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                 $mImageParams, $mImageParamsMagicArray, $mStripList, $mMarkerIndex, $mPreprocessor,
  95                 $mExtLinkBracketedRegex, $mUrlProtocols, $mDefaultStripList, $mVarCache, $mConf,
  96                 $mFunctionTagHooks, $mDoTitleConvert, $mDoContentConvert;
  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( 'nowiki', 'gallery', 'a' );
 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                 $this->mDoTitleConvert = true;
 151                 $this->mDoContentConvert = true;
 152         }
 153
 154         /**
 155          * Reduce memory usage to reduce the impact of circular references
 156          */
 157         function __destruct() {
 158                 if ( isset( $this->mLinkHolders ) ) {
 159                         $this->mLinkHolders->__destruct();
 160                 }
 161                 foreach ( $this as $name => $value ) {
 162                         unset( $this->$name );
 163                 }
 164         }
 165
 166         /**
 167          * Do various kinds of initialisation on the first call of the parser
 168          */
 169         function firstCallInit() {
 170                 if ( !$this->mFirstCall ) {
 171                         return;
 172                 }
 173                 $this->mFirstCall = false;
 174
 175                 wfProfileIn( __METHOD__ );
 176
 177                 $this->setHook( 'pre', array( $this, 'renderPreTag' ) );
 178                 CoreParserFunctions::register( $this );
 179                 $this->initialiseVariables();
 180
 181                 wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
 182                 wfProfileOut( __METHOD__ );
 183         }
 184
 185         /**
 186          * Clear Parser state
 187          *
 188          * @private
 189          */
 190         function clearState() {
 191                 wfProfileIn( __METHOD__ );
 192                 if ( $this->mFirstCall ) {
 193                         $this->firstCallInit();
 194                 }
 195                 $this->mOutput = new ParserOutput;
 196                 $this->mAutonumber = 0;
 197                 $this->mLastSection = '';
 198                 $this->mDTopen = false;
 199                 $this->mIncludeCount = array();
 200                 $this->mStripState = new StripState;
 201                 $this->mArgStack = false;
 202                 $this->mInPre = false;
 203                 $this->mLinkHolders = new LinkHolderArray( $this );
 204                 $this->mLinkID = 0;
 205                 $this->mRevisionTimestamp = $this->mRevisionId = null;
 206                 $this->mVarCache = array();
 207
 208                 /**
 209                  * Prefix for temporary replacement strings for the multipass parser.
 210                  * \x07 should never appear in input as it's disallowed in XML.
 211                  * Using it at the front also gives us a little extra robustness
 212                  * since it shouldn't match when butted up against identifier-like
 213                  * string constructs.
 214                  *
 215                  * Must not consist of all title characters, or else it will change
 216                  * the behaviour of <nowiki> in a link.
 217                  */
 218                 #$this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 219                 # Changed to \x7f to allow XML double-parsing -- TS
 220                 $this->mUniqPrefix = "\x7fUNIQ" . self::getRandomString();
 221
 222
 223                 # Clear these on every parse, bug 4549
 224                 $this->mTplExpandCache = $this->mTplRedirCache = $this->mTplDomCache = array();
 225
 226                 $this->mShowToc = true;
 227                 $this->mForceTocPosition = false;
 228                 $this->mIncludeSizes = array(
 229                         'post-expand' => 0,
 230                         'arg' => 0,
 231                 );
 232                 $this->mPPNodeCount = 0;
 233                 $this->mDefaultSort = false;
 234                 $this->mHeadings = array();
 235                 $this->mDoubleUnderscores = array();
 236                 $this->mExpensiveFunctionCount = 0;
 237
 238                 # Fix cloning
 239                 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
 240                         $this->mPreprocessor = null;
 241                 }
 242
 243                 wfRunHooks( 'ParserClearState', array( &$this ) );
 244                 wfProfileOut( __METHOD__ );
 245         }
 246
 247         function setOutputType( $ot ) {
 248                 $this->mOutputType = $ot;
 249                 // Shortcut alias
 250                 $this->ot = array(
 251                         'html' => $ot == self::OT_HTML,
 252                         'wiki' => $ot == self::OT_WIKI,
 253                         'pre' => $ot == self::OT_PREPROCESS,
 254                 );
 255         }
 256
 257         /**
 258          * Set the context title
 259          */
 260         function setTitle( $t ) {
 261                 if ( $t && !($t instanceOf FakeTitle)
 262                          && !($t instanceOf Title)
 263                          && is_string( "$t" ) ) {
 264                         // If don't have a Title object, make sure we can convert
 265                         // whatever we've been passed to a string.
