.": [[MediaWiki:$msg]] is not a valid title!\n" );
3538                         return false;
3539                 }
3540         }
3541
3542         /**
3543          * This function accomplishes several tasks:
3544          * 1) Auto-number headings if that option is enabled
3545          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3546          * 3) Add a Table of contents on the top for users who have enabled the option
3547          * 4) Auto-anchor headings
3548          *
3549          * It loops through all headlines, collects the necessary data, then splits up the
3550          * string and re-inserts the newly formatted headlines.
3551          *
3552          * @param string $text
3553          * @param string $origText Original, untouched wikitext
3554          * @param boolean $isMain
3555          * @private
3556          */
3557         function formatHeadings( $text, $origText, $isMain=true ) {
3558                 global $wgMaxTocLevel, $wgContLang, $wgHtml5, $wgExperimentalHtmlIds;
3559
3560                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3561                 $showEditLink = $this->mOptions->getEditSection();
3562
3563                 // Do not call quickUserCan unless necessary
3564                 if( $showEditLink && !$this->mTitle->quickUserCan( 'edit' ) ) {
3565                         $showEditLink = 0;
3566                 }
3567
3568                 # Inhibit editsection links if requested in the page
3569                 if ( isset( $this->mDoubleUnderscores['noeditsection'] )  || $this->mOptions->getIsPrintable() ) {
3570                         $showEditLink = 0;
3571                 }
3572
3573                 # Get all headlines for numbering them and adding funky stuff like [edit]
3574                 # links - this is for later, but we need the number of headlines right now
3575                 $matches = array();
3576                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3577
3578                 # if there are fewer than 4 headlines in the article, do not show TOC
3579                 # unless it's been explicitly enabled.
3580                 $enoughToc = $this->mShowToc &&
3581                         (($numMatches >= 4) || $this->mForceTocPosition);
3582
3583                 # Allow user to stipulate that a page should have a "new section"
3584                 # link added via __NEWSECTIONLINK__
3585                 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
3586                         $this->mOutput->setNewSection( true );
3587                 }
3588
3589                 # Allow user to remove the "new section"
3590                 # link via __NONEWSECTIONLINK__
3591                 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
3592                         $this->mOutput->hideNewSection( true );
3593                 }
3594
3595                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3596                 # override above conditions and always show TOC above first header
3597                 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
3598                         $this->mShowToc = true;
3599                         $enoughToc = true;
3600                 }
3601
3602                 # We need this to perform operations on the HTML
3603                 $sk = $this->mOptions->getSkin();
3604
3605                 # headline counter
3606                 $headlineCount = 0;
3607                 $numVisible = 0;
3608
3609                 # Ugh .. the TOC should have neat indentation levels which can be
3610                 # passed to the skin functions. These are determined here
3611                 $toc = '';
3612                 $full = '';
3613                 $head = array();
3614                 $sublevelCount = array();
3615                 $levelCount = array();
3616                 $toclevel = 0;
3617                 $level = 0;
3618                 $prevlevel = 0;
3619                 $toclevel = 0;
3620                 $prevtoclevel = 0;
3621                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
3622                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3623                 $oldType = $this->mOutputType;
3624                 $this->setOutputType( self::OT_WIKI );
3625                 $frame = $this->getPreprocessor()->newFrame();
3626                 $root = $this->preprocessToDom( $origText );
3627                 $node = $root->getFirstChild();
3628                 $byteOffset = 0;
3629                 $tocraw = array();
3630
3631                 foreach( $matches[3] as $headline ) {
3632                         $isTemplate = false;
3633                         $titleText = false;
3634                         $sectionIndex = false;
3635                         $numbering = '';
3636                         $markerMatches = array();
3637                         if (preg_match("/^$markerRegex/", $headline, $markerMatches)) {
3638                                 $serial = $markerMatches[1];
3639                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3640                                 $isTemplate = ($titleText != $baseTitleText);
3641                                 $headline = preg_replace("/^$markerRegex/", "", $headline);
3642                         }
3643
3644                         if( $toclevel ) {
3645                                 $prevlevel = $level;
3646                                 $prevtoclevel = $toclevel;
3647                         }
3648                         $level = $matches[1][$headlineCount];
3649
3650                         if ( $level > $prevlevel ) {
3651                                 # Increase TOC level
3652                                 $toclevel++;
3653                                 $sublevelCount[$toclevel] = 0;
3654                                 if( $toclevel<$wgMaxTocLevel ) {
3655                                         $prevtoclevel = $toclevel;
3656                                         $toc .= $sk->tocIndent();
3657                                         $numVisible++;
3658                                 }
3659                         }
3660                         elseif ( $level < $prevlevel && $toclevel > 1 ) {
3661                                 # Decrease TOC level, find level to jump to
3662
3663                                 for ($i = $toclevel; $i > 0; $i--) {
3664                                         if ( $levelCount[$i] == $level ) {
3665                                                 # Found last matching level
3666                                                 $toclevel = $i;
3667                                                 break;
3668                                         }
3669                                         elseif ( $levelCount[$i] < $level ) {
3670                                                 # Found first matching level below current level
3671                                                 $toclevel = $i + 1;
3672                                                 break;
3673                                         }
3674                                 }
3675                                 if( $i == 0 ) $toclevel = 1;
3676                                 if( $toclevel<$wgMaxTocLevel ) {
3677                                         if($prevtoclevel < $wgMaxTocLevel) {
3678                                                 # Unindent only if the previous toc level was shown :p
3679                                                 $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3680                                                 $prevtoclevel = $toclevel;
3681                                         } else {
3682                                                 $toc .= $sk->tocLineEnd();
3683                                         }
3684                                 }
3685                         }
3686                         else {
3687                                 # No change in level, end TOC line
3688                                 if( $toclevel<$wgMaxTocLevel ) {
3689                                         $toc .= $sk->tocLineEnd();
3690                                 }
3691                         }
3692
3693                         $levelCount[$toclevel] = $level;
3694
3695                         # count number of headlines for each level
3696                         @$sublevelCount[$toclevel]++;
3697                         $dot = 0;
3698                         for( $i = 1; $i <= $toclevel; $i++ ) {
3699                                 if( !empty( $sublevelCount[$i] ) ) {
3700                                         if( $dot ) {
3701                                                 $numbering .= '.';
3702                                         }
3703                                         $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3704                                         $dot = 1;
3705                                 }
3706                         }
3707
3708                         # The safe header is a version of the header text safe to use for links
3709                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3710                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
3711
3712                         # Remove link placeholders by the link text.
3713                         #     <!--LINK number-->
3714                         # turns into
3715                         #     link text with suffix
3716                         $safeHeadline = $this->replaceLinkHoldersText( $safeHeadline );
3717
3718                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3719                         $tocline = preg_replace(
3720                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3721                                 array( '',                          '<$1>'),
3722                                 $safeHeadline
3723                         );
3724                         $tocline = trim( $tocline );
3725
3726                         # For the anchor, strip out HTML-y stuff period
3727                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
3728                         $safeHeadline = preg_replace( '/[ _]+/', ' ', $safeHeadline );
3729                         $safeHeadline = trim( $safeHeadline );
3730
3731                         # Save headline for section edit hint before it's escaped
3732                         $headlineHint = $safeHeadline;
3733
3734                         if ( $wgHtml5 && $wgExperimentalHtmlIds ) {
3735                                 # For reverse compatibility, provide an id that's
3736                                 # HTML4-compatible, like we used to.
3737                                 #
3738                                 # It may be worth noting, academically, that it's possible for
3739                                 # the legacy anchor to conflict with a non-legacy headline
3740                                 # anchor on the page.  In this case likely the "correct" thing
3741                                 # would be to either drop the legacy anchors or make sure
3742                                 # they're numbered first.  However, this would require people
3743                                 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
3744                                 # manually, so let's not bother worrying about it.
3745                                 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
3746                                         array( 'noninitial', 'legacy' ) );
3747                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline );
3748
3749                                 if ( $legacyHeadline == $safeHeadline ) {
3750                                         # No reason to have both (in fact, we can't)
3751                                         $legacyHeadline = false;
3752                                 }
3753                         } else {
3754                                 $legacyHeadline = false;
3755                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
3756                                         'noninitial' );
3757                         }
3758
3759                         # HTML names must be case-insensitively unique (bug 10721).  FIXME:
3760                         # Does this apply to Unicode characters?  Because we aren't
3761                         # handling those here.
3762                         $arrayKey = strtolower( $safeHeadline );
3763                         if ( $legacyHeadline === false ) {
3764                                 $legacyArrayKey = false;
3765                         } else {
3766                                 $legacyArrayKey = strtolower( $legacyHeadline );
3767                         }
3768
3769                         # count how many in assoc. array so we can track dupes in anchors
3770                         if ( isset( $refers[$arrayKey] ) ) {
3771                                 $refers[$arrayKey]++;
3772                         } else {
3773                                 $refers[$arrayKey] = 1;
3774                         }
3775                         if ( isset( $refers[$legacyArrayKey] ) ) {
3776                                 $refers[$legacyArrayKey]++;
3777                         } else {
3778                                 $refers[$legacyArrayKey] = 1;
3779                         }
3780
3781                         # Don't number the heading if it is the only one (looks silly)
3782                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3783                                 # the two are different if the line contains a link
3784                                 $headline = $numbering . ' ' . $headline;
3785                         }
3786
3787                         # Create the anchor for linking from the TOC to the section
3788                         $anchor = $safeHeadline;
3789                         $legacyAnchor = $legacyHeadline;
3790                         if ( $refers[$arrayKey] > 1 ) {
3791                                 $anchor .= '_' . $refers[$arrayKey];
3792                         }
3793                         if ( $legacyHeadline !== false && $refers[$legacyArrayKey] > 1 ) {
3794                                 $legacyAnchor .= '_' . $refers[$legacyArrayKey];
3795                         }
3796                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3797                                 $toc .= $sk->tocLine($anchor, $tocline,
3798                                         $numbering, $toclevel, ($isTemplate ? false : $sectionIndex));
3799                         }
3800
3801                         # Add the section to the section tree
3802                         # Find the DOM node for this header
3803                         while ( $node && !$isTemplate ) {
3804                                 if ( $node->getName() === 'h' ) {
3805                                         $bits = $node->splitHeading();
3806                                         if ( $bits['i'] == $sectionIndex )
3807                                                 break;
3808                                 }
3809                                 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
3810                                         $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
3811                                 $node = $node->getNextSibling();
3812                         }
3813                         $tocraw[] = array(
3814                                 'toclevel' => $toclevel,
3815                                 'level' => $level,
3816                                 'line' => $tocline,
3817                                 'number' => $numbering,
3818                                 'index' => ($isTemplate ? 'T-' : '' ) . $sectionIndex,
3819                                 'fromtitle' => $titleText,
3820                                 'byteoffset' => ( $isTemplate ? null : $byteOffset ),
3821                                 'anchor' => $anchor,
3822                         );
3823
3824                         # give headline the correct <h#> tag
3825                         if( $showEditLink && $sectionIndex !== false ) {
3826                                 if( $isTemplate ) {
3827                                         # Put a T flag in the section identifier, to indicate to extractSections()
3828                                         # that sections inside <includeonly> should be counted.
3829                                         $editlink = $sk->doEditSectionLink(Title::newFromText( $titleText ), "T-$sectionIndex");
3830                                 } else {
3831                                         $editlink = $sk->doEditSectionLink($this->mTitle, $sectionIndex, $headlineHint);
3832                                 }
3833                         } else {
3834                                 $editlink = '';
3835                         }
3836                         $head[$headlineCount] = $sk->makeHeadline( $level,
3837                                 $matches['attrib'][$headlineCount], $anchor, $headline,
3838                                 $editlink, $legacyAnchor );
3839
3840                         $headlineCount++;
3841                 }
3842
3843                 $this->setOutputType( $oldType );
3844
3845                 # Never ever show TOC if no headers
3846                 if( $numVisible < 1 ) {
3847                         $enoughToc = false;
3848                 }
3849
3850                 if( $enoughToc ) {
3851                         if( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
3852                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
3853                         }
3854                         $toc = $sk->tocList( $toc );
3855                         $this->mOutput->setTOCHTML( $toc );
3856                 }
3857
3858                 if ( $isMain ) {
3859                         $this->mOutput->setSections( $tocraw );
3860                 }
3861
3862                 # split up and insert constructed headlines
3863
3864                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3865                 $i = 0;
3866
3867                 foreach( $blocks as $block ) {
3868                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block !== "\n" ) {
3869                                 # This is the [edit] link that appears for the top block of text when
3870                                 # section editing is enabled
3871
3872                                 # Disabled because it broke block formatting
3873                                 # For example, a bullet point in the top line
3874                                 # $full .= $sk->editSectionLink(0);
3875                         }
3876                         $full .= $block;
3877                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3878                                 # Top anchor now in skin
3879                                 $full = $full.$toc;
3880                         }
3881
3882                         if( !empty( $head[$i] ) ) {
3883                                 $full .= $head[$i];
3884                         }
3885                         $i++;
3886                 }
3887                 if( $this->mForceTocPosition ) {
3888                         return str_replace( '<!--MWTOC-->', $toc, $full );
3889                 } else {
3890                         return $full;
3891                 }
3892         }
3893
3894         /**
3895          * Merge $tree2 into $tree1 by replacing the section with index
3896          * $section in $tree1 and its descendants with the sections in $tree2.
3897          * Note that in the returned section tree, only the 'index' and
3898          * 'byteoffset' fields are guaranteed to be correct.
3899          * @param $tree1 array Section tree from ParserOutput::getSectons()
3900         *  @param $tree2 array Section tree
3901          * @param $section int Section index
3902          * @param $title Title Title both section trees come from
3903          * @param $len2 int Length of the original wikitext for $tree2
3904          * @return array Merged section tree
3905          */
3906         public static function mergeSectionTrees( $tree1, $tree2, $section, $title, $len2 ) {
3907                 global $wgContLang;
3908                 $newTree = array();
3909                 $targetLevel = false;
3910                 $merged = false;
3911                 $lastLevel = 1;
3912                 $nextIndex = 1;
3913                 $numbering = array( 0 );
3914                 $titletext = $title->getPrefixedDBkey();
3915                 foreach ( $tree1 as $s ) {
3916                         if ( $targetLevel !== false ) {
3917                                 if ( $s['level'] <= $targetLevel )
3918                                         // We've skipped enough
3919                                         $targetLevel = false;
3920                                 else
3921                                         continue;
3922                         }
3923                         if ( $s['index'] != $section ||
3924                                         $s['fromtitle'] != $titletext ) {
3925                                 self::incrementNumbering( $numbering,
3926                                         $s['toclevel'], $lastLevel );
3927
3928                                 // Rewrite index, byteoffset and number
3929                                 if ( $s['fromtitle'] == $titletext ) {
3930                                         $s['index'] = $nextIndex++;
3931                                         if ( $merged )
3932                                                 $s['byteoffset'] += $len2;
3933                                 }
3934                                 $s['number']  = implode( '.', array_map(
3935                                         array( $wgContLang, 'formatnum' ),
3936                                         $numbering ) );
3937                                 $lastLevel = $s['toclevel'];
3938                                 $newTree[] = $s;
3939                         } else {
3940                                 // We're at $section
3941                                 // Insert sections from $tree2 here
3942                                 foreach ( $tree2 as $s2 ) {
3943                                         // Rewrite the fields in $s2
3944                                         // before inserting it
3945                                         $s2['toclevel'] += $s['toclevel'] - 1;
3946                                         $s2['level'] += $s['level'] - 1;
3947                                         $s2['index'] = $nextIndex++;
3948                                         $s2['byteoffset'] += $s['byteoffset'];
3949
3950                                         self::incrementNumbering( $numbering,
3951                                                 $s2['toclevel'], $lastLevel );
3952                                         $s2['number']  = implode( '.', array_map(
3953                                                 array( $wgContLang, 'formatnum' ),
3954                                                 $numbering ) );
3955                                         $lastLevel = $s2['toclevel'];
3956                                         $newTree[] = $s2;
3957                                 }
3958                                 // Skip all descendants of $section in $tree1
3959                                 $targetLevel = $s['level'];
3960                                 $merged = true;
3961                         }
3962                 }
3963                 return $newTree;
3964         }
3965
3966         /**
3967          * Increment a section number. Helper function for mergeSectionTrees()
3968          * @param $number array Array representing a section number
3969          * @param $level int Current TOC level (depth)
3970          * @param $lastLevel int Level of previous TOC entry
3971          */
3972         private static function incrementNumbering( &$number, $level, $lastLevel ) {
3973                 if ( $level > $lastLevel )
3974                         $number[$level - 1] = 1;
3975                 else if ( $level < $lastLevel ) {
3976                         foreach ( $number as $key => $unused )
3977                                 if ( $key >= $level )
3978                                         unset( $number[$key] );
3979                         $number[$level - 1]++;
3980                 } else
3981                         $number[$level - 1]++;
3982         }
3983
3984         /**
3985          * Transform wiki markup when saving a page by doing \r\n -> \n
3986          * conversion, substitting signatures, {{subst:}} templates, etc.