 266                         $t = Title::newFromText( "$t" );
 267                 }
 268
 269                 if ( !($t instanceOf Title) ) {
 270                         $t = Title::newFromText( 'NO TITLE' );
 271                 }
 272
 273                 if ( strval( $t->getFragment() ) !== '' ) {
 274                         # Strip the fragment to avoid various odd effects
 275                         $this->mTitle = clone $t;
 276                         $this->mTitle->setFragment( '' );
 277                 } else {
 278                         $this->mTitle = $t;
 279                 }
 280         }
 281
 282         /**
 283          * Accessor for mUniqPrefix.
 284          *
 285          * @public
 286          */
 287         function uniqPrefix() {
 288                 if( !isset( $this->mUniqPrefix ) ) {
 289                         // @todo Fixme: this is probably *horribly wrong*
 290                         // LanguageConverter seems to want $wgParser's uniqPrefix, however
 291                         // if this is called for a parser cache hit, the parser may not
 292                         // have ever been initialized in the first place.
 293                         // Not really sure what the heck is supposed to be going on here.
 294                         return '';
 295                         //throw new MWException( "Accessing uninitialized mUniqPrefix" );
 296                 }
 297                 return $this->mUniqPrefix;
 298         }
 299
 300         /**
 301          * Convert wikitext to HTML
 302          * Do not call this function recursively.
 303          *
 304          * @param $text String: text we want to parse
 305          * @param $title A title object
 306          * @param $options ParserOptions
 307          * @param $linestart boolean
 308          * @param $clearState boolean
 309          * @param $revid Int: number to pass in {{REVISIONID}}
 310          * @return ParserOutput a ParserOutput
 311          */
 312         public function parse( $text, Title $title, ParserOptions $options, $linestart = true, $clearState = true, $revid = null ) {
 313                 /**
 314                  * First pass--just handle <nowiki> sections, pass the rest off
 315                  * to internalParse() which does all the real work.
 316                  */
 317
 318                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
 319                 $fname = __METHOD__.'-' . wfGetCaller();
 320                 wfProfileIn( __METHOD__ );
 321                 wfProfileIn( $fname );
 322
 323                 if ( $clearState ) {
 324                         $this->clearState();
 325                 }
 326
 327                 $this->mOptions = $options;
 328                 $this->setTitle( $title ); // Page title has to be set for the pre-processor
 329
 330                 $oldRevisionId = $this->mRevisionId;
 331                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 332                 if( $revid !== null ) {
 333                         $this->mRevisionId = $revid;
 334                         $this->mRevisionTimestamp = null;
 335                 }
 336                 $this->setOutputType( self::OT_HTML );
 337                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 338                 # No more strip!
 339                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 340                 $text = $this->internalParse( $text );
 341
 342                 $text = $this->mStripState->unstripGeneral( $text );
 343
 344                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 345                 $fixtags = array(
 346                         # french spaces, last one Guillemet-left
 347                         # only if there is something before the space
 348                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&nbsp;\\2',
 349                         # french spaces, Guillemet-right
 350                         '/(\\302\\253) /' => '\\1&nbsp;',
 351                         '/&nbsp;(!\s*important)/' => ' \\1', #Beware of CSS magic word !important, bug #11874.
 352                 );
 353                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 354
 355                 $text = $this->doBlockLevels( $text, $linestart );
 356
 357                 $this->replaceLinkHolders( $text );
 358
 359                 // The position of the convert() call should not be changed. it
 360                 // assumes that the links are all replaced and the only thing left
 361                 // is the <nowiki> mark.
 362                 if ( $this->mDoContentConvert && !$this->mTitle->isConversionTable()) {
 363                         $text = $wgContLang->convert( $text );
 364                 }
 365
 366                 // A title may have been set in a conversion rule.
 367                 // Note that if a user tries to set a title in a conversion
 368                 // rule but content conversion was not done, then the parser
 369                 // won't pick it up.  This is probably expected behavior.