3987          *
3988          * @param string $text the text to transform
3989          * @param Title &$title the Title object for the current article
3990          * @param User $user the User object describing the current user
3991          * @param ParserOptions $options parsing options
3992          * @param bool $clearState whether to clear the parser state first
3993          * @return string the altered wiki markup
3994          * @public
3995          */
3996         function preSaveTransform( $text, Title $title, $user, $options, $clearState = true ) {
3997                 $this->mOptions = $options;
3998                 $this->setTitle( $title );
3999                 $this->setOutputType( self::OT_WIKI );
4000
4001                 if ( $clearState ) {
4002                         $this->clearState();
4003                 }
4004
4005                 $pairs = array(
4006                         "\r\n" => "\n",
4007                 );
4008                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
4009                 $text = $this->pstPass2( $text, $user );
4010                 $text = $this->mStripState->unstripBoth( $text );
4011                 return $text;
4012         }
4013
4014         /**
4015          * Pre-save transform helper function
4016          * @private
4017          */
4018         function pstPass2( $text, $user ) {
4019                 global $wgContLang, $wgLocaltimezone;
4020
4021                 /* Note: This is the timestamp saved as hardcoded wikitext to
4022                  * the database, we use $wgContLang here in order to give
4023                  * everyone the same signature and use the default one rather
4024                  * than the one selected in each user's preferences.
4025                  *
4026                  * (see also bug 12815)
4027                  */
4028                 $ts = $this->mOptions->getTimestamp();
4029                 if ( isset( $wgLocaltimezone ) ) {
4030                         $tz = $wgLocaltimezone;
4031                 } else {
4032                         $tz = date_default_timezone_get();
4033                 }
4034
4035                 $unixts = wfTimestamp( TS_UNIX, $ts );
4036                 $oldtz = date_default_timezone_get();
4037                 date_default_timezone_set( $tz );
4038                 $ts = date( 'YmdHis', $unixts );
4039                 $tzMsg = date( 'T', $unixts );  # might vary on DST changeover!
4040
4041                 /* Allow translation of timezones trough wiki. date() can return
4042                  * whatever crap the system uses, localised or not, so we cannot
4043                  * ship premade translations.
4044                  */
4045                 $key = 'timezone-' . strtolower( trim( $tzMsg ) );
4046                 $value = wfMsgForContent( $key );
4047                 if ( !wfEmptyMsg( $key, $value ) ) $tzMsg = $value;
4048
4049                 date_default_timezone_set( $oldtz );
4050
4051                 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4052
4053                 # Variable replacement
4054                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4055                 $text = $this->replaceVariables( $text );
4056
4057                 # Signatures
4058                 $sigText = $this->getUserSig( $user );
4059                 $text = strtr( $text, array(
4060                         '~~~~~' => $d,
4061                         '~~~~' => "$sigText $d",
4062                         '~~~' => $sigText
4063                 ) );
4064
4065                 # Links of the form [[|<blah>]] or [[<blah>|]] perform pipe tricks
4066                 # Note this only allows the # in the one position it works.
4067                 list( $tc, $nc ) = Parser::getPipeTrickCharacterClasses();
4068                 $pipeTrickRe = "/\[\[(?:(\\|)($tc+)|(#?$tc+)\\|)\]\]/";
4069                 $text = preg_replace_callback( $pipeTrickRe, array( $this, 'pstPipeTrickCallback' ), $text );
4070
4071                 # Trim trailing whitespace
4072                 $text = rtrim( $text );
4073
4074                 return $text;
4075         }
4076
4077         /**
4078          * Called from pstPass2 to perform the pipe trick on links.
4079          * Original was either [[|text]] or [[link|]]
4080          *
4081          * @param Array ("|" or "", text, link) $m
4082          */
4083         function pstPipeTrickCallback( $m )
4084         {
4085                 if( $m[1] ) { # [[|<blah>]]
4086                         $text = $m[2];
4087                         $link = $this->getPipeTrickLink( $text );
4088                 } else { # [[<blah>|]]
4089                         $link = $m[3];
4090                         $text = $this->getPipeTrickText( $link );
4091                 }
4092
4093                 return $link === $text ? "[[$link]]" : "[[$link|$text]]";
4094         }
4095
4096         /**
4097          * Fetch the user's signature text, if any, and normalize to
4098          * validated, ready-to-insert wikitext.
4099          * If you have pre-fetched the nickname or the fancySig option, you can
4100          * specify them here to save a database query.
4101          *
4102          * @param User $user
4103          * @return string
4104          */
4105         function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4106                 global $wgMaxSigChars;
4107
4108                 $username = $user->getName();
4109
4110                 // If not given, retrieve from the user object.
4111                 if ( $nickname === false )
4112                         $nickname = $user->getOption( 'nickname' );
4113
4114                 if ( is_null( $fancySig ) )
4115                         $fancySig = $user->getBoolOption( 'fancysig' );
4116
4117                 $nickname = $nickname == null ? $username : $nickname;
4118
4119                 if( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4120                         $nickname = $username;
4121                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4122                 } elseif( $fancySig !== false ) {
4123                         # Sig. might contain markup; validate this
4124                         if( $this->validateSig( $nickname ) !== false ) {
4125                                 # Validated; clean up (if needed) and return it
4126                                 return $this->cleanSig( $nickname, true );
4127                         } else {
4128                                 # Failed to validate; fall back to the default
4129                                 $nickname = $username;
4130                                 wfDebug( __METHOD__.": $username has bad XML tags in signature.\n" );
4131                         }
4132                 }
4133
4134                 // Make sure nickname doesnt get a sig in a sig
4135                 $nickname = $this->cleanSigInSig( $nickname );
4136
4137                 # If we're still here, make it a link to the user page
4138                 $userText = wfEscapeWikiText( $username );
4139                 $nickText = wfEscapeWikiText( $nickname );
4140                 if ( $user->isAnon() )  {
4141                         return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
4142                 } else {
4143                         return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
4144                 }
4145         }
4146
4147         /**
4148          * Check that the user's signature contains no bad XML
4149          *
4150          * @param string $text
4151          * @return mixed An expanded string, or false if invalid.
4152          */
4153         function validateSig( $text ) {
4154                 return( Xml::isWellFormedXmlFragment( $text ) ? $text : false );
4155         }
4156
4157         /**
4158          * Clean up signature text
4159          *
4160          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
4161          * 2) Substitute all transclusions
4162          *
4163          * @param string $text
4164          * @param $parsing Whether we're cleaning (preferences save) or parsing
4165          * @return string Signature text
4166          */
4167         function cleanSig( $text, $parsing = false ) {
4168                 if ( !$parsing ) {
4169                         global $wgTitle;
4170                         $this->clearState();
4171                         $this->setTitle( $wgTitle );
4172                         $this->mOptions = new ParserOptions;
4173                         $this->setOutputType = self::OT_PREPROCESS;
4174                 }
4175
4176                 # Option to disable this feature
4177                 if ( !$this->mOptions->getCleanSignatures() ) {
4178                         return $text;
4179                 }
4180
4181                 # FIXME: regex doesn't respect extension tags or nowiki
4182                 #  => Move this logic to braceSubstitution()
4183                 $substWord = MagicWord::get( 'subst' );
4184                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4185                 $substText = '{{' . $substWord->getSynonym( 0 );
4186
4187                 $text = preg_replace( $substRegex, $substText, $text );
4188                 $text = $this->cleanSigInSig( $text );
4189                 $dom = $this->preprocessToDom( $text );
4190                 $frame = $this->getPreprocessor()->newFrame();
4191                 $text = $frame->expand( $dom );
4192
4193                 if ( !$parsing ) {
4194                         $text = $this->mStripState->unstripBoth( $text );
4195                 }
4196
4197                 return $text;
4198         }
4199
4200         /**
4201          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4202          * @param string $text
4203          * @return string Signature text with /~{3,5}/ removed
4204          */
4205         function cleanSigInSig( $text ) {
4206                 $text = preg_replace( '/~{3,5}/', '', $text );
4207                 return $text;
4208         }
4209
4210         /**
4211          * Set up some variables which are usually set up in parse()
4212          * so that an external function can call some class members with confidence
4213          * @public
4214          */
4215         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
4216                 $this->setTitle( $title );
4217                 $this->mOptions = $options;
4218                 $this->setOutputType( $outputType );
4219                 if ( $clearState ) {
4220                         $this->clearState();
4221                 }
4222         }
4223
4224         /**
4225          * Wrapper for preprocess()
4226          *
4227          * @param string $text the text to preprocess
4228          * @param ParserOptions $options  options
4229          * @return string
4230          * @public
4231          */
4232         function transformMsg( $text, $options ) {
4233                 global $wgTitle;
4234                 static $executing = false;
4235
4236                 # Guard against infinite recursion
4237                 if ( $executing ) {
4238                         return $text;
4239                 }
4240                 $executing = true;
4241
4242                 wfProfileIn(__METHOD__);
4243                 $text = $this->preprocess( $text, $wgTitle, $options );
4244
4245                 $executing = false;
4246                 wfProfileOut(__METHOD__);
4247                 return $text;
4248         }
4249
4250         /**
4251          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
4252          * The callback should have the following form:
4253          *    function myParserHook( $text, $params, &$parser ) { ... }
4254          *
4255          * Transform and return $text. Use $parser for any required context, e.g. use
4256          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4257          *
4258          * @public
4259          *
4260          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
4261          * @param mixed $callback The callback function (and object) to use for the tag
4262          *
4263          * @return The old value of the mTagHooks array associated with the hook
4264          */
4265         function setHook( $tag, $callback ) {
4266                 $tag = strtolower( $tag );
4267                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4268                 $this->mTagHooks[$tag] = $callback;
4269                 if( !in_array( $tag, $this->mStripList ) ) {
4270                         $this->mStripList[] = $tag;
4271                 }
4272
4273                 return $oldVal;
4274         }
4275
4276         function setTransparentTagHook( $tag, $callback ) {
4277                 $tag = strtolower( $tag );
4278                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4279                 $this->mTransparentTagHooks[$tag] = $callback;
4280
4281                 return $oldVal;
4282         }
4283
4284         /**
4285          * Remove all tag hooks
4286          */
4287         function clearTagHooks() {
4288                 $this->mTagHooks = array();
4289                 $this->mStripList = $this->mDefaultStripList;
4290         }
4291
4292         /**
4293          * Create a function, e.g. {{sum:1|2|3}}
4294          * The callback function should have the form:
4295          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4296          *
4297          * Or with SFH_OBJECT_ARGS:
4298          *    function myParserFunction( $parser, $frame, $args ) { ... }
4299          *
4300          * The callback may either return the text result of the function, or an array with the text
4301          * in element 0, and a number of flags in the other elements. The names of the flags are
4302          * specified in the keys. Valid flags are:
4303          *   found                     The text returned is valid, stop processing the template. This
4304          *                             is on by default.
4305          *   nowiki                    Wiki markup in the return value should be escaped
4306          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4307          *
4308          * @public
4309          *
4310          * @param string $id The magic word ID
4311          * @param mixed $callback The callback function (and object) to use
4312          * @param integer $flags a combination of the following flags:
4313          *     SFH_NO_HASH   No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4314          *
4315          *     SFH_OBJECT_ARGS   Pass the template arguments as PPNode objects instead of text. This
4316          *     allows for conditional expansion of the parse tree, allowing you to eliminate dead
4317          *     branches and thus speed up parsing. It is also possible to analyse the parse tree of
4318          *     the arguments, and to control the way they are expanded.
4319          *
4320          *     The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4321          *     arguments, for instance:
4322          *         $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4323          *
4324          *     For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4325          *     future versions. Please call $frame->expand() on it anyway so that your code keeps
4326          *     working if/when this is changed.
4327          *
4328          *     If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4329          *     expansion.
4330          *
4331          *     Please read the documentation in includes/parser/Preprocessor.php for more information
4332          *     about the methods available in PPFrame and PPNode.
4333          *
4334          * @return The old callback function for this name, if any
4335          */
4336         function setFunctionHook( $id, $callback, $flags = 0 ) {
4337                 global $wgContLang;
4338
4339                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4340                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4341
4342                 # Add to function cache
4343                 $mw = MagicWord::get( $id );
4344                 if( !$mw )
4345                         throw new MWException( __METHOD__.'() expecting a magic word identifier.' );
4346
4347                 $synonyms = $mw->getSynonyms();
4348                 $sensitive = intval( $mw->isCaseSensitive() );
4349
4350                 foreach ( $synonyms as $syn ) {
4351                         # Case
4352                         if ( !$sensitive ) {
4353                                 $syn = $wgContLang->lc( $syn );
4354                         }
4355                         # Add leading hash
4356                         if ( !( $flags & SFH_NO_HASH ) ) {
4357                                 $syn = '#' . $syn;
4358                         }
4359                         # Remove trailing colon
4360                         if ( substr( $syn, -1, 1 ) === ':' ) {
4361                                 $syn = substr( $syn, 0, -1 );
4362                         }
4363                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4364                 }
4365                 return $oldVal;
4366         }
4367
4368         /**
4369          * Get all registered function hook identifiers
4370          *
4371          * @return array
4372          */
4373         function getFunctionHooks() {
4374                 return array_keys( $this->mFunctionHooks );
4375         }
4376
4377         /**
4378          * Create a tag function, e.g. <test>some stuff</test>.
4379          * Unlike tag hooks, tag functions are parsed at preprocessor level.
4380          * Unlike parser functions, their content is not preprocessed.
4381          */
4382         function setFunctionTagHook( $tag, $callback, $flags ) {
4383                 $tag = strtolower( $tag );
4384                 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
4385                         $this->mFunctionTagHooks[$tag] : null;
4386                 $this->mFunctionTagHooks[$tag] = array( $callback, $flags );
4387
4388                 if( !in_array( $tag, $this->mStripList ) ) {
4389                         $this->mStripList[] = $tag;
4390                 }
4391
4392                 return $old;
4393         }
4394
4395         /**
4396          * FIXME: update documentation. makeLinkObj() is deprecated.
4397          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4398          * Placeholders created in Skin::makeLinkObj()
4399          * Returns an array of link CSS classes, indexed by PDBK.
4400          */
4401         function replaceLinkHolders( &$text, $options = 0 ) {
4402                 return $this->mLinkHolders->replace( $text );
4403         }
4404
4405         /**
4406          * Replace <!--LINK--> link placeholders with plain text of links
4407          * (not HTML-formatted).
4408          * @param string $text
4409          * @return string
4410          */
4411         function replaceLinkHoldersText( $text ) {
4412                 return $this->mLinkHolders->replaceText( $text );
4413         }
4414
4415         /**
4416          * Renders an image gallery from a text with one line per image.
4417          * text labels may be given by using |-style alternative text. E.g.
4418          *   Image:one.jpg|The number "1"
4419          *   Image:tree.jpg|A tree
4420          * given as text will return the HTML of a gallery with two images,
4421          * labeled 'The number "1"' and
4422          * 'A tree'.