 370                 if ( $wgContLang->getConvRuleTitle() ) {
 371                         $this->setTitle( $wgContLang->getConvRuleTitle() );
 372                 } elseif ( $this->mDoTitleConvert && !$this->mTitle->isConversionTable() ) {
 373                         $this->setTitle( $wgContLang->convert( $title ) );
 374                 }
 375
 376                 $text = $this->mStripState->unstripNoWiki( $text );
 377
 378                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 379
 380 //!JF Move to its own function
 381
 382                 $uniq_prefix = $this->mUniqPrefix;
 383                 $matches = array();
 384                 $elements = array_keys( $this->mTransparentTagHooks );
 385                 $text = self::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 386
 387                 foreach( $matches as $marker => $data ) {
 388                         list( $element, $content, $params, $tag ) = $data;
 389                         $tagName = strtolower( $element );
 390                         if( isset( $this->mTransparentTagHooks[$tagName] ) ) {
 391                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName],
 392                                         array( $content, $params, $this ) );
 393                         } else {
 394                                 $output = $tag;
 395                         }
 396                         $this->mStripState->general->setPair( $marker, $output );
 397                 }
 398                 $text = $this->mStripState->unstripGeneral( $text );
 399
 400                 $text = Sanitizer::normalizeCharReferences( $text );
 401
 402                 if ( ( $wgUseTidy && $this->mOptions->mTidy ) || $wgAlwaysUseTidy ) {
 403                         $text = MWTidy::tidy( $text );
 404                 } else {
 405                         # attempt to sanitize at least some nesting problems
 406                         # (bug #2702 and quite a few others)
 407                         $tidyregs = array(
 408                                 # ''Something [http://www.cool.com cool''] -->
 409                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 410                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 411                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 412                                 # fix up an anchor inside another anchor, only
 413                                 # at least for a single single nested link (bug 3695)
 414                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 415                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 416                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 417                                 # contains a div, so fix it up here; replace
 418                                 # div with escaped text
 419                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 420                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 421                                 # remove empty italic or bold tag pairs, some
 422                                 # introduced by rules above
 423                                 '/<([bi])><\/\\1>/' => '',
 424                         );
 425
 426                         $text = preg_replace(
 427                                 array_keys( $tidyregs ),
 428                                 array_values( $tidyregs ),
 429                                 $text );
 430                 }
 431                 global $wgExpensiveParserFunctionLimit;
 432                 if ( $this->mExpensiveFunctionCount > $wgExpensiveParserFunctionLimit ) {
 433                         $this->limitationWarn( 'expensive-parserfunction', $this->mExpensiveFunctionCount, $wgExpensiveParserFunctionLimit );
 434                 }
 435
 436                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 437
 438                 # Information on include size limits, for the benefit of users who try to skirt them
 439                 if ( $this->mOptions->getEnableLimitReport() ) {
 440                         global $wgExpensiveParserFunctionLimit;
 441                         $max = $this->mOptions->getMaxIncludeSize();
 442                         $PFreport = "Expensive parser function count: {$this->mExpensiveFunctionCount}/$wgExpensiveParserFunctionLimit\n";
 443                         $limitReport =
 444                                 "NewPP limit report\n" .
 445                                 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->mMaxPPNodeCount}\n" .
 446                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 447                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n".
 448                                 $PFreport;
 449                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 450                         $text .= "\n<!-- \n$limitReport-->\n";
 451                 }
 452                 $this->mOutput->setText( $text );
 453                 $this->mRevisionId = $oldRevisionId;
 454                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 455                 wfProfileOut( $fname );
 456                 wfProfileOut( __METHOD__ );
 457
 458                 return $this->mOutput;
 459         }
 460
 461         /**
 462          * Recursive parser entry point that can be called from an extension tag
 463          * hook.
 464          *
 465          * If $frame is not provided, then template variables (e.g., {{{1}}}) within $text are not expanded
 466          *
 467          * @param $text String: text extension wants to have parsed
 468          * @param PPFrame $frame: The frame to use for expanding any template variables
 469          */
 470         function recursiveTagParse( $text, $frame=false ) {
 471                 wfProfileIn( __METHOD__ );
 472                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 473                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 474                 $text = $this->internalParse( $text, false, $frame );
 475                 wfProfileOut( __METHOD__ );
 476                 return $text;
 477         }
 478
 479         /**
 480          * Expand templates and variables in the text, producing valid, static wikitext.
 481          * Also removes comments.