4423          */
4424         function renderImageGallery( $text, $params ) {
4425                 $ig = new ImageGallery();
4426                 $ig->setContextTitle( $this->mTitle );
4427                 $ig->setShowBytes( false );
4428                 $ig->setShowFilename( false );
4429                 $ig->setParser( $this );
4430                 $ig->setHideBadImages();
4431                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4432                 $ig->useSkin( $this->mOptions->getSkin() );
4433                 $ig->mRevisionId = $this->mRevisionId;
4434
4435                 if( isset( $params['caption'] ) ) {
4436                         $caption = $params['caption'];
4437                         $caption = htmlspecialchars( $caption );
4438                         $caption = $this->replaceInternalLinks( $caption );
4439                         $ig->setCaptionHtml( $caption );
4440                 }
4441                 if( isset( $params['perrow'] ) ) {
4442                         $ig->setPerRow( $params['perrow'] );
4443                 }
4444                 if( isset( $params['widths'] ) ) {
4445                         $ig->setWidths( $params['widths'] );
4446                 }
4447                 if( isset( $params['heights'] ) ) {
4448                         $ig->setHeights( $params['heights'] );
4449                 }
4450
4451                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4452
4453                 $lines = StringUtils::explode( "\n", $text );
4454                 foreach ( $lines as $line ) {
4455                         # match lines like these:
4456                         # Image:someimage.jpg|This is some image
4457                         $matches = array();
4458                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4459                         # Skip empty lines
4460                         if ( count( $matches ) == 0 ) {
4461                                 continue;
4462                         }
4463
4464                         if ( strpos( $matches[0], '%' ) !== false )
4465                                 $matches[1] = urldecode( $matches[1] );
4466                         $tp = Title::newFromText( $matches[1]/*, NS_FILE*/ );
4467                         $nt =& $tp;
4468                         if( is_null( $nt ) ) {
4469                                 # Bogus title. Ignore these so we don't bomb out later.
4470                                 continue;
4471                         }
4472                         if ( isset( $matches[3] ) ) {
4473                                 $label = $matches[3];
4474                         } else {
4475                                 $label = '';
4476                         }
4477
4478                         $html = $this->recursiveTagParse( trim( $label ) );
4479
4480                         $ig->add( $nt, $html );
4481
4482                         # Only add real images (bug #5586)
4483                         if ( $nt->getNamespace() == NS_FILE ) {
4484                                 $this->mOutput->addImage( $nt->getDBkey() );
4485                         }
4486                 }
4487                 return $ig->toHTML();
4488         }
4489
4490         function getImageParams( $handler ) {
4491                 if ( $handler ) {
4492                         $handlerClass = get_class( $handler );
4493                 } else {
4494                         $handlerClass = '';
4495                 }
4496                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4497                         // Initialise static lists
4498                         static $internalParamNames = array(
4499                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4500                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4501                                         'bottom', 'text-bottom' ),
4502                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4503                                         'upright', 'border', 'link', 'alt' ),
4504                         );
4505                         static $internalParamMap;
4506                         if ( !$internalParamMap ) {
4507                                 $internalParamMap = array();
4508                                 foreach ( $internalParamNames as $type => $names ) {
4509                                         foreach ( $names as $name ) {
4510                                                 $magicName = str_replace( '-', '_', "img_$name" );
4511                                                 $internalParamMap[$magicName] = array( $type, $name );
4512                                         }
4513                                 }
4514                         }
4515
4516                         // Add handler params
4517                         $paramMap = $internalParamMap;
4518                         if ( $handler ) {
4519                                 $handlerParamMap = $handler->getParamMap();
4520                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4521                                         $paramMap[$magic] = array( 'handler', $paramName );
4522                                 }
4523                         }
4524                         $this->mImageParams[$handlerClass] = $paramMap;
4525                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4526                 }
4527                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4528         }
4529
4530         /**
4531          * Parse image options text and use it to make an image
4532          * @param Title $title
4533          * @param string $options
4534          * @param LinkHolderArray $holders
4535          */
4536         function makeImage( $title, $options, $holders = false ) {
4537                 # Check if the options text is of the form "options|alt text"
4538                 # Options are:
4539                 #  * thumbnail  make a thumbnail with enlarge-icon and caption, alignment depends on lang
4540                 #  * left       no resizing, just left align. label is used for alt= only
4541                 #  * right      same, but right aligned
4542                 #  * none       same, but not aligned
4543                 #  * ___px      scale to ___ pixels width, no aligning. e.g. use in taxobox
4544                 #  * center     center the image
4545                 #  * frame      Keep original image size, no magnify-button.
4546                 #  * framed     Same as "frame"
4547                 #  * frameless  like 'thumb' but without a frame. Keeps user preferences for width
4548                 #  * upright    reduce width for upright images, rounded to full __0 px
4549                 #  * border     draw a 1px border around the image
4550                 #  * alt        Text for HTML alt attribute (defaults to empty)
4551                 #  * link       Set the target of the image link. Can be external, interwiki, or local
4552                 # vertical-align values (no % or length right now):
4553                 #  * baseline
4554                 #  * sub
4555                 #  * super
4556                 #  * top
4557                 #  * text-top
4558                 #  * middle
4559                 #  * bottom
4560                 #  * text-bottom
4561
4562                 $parts = StringUtils::explode( "|", $options );
4563                 $sk = $this->mOptions->getSkin();
4564
4565                 # Give extensions a chance to select the file revision for us
4566                 $skip = $time = $descQuery = false;
4567                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time, &$descQuery ) );
4568
4569                 if ( $skip ) {
4570                         return $sk->link( $title );
4571                 }
4572
4573                 # Get the file
4574                 $imagename = $title->getDBkey();
4575                 $file = wfFindFile( $title, array( 'time' => $time ) );
4576                 # Get parameter map
4577                 $handler = $file ? $file->getHandler() : false;
4578
4579                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4580
4581                 # Process the input parameters
4582                 $caption = '';
4583                 $params = array( 'frame' => array(), 'handler' => array(),
4584                         'horizAlign' => array(), 'vertAlign' => array() );
4585                 foreach( $parts as $part ) {
4586                         $part = trim( $part );
4587                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4588                         $validated = false;
4589                         if( isset( $paramMap[$magicName] ) ) {
4590                                 list( $type, $paramName ) = $paramMap[$magicName];
4591
4592                                 // Special case; width and height come in one variable together
4593                                 if( $type === 'handler' && $paramName === 'width' ) {
4594                                         $m = array();
4595                                         # (bug 13500) In both cases (width/height and width only),
4596                                         # permit trailing "px" for backward compatibility.
4597                                         if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
4598                                                 $width = intval( $m[1] );
4599                                                 $height = intval( $m[2] );
4600                                                 if ( $handler->validateParam( 'width', $width ) ) {
4601                                                         $params[$type]['width'] = $width;
4602                                                         $validated = true;
4603                                                 }
4604                                                 if ( $handler->validateParam( 'height', $height ) ) {
4605                                                         $params[$type]['height'] = $height;
4606                                                         $validated = true;
4607                                                 }
4608                                         } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
4609                                                 $width = intval( $value );
4610                                                 if ( $handler->validateParam( 'width', $width ) ) {
4611                                                         $params[$type]['width'] = $width;
4612                                                         $validated = true;
4613                                                 }
4614                                         } // else no validation -- bug 13436
4615                                 } else {
4616                                         if ( $type === 'handler' ) {
4617                                                 # Validate handler parameter
4618                                                 $validated = $handler->validateParam( $paramName, $value );
4619                                         } else {
4620                                                 # Validate internal parameters
4621                                                 switch( $paramName ) {
4622                                                 case 'manualthumb':
4623                                                 case 'alt':
4624                                                         // @todo Fixme: possibly check validity here for
4625                                                         // manualthumb? downstream behavior seems odd with
4626                                                         // missing manual thumbs.
4627                                                         $validated = true;
4628                                                         $value = $this->stripAltText( $value, $holders );
4629                                                         break;
4630                                                 case 'link':
4631                                                         $chars = self::EXT_LINK_URL_CLASS;
4632                                                         $prots = $this->mUrlProtocols;
4633                                                         if ( $value === '' ) {
4634                                                                 $paramName = 'no-link';
4635                                                                 $value = true;
4636                                                                 $validated = true;
4637                                                         } elseif ( preg_match( "/^$prots/", $value ) ) {
4638                                                                 if ( preg_match( "/^($prots)$chars+$/", $value, $m ) ) {
4639                                                                         $paramName = 'link-url';
4640                                                                         $this->mOutput->addExternalLink( $value );
4641                                                                         $validated = true;
4642                                                                 }
4643                                                         } else {
4644                                                                 $linkTitle = Title::newFromText( $value );
4645                                                                 if ( $linkTitle ) {
4646                                                                         $paramName = 'link-title';
4647                                                                         $value = $linkTitle;
4648                                                                         $this->mOutput->addLink( $linkTitle );
4649                                                                         $validated = true;
4650                                                                 }
4651                                                         }
4652                                                         break;
4653                                                 default:
4654                                                         // Most other things appear to be empty or numeric...
4655                                                         $validated = ( $value === false || is_numeric( trim( $value ) ) );
4656                                                 }
4657                                         }
4658
4659                                         if ( $validated ) {
4660                                                 $params[$type][$paramName] = $value;
4661                                         }
4662                                 }
4663                         }
4664                         if ( !$validated ) {
4665                                 $caption = $part;
4666                         }
4667                 }
4668
4669                 # Process alignment parameters
4670                 if ( $params['horizAlign'] ) {
4671                         $params['frame']['align'] = key( $params['horizAlign'] );
4672                 }
4673                 if ( $params['vertAlign'] ) {
4674                         $params['frame']['valign'] = key( $params['vertAlign'] );
4675                 }
4676
4677                 $params['frame']['caption'] = $caption;
4678
4679                 # Will the image be presented in a frame, with the caption below?
4680                 $imageIsFramed = isset( $params['frame']['frame'] ) ||
4681                                  isset( $params['frame']['framed'] ) ||
4682                                  isset( $params['frame']['thumbnail'] ) ||
4683                                  isset( $params['frame']['manualthumb'] );
4684
4685                 # In the old days, [[Image:Foo|text...]] would set alt text.  Later it
4686                 # came to also set the caption, ordinary text after the image -- which
4687                 # makes no sense, because that just repeats the text multiple times in
4688                 # screen readers.  It *also* came to set the title attribute.
4689                 #
4690                 # Now that we have an alt attribute, we should not set the alt text to
4691                 # equal the caption: that's worse than useless, it just repeats the
4692                 # text.  This is the framed/thumbnail case.  If there's no caption, we
4693                 # use the unnamed parameter for alt text as well, just for the time be-
4694                 # ing, if the unnamed param is set and the alt param is not.
4695                 #
4696                 # For the future, we need to figure out if we want to tweak this more,
4697                 # e.g., introducing a title= parameter for the title; ignoring the un-
4698                 # named parameter entirely for images without a caption; adding an ex-
4699                 # plicit caption= parameter and preserving the old magic unnamed para-
4700                 # meter for BC; ...
4701                 if ( $imageIsFramed ) { # Framed image
4702                         if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
4703                                 # No caption or alt text, add the filename as the alt text so
4704                                 # that screen readers at least get some description of the image
4705                                 $params['frame']['alt'] = $title->getText();
4706                         }
4707                         # Do not set $params['frame']['title'] because tooltips don't make sense
4708                         # for framed images
4709                 } else { # Inline image
4710                         if ( !isset( $params['frame']['alt'] ) ) {
4711                                 # No alt text, use the "caption" for the alt text
4712                                 if ( $caption !== '') {
4713                                         $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
4714                                 } else {
4715                                         # No caption, fall back to using the filename for the
4716                                         # alt text
4717                                         $params['frame']['alt'] = $title->getText();
4718                                 }
4719                         }
4720                         # Use the "caption" for the tooltip text
4721                         $params['frame']['title'] = $this->stripAltText( $caption, $holders );
4722                 }
4723
4724                 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params ) );
4725
4726                 # Linker does the rest
4727                 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'], $time, $descQuery );
4728
4729                 # Give the handler a chance to modify the parser object
4730                 if ( $handler ) {
4731                         $handler->parserTransformHook( $this, $file );
4732                 }
4733
4734                 return $ret;
4735         }
4736
4737         protected function stripAltText( $caption, $holders ) {
4738                 # Strip bad stuff out of the title (tooltip).  We can't just use
4739                 # replaceLinkHoldersText() here, because if this function is called
4740                 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
4741                 if ( $holders ) {
4742                         $tooltip = $holders->replaceText( $caption );
4743                 } else {
4744                         $tooltip = $this->replaceLinkHoldersText( $caption );
4745                 }
4746
4747                 # make sure there are no placeholders in thumbnail attributes
4748                 # that are later expanded to html- so expand them now and
4749                 # remove the tags
4750                 $tooltip = $this->mStripState->unstripBoth( $tooltip );
4751                 $tooltip = Sanitizer::stripAllTags( $tooltip );
4752
4753                 return $tooltip;
4754         }
4755
4756         /**
4757          * Set a flag in the output object indicating that the content is dynamic and
4758          * shouldn't be cached.
4759          */
4760         function disableCache() {
4761                 wfDebug( "Parser output marked as uncacheable.\n" );
4762                 $this->mOutput->mCacheTime = -1;
4763         }
4764
4765         /**#@+
4766          * Callback from the Sanitizer for expanding items found in HTML attribute
4767          * values, so they can be safely tested and escaped.
4768          * @param string $text
4769          * @param PPFrame $frame
4770          * @return string
4771          * @private
4772          */
4773         function attributeStripCallback( &$text, $frame = false ) {
4774                 $text = $this->replaceVariables( $text, $frame );
4775                 $text = $this->mStripState->unstripBoth( $text );
4776                 return $text;
4777         }
4778
4779         /**#@-*/
4780
4781         /**#@+
4782          * Accessor/mutator
4783          */
4784         function Title( $x = null ) { return wfSetVar( $this->mTitle, $x ); }
4785         function Options( $x = null ) { return wfSetVar( $this->mOptions, $x ); }
4786         function OutputType( $x = null ) { return wfSetVar( $this->mOutputType, $x ); }
4787         /**#@-*/
4788
4789         /**#@+
4790          * Accessor
4791          */
4792         function getTags() { return array_merge( array_keys($this->mTransparentTagHooks), array_keys( $this->mTagHooks ) ); }
4793         /**#@-*/
4794
4795
4796         /**
4797          * Break wikitext input into sections, and either pull or replace
4798          * some particular section's text.
4799          *
4800          * External callers should use the getSection and replaceSection methods.
4801          *
4802          * @param string $text Page wikitext
4803          * @param string $section A section identifier string of the form:
4804          *   <flag1> - <flag2> - ... - <section number>
4805          *
4806          * Currently the only recognised flag is "T", which means the target section number
4807          * was derived during a template inclusion parse, in other words this is a template
4808          * section edit link. If no flags are given, it was an ordinary section edit link.
4809          * This flag is required to avoid a section numbering mismatch when a section is
4810          * enclosed by <includeonly> (bug 6563).
4811          *
4812          * The section number 0 pulls the text before the first heading; other numbers will
4813          * pull the given section along with its lower-level subsections. If the section is
4814          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
4815          *
4816          * @param string $mode One of "get" or "replace"
4817          * @param string $newText Replacement text for section data.
4818          * @return string for "get", the extracted section text.
4819          *                for "replace", the whole page with the section replaced.