 482          */
 483         function preprocess( $text, $title, $options, $revid = null ) {
 484                 wfProfileIn( __METHOD__ );
 485                 $this->clearState();
 486                 $this->setOutputType( self::OT_PREPROCESS );
 487                 $this->mOptions = $options;
 488                 $this->setTitle( $title );
 489                 if( $revid !== null ) {
 490                         $this->mRevisionId = $revid;
 491                 }
 492                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 493                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 494                 $text = $this->replaceVariables( $text );
 495                 $text = $this->mStripState->unstripBoth( $text );
 496                 wfProfileOut( __METHOD__ );
 497                 return $text;
 498         }
 499
 500         /**
 501          * Get a random string
 502          *
 503          * @private
 504          * @static
 505          */
 506         function getRandomString() {
 507                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 508         }
 509
 510         function &getTitle() { return $this->mTitle; }
 511         function getOptions() { return $this->mOptions; }
 512         function getRevisionId() { return $this->mRevisionId; }
 513         function getOutput() { return $this->mOutput; }
 514         function nextLinkID() { return $this->mLinkID++; }
 515
 516         function getFunctionLang() {
 517                 global $wgLang, $wgContLang;
 518
 519                 $target = $this->mOptions->getTargetLanguage();
 520                 if ( $target !== null ) {
 521                         return $target;
 522                 } else {
 523                         return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 524                 }
 525         }
 526
 527         /**
 528          * Get a preprocessor object
 529          */
 530         function getPreprocessor() {
 531                 if ( !isset( $this->mPreprocessor ) ) {
 532                         $class = $this->mPreprocessorClass;
 533                         $this->mPreprocessor = new $class( $this );
 534                 }
 535                 return $this->mPreprocessor;
 536         }
 537
 538         /**
 539          * Replaces all occurrences of HTML-style comments and the given tags
 540          * in the text with a random marker and returns the next text. The output
 541          * parameter $matches will be an associative array filled with data in
 542          * the form:
 543          *   'UNIQ-xxxxx' => array(
 544          *     'element',
 545          *     'tag content',
 546          *     array( 'param' => 'x' ),
 547          *     '<element param="x">tag content</element>' ) )
 548          *
 549          * @param $elements list of element names. Comments are always extracted.
 550          * @param $text Source text string.
 551          * @param $uniq_prefix
 552          *
 553          * @public
 554          * @static
 555          */
 556         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 557                 static $n = 1;
 558                 $stripped = '';
 559                 $matches = array();
 560
 561                 $taglist = implode( '|', $elements );
 562                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
 563
 564                 while ( $text != '' ) {
 565                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 566                         $stripped .= $p[0];
 567                         if( count( $p ) < 5 ) {
 568                                 break;
 569                         }
 570                         if( count( $p ) > 5 ) {
 571                                 // comment
 572                                 $element    = $p[4];
 573                                 $attributes = '';
 574                                 $close      = '';
 575                                 $inside     = $p[5];
 576                         } else {
 577                                 // tag
 578                                 $element    = $p[1];
 579                                 $attributes = $p[2];
 580                                 $close      = $p[3];
 581                                 $inside     = $p[4];
 582                         }
 583
 584                         $marker = "$uniq_prefix-$element-" . sprintf('%08X', $n++) . self::MARKER_SUFFIX;
 585                         $stripped .= $marker;
 586
 587                         if ( $close === '/>' ) {
 588                                 // Empty element tag, <tag />
 589                                 $content = null;
 590                                 $text = $inside;
 591                                 $tail = null;
 592                         } else {
 593                                 if( $element === '!--' ) {
 594                                         $end = '/(-->)/';
 595                                 } else {
 596                                         $end = "/(<\\/$element\\s*>)/i";
 597                                 }
 598                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 599                                 $content = $q[0];
 600                                 if( count( $q ) < 3 ) {
 601                                         # No end tag -- let it run out to the end of the text.