4820          */
4821         private function extractSections( $text, $section, $mode, $newText='' ) {
4822                 global $wgTitle;
4823                 $this->clearState();
4824                 $this->setTitle( $wgTitle ); // not generally used but removes an ugly failure mode
4825                 $this->mOptions = new ParserOptions;
4826                 $this->setOutputType( self::OT_WIKI );
4827                 $outText = '';
4828                 $frame = $this->getPreprocessor()->newFrame();
4829
4830                 // Process section extraction flags
4831                 $flags = 0;
4832                 $sectionParts = explode( '-', $section );
4833                 $sectionIndex = array_pop( $sectionParts );
4834                 foreach ( $sectionParts as $part ) {
4835                         if ( $part === 'T' ) {
4836                                 $flags |= self::PTD_FOR_INCLUSION;
4837                         }
4838                 }
4839                 // Preprocess the text
4840                 $root = $this->preprocessToDom( $text, $flags );
4841
4842                 // <h> nodes indicate section breaks
4843                 // They can only occur at the top level, so we can find them by iterating the root's children
4844                 $node = $root->getFirstChild();
4845
4846                 // Find the target section
4847                 if ( $sectionIndex == 0 ) {
4848                         // Section zero doesn't nest, level=big
4849                         $targetLevel = 1000;
4850                 } else {
4851             while ( $node ) {
4852                 if ( $node->getName() === 'h' ) {
4853                     $bits = $node->splitHeading();
4854                                         if ( $bits['i'] == $sectionIndex ) {
4855                                         $targetLevel = $bits['level'];
4856                                                 break;
4857                                         }
4858                                 }
4859                                 if ( $mode === 'replace' ) {
4860                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4861                                 }
4862                                 $node = $node->getNextSibling();
4863                         }
4864                 }
4865
4866                 if ( !$node ) {
4867                         // Not found
4868                         if ( $mode === 'get' ) {
4869                                 return $newText;
4870                         } else {
4871                                 return $text;
4872                         }
4873                 }
4874
4875                 // Find the end of the section, including nested sections
4876                 do {
4877                         if ( $node->getName() === 'h' ) {
4878                                 $bits = $node->splitHeading();
4879                                 $curLevel = $bits['level'];
4880                                 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
4881                                         break;
4882                                 }
4883                         }
4884                         if ( $mode === 'get' ) {
4885                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4886                         }
4887                         $node = $node->getNextSibling();
4888                 } while ( $node );
4889
4890                 // Write out the remainder (in replace mode only)
4891                 if ( $mode === 'replace' ) {
4892                         // Output the replacement text
4893                         // Add two newlines on -- trailing whitespace in $newText is conventionally
4894                         // stripped by the editor, so we need both newlines to restore the paragraph gap
4895                         // Only add trailing whitespace if there is newText
4896                         if($newText != "") {
4897                                 $outText .= $newText . "\n\n";
4898                         }
4899
4900                         while ( $node ) {
4901                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4902                                 $node = $node->getNextSibling();
4903                         }
4904                 }
4905
4906                 if ( is_string( $outText ) ) {
4907                         // Re-insert stripped tags
4908                         $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
4909                 }
4910
4911                 return $outText;
4912         }
4913
4914         /**
4915          * This function returns the text of a section, specified by a number ($section).
4916          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4917          * the first section before any such heading (section 0).
4918          *
4919          * If a section contains subsections, these are also returned.
4920          *
4921          * @param string $text text to look in
4922          * @param string $section section identifier
4923          * @param string $deftext default to return if section is not found
4924          * @return string text of the requested section
4925          */
4926         public function getSection( $text, $section, $deftext='' ) {
4927                 return $this->extractSections( $text, $section, "get", $deftext );
4928         }
4929
4930         public function replaceSection( $oldtext, $section, $text ) {
4931                 return $this->extractSections( $oldtext, $section, "replace", $text );
4932         }
4933
4934         /**
4935          * Get the timestamp associated with the current revision, adjusted for
4936          * the default server-local timestamp
4937          */
4938         function getRevisionTimestamp() {
4939                 if ( is_null( $this->mRevisionTimestamp ) ) {
4940                         wfProfileIn( __METHOD__ );
4941                         global $wgContLang;
4942                         $dbr = wfGetDB( DB_SLAVE );
4943                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
4944                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
4945
4946                         // Normalize timestamp to internal MW format for timezone processing.
4947                         // This has the added side-effect of replacing a null value with
4948                         // the current time, which gives us more sensible behavior for
4949                         // previews.
4950                         $timestamp = wfTimestamp( TS_MW, $timestamp );
4951
4952                         // The cryptic '' timezone parameter tells to use the site-default
4953                         // timezone offset instead of the user settings.
4954                         //
4955                         // Since this value will be saved into the parser cache, served
4956                         // to other users, and potentially even used inside links and such,
4957                         // it needs to be consistent for all visitors.
4958                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
4959
4960                         wfProfileOut( __METHOD__ );
4961                 }
4962                 return $this->mRevisionTimestamp;
4963         }
4964
4965         /**
4966          * Get the name of the user that edited the last revision
4967          */
4968         function getRevisionUser() {
4969                 // if this template is subst: the revision id will be blank,
4970                 // so just use the current user's name
4971                 if( $this->mRevisionId ) {
4972                         $revision = Revision::newFromId( $this->mRevisionId );
4973                         $revuser = $revision->getUserText();
4974                 } else {
4975                         global $wgUser;
4976                         $revuser = $wgUser->getName();
4977                 }
4978                 return $revuser;
4979         }
4980
4981         /**
4982          * Mutator for $mDefaultSort
4983          *
4984          * @param $sort New value
4985          */
4986         public function setDefaultSort( $sort ) {
4987                 $this->mDefaultSort = $sort;
4988         }
4989
4990         /**
4991          * Accessor for $mDefaultSort
4992          * Will use the title/prefixed title if none is set
4993          *
4994          * @return string
4995          */
4996         public function getDefaultSort() {
4997                 global $wgCategoryPrefixedDefaultSortkey;
4998                 if( $this->mDefaultSort !== false ) {
4999                         return $this->mDefaultSort;
5000                 } elseif ($this->mTitle->getNamespace() == NS_CATEGORY ||
5001                         !$wgCategoryPrefixedDefaultSortkey) {
5002                         return $this->mTitle->getText();
5003                 } else {
5004                         return $this->mTitle->getPrefixedText();
5005                 }
5006         }
5007
5008         /**
5009          * Accessor for $mDefaultSort
5010          * Unlike getDefaultSort(), will return false if none is set
5011          *
5012          * @return string or false
5013          */
5014         public function getCustomDefaultSort() {
5015                 return $this->mDefaultSort;
5016         }
5017
5018         /**
5019          * Try to guess the section anchor name based on a wikitext fragment
5020          * presumably extracted from a heading, for example "Header" from
5021          * "== Header ==".
5022          */
5023         public function guessSectionNameFromWikiText( $text ) {
5024                 # Strip out wikitext links(they break the anchor)
5025                 $text = $this->stripSectionName( $text );
5026                 $headline = Sanitizer::decodeCharReferences( $text );
5027                 # strip out HTML
5028                 $headline = StringUtils::delimiterReplace( '<', '>', '', $headline );
5029                 $headline = trim( $headline );
5030                 $sectionanchor = '#' . urlencode( str_replace( ' ', '_', $headline ) );
5031                 $replacearray = array(
5032                         '%3A' => ':',
5033                         '%' => '.'
5034                 );
5035                 return str_replace(
5036                         array_keys( $replacearray ),
5037                         array_values( $replacearray ),
5038                         $sectionanchor );
5039         }
5040
5041         /**
5042          * Strips a text string of wikitext for use in a section anchor
5043          *
5044          * Accepts a text string and then removes all wikitext from the
5045          * string and leaves only the resultant text (i.e. the result of
5046          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5047          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5048          * to create valid section anchors by mimicing the output of the
5049          * parser when headings are parsed.
5050          *
5051          * @param $text string Text string to be stripped of wikitext
5052          * for use in a Section anchor
5053          * @return Filtered text string
5054          */
5055         public function stripSectionName( $text ) {
5056                 # Strip internal link markup
5057                 $text = preg_replace('/\[\[:?([^[|]+)\|([^[]+)\]\]/','$2',$text);
5058                 $text = preg_replace('/\[\[:?([^[]+)\|?\]\]/','$1',$text);
5059
5060                 # Strip external link markup (FIXME: Not Tolerant to blank link text
5061                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
5062                 # on how many empty links there are on the page - need to figure that out.