 602                                         $tail = '';
 603                                         $text = '';
 604                                 } else {
 605                                         $tail = $q[1];
 606                                         $text = $q[2];
 607                                 }
 608                         }
 609
 610                         $matches[$marker] = array( $element,
 611                                 $content,
 612                                 Sanitizer::decodeTagAttributes( $attributes ),
 613                                 "<$element$attributes$close$content$tail" );
 614                 }
 615                 return $stripped;
 616         }
 617
 618         /**
 619          * Get a list of strippable XML-like elements
 620          */
 621         function getStripList() {
 622                 global $wgRawHtml;
 623                 $elements = $this->mStripList;
 624                 if( $wgRawHtml ) {
 625                         $elements[] = 'html';
 626                 }
 627                 if( $this->mOptions->getUseTeX() ) {
 628                         $elements[] = 'math';
 629                 }
 630                 return $elements;
 631         }
 632
 633         /**
 634          * @deprecated use replaceVariables
 635          */
 636         function strip( $text, $state, $stripcomments = false , $dontstrip = array () ) {
 637                 return $text;
 638         }
 639
 640         /**
 641          * Restores pre, math, and other extensions removed by strip()
 642          *
 643          * always call unstripNoWiki() after this one
 644          * @private
 645          * @deprecated use $this->mStripState->unstrip()
 646          */
 647         function unstrip( $text, $state ) {
 648                 return $state->unstripGeneral( $text );
 649         }
 650
 651         /**
 652          * Always call this after unstrip() to preserve the order
 653          *
 654          * @private
 655          * @deprecated use $this->mStripState->unstrip()
 656          */
 657         function unstripNoWiki( $text, $state ) {
 658                 return $state->unstripNoWiki( $text );
 659         }
 660
 661         /**
 662          * @deprecated use $this->mStripState->unstripBoth()
 663          */
 664         function unstripForHTML( $text ) {
 665                 return $this->mStripState->unstripBoth( $text );
 666         }
 667
 668         /**
 669          * Add an item to the strip state
 670          * Returns the unique tag which must be inserted into the stripped text
 671          * The tag will be replaced with the original text in unstrip()
 672          *
 673          * @private
 674          */
 675         function insertStripItem( $text ) {
 676                 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
 677                 $this->mMarkerIndex++;
 678                 $this->mStripState->general->setPair( $rnd, $text );
 679                 return $rnd;
 680         }
 681
 682         /**
 683          * Interface with html tidy
 684          * @deprecated Use MWTidy::tidy()
 685          */
 686         public static function tidy( $text ) {
 687                 wfDeprecated( __METHOD__ );
 688                 return MWTidy::tidy( $text );
 689         }
 690
 691         /**
 692          * parse the wiki syntax used to render tables
 693          *
 694          * @private
 695          */
 696         function doTableStuff ( $text ) {
 697                 wfProfileIn( __METHOD__ );
 698
 699                 $lines = StringUtils::explode( "\n", $text );
 700                 $out = '';
 701                 $td_history = array (); // Is currently a td tag open?
 702                 $last_tag_history = array (); // Save history of last lag activated (td, th or caption)
 703                 $tr_history = array (); // Is currently a tr tag open?
 704                 $tr_attributes = array (); // history of tr attributes
 705                 $has_opened_tr = array(); // Did this table open a <tr> element?
 706                 $indent_level = 0; // indent level of the table
 707
 708                 foreach ( $lines as $outLine ) {
 709                         $line = trim( $outLine );
 710
 711                         if( $line == '' ) { // empty line, go to next line
 712                                 $out .= $outLine."\n";
 713                                 continue;
 714                         }
 715                         $first_character = $line[0];
 716                         $matches = array();
 717
 718                         if ( preg_match( '/^(:*)\{\|(.*)$/', $line , $matches ) ) {
 719                                 // First check if we are starting a new table
 720                                 $indent_level = strlen( $matches[1] );
 721
 722                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 723                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'table' );
 724
 725                                 $outLine = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 726                                 array_push ( $td_history , false );
 727                                 array_push ( $last_tag_history , '' );
 728                                 array_push ( $tr_history , false );
 729                                 array_push ( $tr_attributes , '' );
 730                                 array_push ( $has_opened_tr , false );
 731                         } else if ( count ( $td_history ) == 0 ) {
 732                                 // Don't do any of the following
 733                                 $out .= $outLine."\n";
 734                                 continue;
 735                         } else if ( substr ( $line , 0 , 2 ) === '|}' ) {
 736                                 // We are ending a table
 737                                 $line = '</table>' . substr ( $line , 2 );
 738                                 $last_tag = array_pop ( $last_tag_history );
 739
 740                                 if ( !array_pop ( $has_opened_tr ) ) {
 741                                         $line = "<tr><td></td></tr>{$line}";
 742                                 }
 743
 744                                 if ( array_pop ( $tr_history ) ) {
 745                                         $line = "</tr>{$line}";
 746                                 }
 747
 748                                 if ( array_pop ( $td_history ) ) {
 749                                         $line = "</{$last_tag}>{$line}";
 750                                 }
 751                                 array_pop ( $tr_attributes );
 752                                 $outLine = $line . str_repeat( '</dd></dl>' , $indent_level );
 753                         } else if ( substr ( $line , 0 , 2 ) === '|-' ) {
 754                                 // Now we have a table row
 755                                 $line = preg_replace( '#^\|-+#', '', $line );
 756
 757                                 // Whats after the tag is now only attributes
 758                                 $attributes = $this->mStripState->unstripBoth( $line );
 759                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'tr' );
 760                                 array_pop ( $tr_attributes );
 761                                 array_push ( $tr_attributes , $attributes );
 762
 763                                 $line = '';
 764                                 $last_tag = array_pop ( $last_tag_history );
 765                                 array_pop ( $has_opened_tr );
 766                                 array_push ( $has_opened_tr , true );
 767
 768                                 if ( array_pop ( $tr_history ) ) {
 769                                         $line = '</tr>';
 770                                 }
 771
 772                                 if ( array_pop ( $td_history ) ) {
 773                                         $line = "</{$last_tag}>{$line}";
 774                                 }
 775
 776                                 $outLine = $line;
 777                                 array_push ( $tr_history , false );
 778                                 array_push ( $td_history , false );
 779                                 array_push ( $last_tag_history , '' );
 780                         }
 781                         else if ( $first_character === '|' || $first_character === '!' || substr ( $line , 0 , 2 )  === '|+' ) {
 782                                 // This might be cell elements, td, th or captions
 783                                 if ( substr ( $line , 0 , 2 ) === '|+' ) {
 784                                         $first_character = '+';
 785                                         $line = substr ( $line , 1 );
 786                                 }
 787
 788                                 $line = substr ( $line , 1 );
 789
 790                                 if ( $first_character === '!' ) {
 791                                         $line = str_replace ( '!!' , '||' , $line );
 792                                 }
 793
 794                                 // Split up multiple cells on the same line.
 795                                 // FIXME : This can result in improper nesting of tags processed
 796                                 // by earlier parser steps, but should avoid splitting up eg
 797                                 // attribute values containing literal "||".
 798                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 799
 800                                 $outLine = '';
 801
 802                                 // Loop through each table cell
 803                                 foreach ( $cells as $cell )
 804                                 {
 805                                         $previous = '';
 806                                         if ( $first_character !== '+' )
 807                                         {
 808                                                 $tr_after = array_pop ( $tr_attributes );
 809                                                 if ( !array_pop ( $tr_history ) ) {
 810                                                         $previous = "<tr{$tr_after}>\n";
 811                                                 }
 812                                                 array_push ( $tr_history , true );
 813                                                 array_push ( $tr_attributes , '' );
 814                                                 array_pop ( $has_opened_tr );
 815                                                 array_push ( $has_opened_tr , true );
 816                                         }
 817
 818                                         $last_tag = array_pop ( $last_tag_history );
 819
 820                                         if ( array_pop ( $td_history ) ) {
 821                                                 $previous = "</{$last_tag}>{$previous}";
 822                                         }
 823
 824                                         if ( $first_character === '|' ) {
 825                                                 $last_tag = 'td';
 826                                         } else if ( $first_character === '!' ) {
 827                                                 $last_tag = 'th';
 828                                         } else if ( $first_character === '+' ) {
 829                                                 $last_tag = 'caption';
 830                                         } else {
 831                                                 $last_tag = '';
 832                                         }
 833
 834                                         array_push ( $last_tag_history , $last_tag );
 835
 836                                         // A cell could contain both parameters and data
 837                                         $cell_data = explode ( '|' , $cell , 2 );
 838
 839                                         // Bug 553: Note that a '|' inside an invalid link should not
 840                                         // be mistaken as delimiting cell parameters
 841                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 842                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 843                                         } else if ( count ( $cell_data ) == 1 )
 844                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 845                                         else {
 846                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 847                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 848                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 849                                         }
 850
 851                                         $outLine .= $cell;
 852                                         array_push ( $td_history , true );
 853                                 }
 854                         }
 855                         $out .= $outLine . "\n";
 856                 }
 857
 858                 // Closing open td, tr && table
 859                 while ( count ( $td_history ) > 0 )
 860                 {
 861                         if ( array_pop ( $td_history ) ) {
 862                                 $out .= "</td>\n";
 863                         }
 864                         if ( array_pop ( $tr_history ) ) {
 865                                 $out .= "</tr>\n";
 866                         }
 867                         if ( !array_pop ( $has_opened_tr ) ) {
 868                                 $out .= "<tr><td></td></tr>\n" ;
 869                         }
 870
 871                         $out .= "</table>\n";
 872                 }
 873
 874                 // Remove trailing line-ending (b/c)
 875                 if ( substr( $out, -1 ) === "\n" ) {
 876                         $out = substr( $out, 0, -1 );
 877                 }
 878
 879                 // special case: don't return empty table
 880                 if( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
 881                         $out = '';
 882                 }
 883
 884                 wfProfileOut( __METHOD__ );
 885
 886                 return $out;
 887         }
 888
 889         /**
 890          * Helper function for parse() that transforms wiki markup into
 891          * HTML. Only called for $mOutputType == self::OT_HTML.