5063                 $text = preg_replace('/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/','$2',$text);
5064
5065                 # Parse wikitext quotes (italics & bold)
5066                 $text = $this->doQuotes($text);
5067
5068                 # Strip HTML tags
5069                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5070                 return $text;
5071         }
5072
5073         function srvus( $text ) {
5074                 return $this->testSrvus( $text, $this->mOutputType );
5075         }
5076
5077         /**
5078          * strip/replaceVariables/unstrip for preprocessor regression testing
5079          */
5080         function testSrvus( $text, $title, $options, $outputType = self::OT_HTML ) {
5081                 $this->clearState();
5082                 if ( ! ( $title instanceof Title ) ) {
5083                         $title = Title::newFromText( $title );
5084                 }
5085                 $this->mTitle = $title;
5086                 $this->mOptions = $options;
5087                 $this->setOutputType( $outputType );
5088                 $text = $this->replaceVariables( $text );
5089                 $text = $this->mStripState->unstripBoth( $text );
5090                 $text = Sanitizer::removeHTMLtags( $text );
5091                 return $text;
5092         }
5093
5094         function testPst( $text, $title, $options ) {
5095                 global $wgUser;
5096                 if ( ! ( $title instanceof Title ) ) {
5097                         $title = Title::newFromText( $title );
5098                 }
5099                 return $this->preSaveTransform( $text, $title, $wgUser, $options );
5100         }
5101
5102         function testPreprocess( $text, $title, $options ) {
5103                 if ( ! ( $title instanceof Title ) ) {
5104                         $title = Title::newFromText( $title );
5105                 }
5106                 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
5107         }
5108
5109         function markerSkipCallback( $s, $callback ) {
5110                 $i = 0;
5111                 $out = '';
5112                 while ( $i < strlen( $s ) ) {
5113                         $markerStart = strpos( $s, $this->mUniqPrefix, $i );
5114                         if ( $markerStart === false ) {
5115                                 $out .= call_user_func( $callback, substr( $s, $i ) );
5116                                 break;
5117                         } else {
5118                                 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5119                                 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
5120                                 if ( $markerEnd === false ) {
5121                                         $out .= substr( $s, $markerStart );
5122                                         break;
5123                                 } else {
5124                                         $markerEnd += strlen( self::MARKER_SUFFIX );
5125                                         $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5126                                         $i = $markerEnd;
5127                                 }
5128                         }
5129                 }
5130                 return $out;
5131         }
5132
5133         function serialiseHalfParsedText( $text ) {
5134                 $data = array();
5135                 $data['text'] = $text;
5136
5137                 // First, find all strip markers, and store their
5138                 //  data in an array.
5139                 $stripState = new StripState;
5140                 $pos = 0;
5141                 while( ( $start_pos = strpos( $text, $this->mUniqPrefix, $pos ) ) && ( $end_pos = strpos( $text, self::MARKER_SUFFIX, $pos ) ) ) {
5142                         $end_pos += strlen( self::MARKER_SUFFIX );
5143                         $marker = substr( $text, $start_pos, $end_pos-$start_pos );
5144
5145                         if ( !empty( $this->mStripState->general->data[$marker] ) ) {
5146                                 $replaceArray = $stripState->general;
5147                                 $stripText = $this->mStripState->general->data[$marker];
5148                         } elseif ( !empty( $this->mStripState->nowiki->data[$marker] ) ) {
5149                                 $replaceArray = $stripState->nowiki;
5150                                 $stripText = $this->mStripState->nowiki->data[$marker];
5151                         } else {
5152                                 throw new MWException( "Hanging strip marker: '$marker'." );
5153                         }
5154
5155                         $replaceArray->setPair( $marker, $stripText );
5156                         $pos = $end_pos;
5157                 }
5158                 $data['stripstate'] = $stripState;
5159
5160                 // Now, find all of our links, and store THEIR
5161                 //  data in an array! :)
5162                 $links = array( 'internal' => array(), 'interwiki' => array() );
5163                 $pos = 0;
5164
5165                 // Internal links
5166                 while( ( $start_pos = strpos( $text, '<!--LINK ', $pos ) ) ) {
5167                         list( $ns, $trail ) = explode( ':', substr( $text, $start_pos + strlen( '<!--LINK ' ) ), 2 );
5168
5169                         $ns = trim($ns);
5170                         if (empty( $links['internal'][$ns] )) {
5171                                 $links['internal'][$ns] = array();
5172                         }
5173
5174                         $key = trim( substr( $trail, 0, strpos( $trail, '-->' ) ) );
5175                         $links['internal'][$ns][] = $this->mLinkHolders->internals[$ns][$key];
5176                         $pos = $start_pos + strlen( "<!--LINK $ns:$key-->" );
5177                 }
5178
5179                 $pos = 0;
5180
5181                 // Interwiki links
5182                 while( ( $start_pos = strpos( $text, '<!--IWLINK ', $pos ) ) ) {
5183                         $data = substr( $text, $start_pos );
5184                         $key = trim( substr( $data, 0, strpos( $data, '-->' ) ) );
5185                         $links['interwiki'][] = $this->mLinkHolders->interwiki[$key];
5186                         $pos = $start_pos + strlen( "<!--IWLINK $key-->" );
5187                 }
5188
5189                 $data['linkholder'] = $links;
5190
5191                 return $data;
5192         }
5193
5194         function unserialiseHalfParsedText( $data, $intPrefix = null /* Unique identifying prefix */ ) {
5195                 if (!$intPrefix)
5196                         $intPrefix = $this->getRandomString();
5197
5198                 // First, extract the strip state.
5199                 $stripState = $data['stripstate'];
5200                 $this->mStripState->general->merge( $stripState->general );
5201                 $this->mStripState->nowiki->merge( $stripState->nowiki );
5202
5203                 // Now, extract the text, and renumber links
5204                 $text = $data['text'];
5205                 $links = $data['linkholder'];
5206
5207                 // Internal...
5208                 foreach( $links['internal'] as $ns => $nsLinks ) {
5209                         foreach( $nsLinks as $key => $entry ) {
5210                                 $newKey = $intPrefix . '-' . $key;
5211                                 $this->mLinkHolders->internals[$ns][$newKey] = $entry;
5212
5213                                 $text = str_replace( "<!--LINK $ns:$key-->", "<!--LINK $ns:$newKey-->", $text );
5214                         }
5215                 }
5216
5217                 // Interwiki...
5218                 foreach( $links['interwiki'] as $key => $entry ) {
5219                         $newKey = "$intPrefix-$key";
5220                         $this->mLinkHolders->interwikis[$newKey] = $entry;
5221
5222                         $text = str_replace( "<!--IWLINK $key-->", "<!--IWLINK $newKey-->", $text );
5223                 }
5224
5225                 // Should be good to go.
5226                 return $text;
5227         }
5228 }
5229
5230 /**
5231  * @todo document, briefly.
5232  * @ingroup Parser
5233  */
5234 class StripState {
5235         var $general, $nowiki;
5236
5237         function __construct() {
5238                 $this->general = new ReplacementArray;
5239                 $this->nowiki = new ReplacementArray;
5240         }
5241
5242         function unstripGeneral( $text ) {
5243                 wfProfileIn( __METHOD__ );
5244                 do {
5245                         $oldText = $text;
5246                         $text = $this->general->replace( $text );
5247                 } while ( $text !== $oldText );
5248                 wfProfileOut( __METHOD__ );
5249                 return $text;
5250         }
5251
5252         function unstripNoWiki( $text ) {
5253                 wfProfileIn( __METHOD__ );
5254                 do {
5255                         $oldText = $text;
5256                         $text = $this->nowiki->replace( $text );
5257                 } while ( $text !== $oldText );
5258                 wfProfileOut( __METHOD__ );
5259                 return $text;
5260         }
5261
5262         function unstripBoth( $text ) {
5263                 wfProfileIn( __METHOD__ );
5264                 do {
5265                         $oldText = $text;
5266                         $text = $this->general->replace( $text );
5267                         $text = $this->nowiki->replace( $text );
5268                 } while ( $text !== $oldText );
5269                 wfProfileOut( __METHOD__ );
5270                 return $text;
5271         }
5272 }
5273
5274 /**
5275  * @todo document, briefly.
5276  * @ingroup Parser
5277  */
5278 class OnlyIncludeReplacer {
5279         var $output = '';
5280
5281         function replace( $matches ) {
5282                 if ( substr( $matches[1], -1 ) === "\n" ) {
5283                         $this->output .= substr( $matches[1], 0, -1 );
5284                 } else {
5285                         $this->output .= $matches[1];
5286                 }
5287         }
5288 }