 892          *
 893          * @private
 894          */
 895         function internalParse( $text, $isMain = true, $frame=false ) {
 896                 wfProfileIn( __METHOD__ );
 897
 898                 $origText = $text;
 899
 900                 # Hook to suspend the parser in this state
 901                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
 902                         wfProfileOut( __METHOD__ );
 903                         return $text ;
 904                 }
 905
 906                 // if $frame is provided, then use $frame for replacing any variables
 907                 if ($frame) {
 908                         // use frame depth to infer how include/noinclude tags should be handled
 909                         // depth=0 means this is the top-level document; otherwise it's an included document
 910                         if( !$frame->depth )
 911                                 $flag = 0;
 912                         else
 913                                 $flag = Parser::PTD_FOR_INCLUSION;
 914                         $dom = $this->preprocessToDom( $text, $flag );
 915                         $text = $frame->expand( $dom );
 916                 }
 917                 // if $frame is not provided, then use old-style replaceVariables
 918                 else {
 919                         $text = $this->replaceVariables( $text );
 920                 }
 921
 922                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
 923                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
 924
 925                 // Tables need to come after variable replacement for things to work
 926                 // properly; putting them before other transformations should keep
 927                 // exciting things like link expansions from showing up in surprising
 928                 // places.
 929                 $text = $this->doTableStuff( $text );
 930
 931                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 932
 933                 $text = $this->doDoubleUnderscore( $text );
 934
 935                 $text = $this->doHeadings( $text );
 936                 if( $this->mOptions->getUseDynamicDates() ) {
 937                         $df = DateFormatter::getInstance();
 938                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 939                 }
 940                 $text = $this->doAllQuotes( $text );
 941                 $text = $this->replaceInternalLinks( $text );
 942                 $text = $this->replaceExternalLinks( $text );
 943
 944                 # replaceInternalLinks may sometimes leave behind
 945                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 946                 $text = str_replace($this->mUniqPrefix.'NOPARSE', '', $text);
 947
 948                 $text = $this->doMagicLinks( $text );
 949                 $text = $this->formatHeadings( $text, $origText, $isMain );
 950
 951                 wfProfileOut( __METHOD__ );
 952                 return $text;
 953         }
 954
 955         /**
 956          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 957          * magic external links.
 958          *
 959          * DML
 960          * @private
 961          */
 962         function doMagicLinks( $text ) {
 963                 wfProfileIn( __METHOD__ );
 964                 $prots = $this->mUrlProtocols;
 965                 $urlChar = self::EXT_LINK_URL_CLASS;
 966                 $text = preg_replace_callback(
 967                         '!(?:                           # Start cases
 968                                 (<a.*?</a>) |               # m[1]: Skip link text
 969                                 (<.*?>) |                   # m[2]: Skip stuff inside HTML elements' . "
 970                                 (\\b(?:$prots)$urlChar+) |  # m[3]: Free external links" . '
 971                                 (?:RFC|PMID)\s+([0-9]+) |   # m[4]: RFC or PMID, capture number
 972                                 ISBN\s+(\b                  # m[5]: ISBN, capture number
 973                                     (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
 974                                     (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
 975                                     [0-9Xx]                 # check digit
 976                                     \b)
 977                         )!x', array( &$this, 'magicLinkCallback' ), $text );
 978                 wfProfileOut( __METHOD__ );
 979                 return $text;
 980         }
 981
 982         function magicLinkCallback( $m ) {
 983                 if ( isset( $m[1] ) && $m[1] !== '' ) {
 984                         # Skip anchor
 985                         return $m[0];
 986                 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
 987                         # Skip HTML element
 988                         return $m[0];
 989                 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
 990                         # Free external link
 991                         return $this->makeFreeExternalLink( $m[0] );
 992                 } elseif ( isset( $m[4] ) && $m[4] !== '' ) {
 993                         # RFC or PMID
 994                         $CssClass = '';
 995                         if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
 996                                 $keyword = 'RFC';
 997                                 $urlmsg = 'rfcurl';
 998                                 $CssClass = 'mw-magiclink-rfc';
 999                                 $id = $m[4];
1000                         } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1001                                 $keyword = 'PMID';
1002                                 $urlmsg = 'pubmedurl';
1003                                 $CssClass = 'mw-magiclink-pmid';
1004                                 $id = $m[4];
1005                         } else {
1006                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1007                                         substr($m[0], 0, 20 ) . '"' );
1008                         }
1009                         $url = wfMsg( $urlmsg, $id);
1010                         $sk = $this->mOptions->getSkin();
1011                         $la = $sk->getExternalLinkAttributes( "external $CssClass" );
1012                         return "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
1013                 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1014                         # ISBN
1015                         $isbn = $m[5];
1016                         $num = strtr( $isbn, array(
1017                                 '-' => '',
1018                                 ' ' => '',
1019                                 'x' => 'X',
1020                         ));
1021                         $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
1022                         return'<a href="' .
1023                                 $titleObj->escapeLocalUrl() .
1024                                 "\" class=\"internal mw-magiclink-isbn\">ISBN $isbn</a>";
1025                 } else {
1026                         return $m[0];
1027                 }
1028         }
1029
1030         /**
1031          * Make a free external link, given a user-supplied URL
1032          * @return HTML
1033          * @private
1034          */
1035         function makeFreeExternalLink( $url ) {
1036                 global $wgContLang;
1037                 wfProfileIn( __METHOD__ );
1038
1039                 $sk = $this->mOptions->getSkin();
1040                 $trail = '';
1041
1042                 # The characters '<' and '>' (which were escaped by
1043                 # removeHTMLtags()) should not be included in
1044                 # URLs, per RFC 2396.
1045                 $m2 = array();
1046                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1047                         $trail = substr($url, $m2[0][1]) . $trail;
1048                         $url = substr($url, 0, $m2[0][1]);
1049                 }
1050
1051                 # Move trailing punctuation to $trail
1052                 $sep = ',;\.:!?';
1053                 # If there is no left bracket, then consider right brackets fair game too
1054                 if ( strpos( $url, '(' ) === false ) {
1055                         $sep .= ')';
1056                 }
1057
1058                 $numSepChars = strspn( strrev( $url ), $sep );
1059                 if ( $numSepChars ) {
1060                         $trail = substr( $url, -$numSepChars ) . $trail;
1061                         $url = substr( $url, 0, -$numSepChars );
1062                 }
1063
1064                 $url = Sanitizer::cleanUrl( $url );
1065
1066                 # Is this an external image?
1067                 $text = $this->maybeMakeExternalImage( $url );
1068                 if ( $text === false ) {
1069                         # Not an image, make a link
1070                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free',
1071                                 $this->getExternalLinkAttribs( $url ) );
1072                         # Register it in the output object...
1073                         # Replace unnecessary URL escape codes with their equivalent characters
1074                         $pasteurized = self::replaceUnusualEscapes( $url );
1075                         $this->mOutput->addExternalLink( $pasteurized );
1076                 }
1077                 wfProfileOut( __METHOD__ );
1078                 return $text . $trail;
1079         }
1080
1081
1082         /**
1083          * Parse headers and return html
1084          *
1085          * @private
1086          */
1087         function doHeadings( $text ) {
1088                 wfProfileIn( __METHOD__ );
1089                 for ( $i = 6; $i >= 1; --$i ) {
1090                         $h = str_repeat( '=', $i );
1091                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1092                           "<h$i>\\1</h$i>", $text );
1093                 }
1094                 wfProfileOut( __METHOD__ );
1095                 return $text;
1096         }
1097
1098         /**
1099          * Replace single quotes with HTML markup
1100          * @private
1101          * @return string the altered text
1102          */
1103         function doAllQuotes( $text ) {
1104                 wfProfileIn( __METHOD__ );
1105                 $outtext = '';
1106                 $lines = StringUtils::explode( "\n", $text );
1107                 foreach ( $lines as $line ) {
1108                         $outtext .= $this->doQuotes( $line ) . "\n";
1109                 }
1110                 $outtext = substr($outtext, 0,-1);
1111                 wfProfileOut( __METHOD__ );
1112                 return $outtext;
1113         }
1114
1115         /**
1116          * Helper function for doAllQuotes()
1117          */
1118         public function doQuotes( $text ) {
1119                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1120                 if ( count( $arr ) == 1 )
1121                         return $text;
1122                 else
1123                 {
1124                         # First, do some preliminary work. This may shift some apostrophes from
1125                         # being mark-up to being text. It also counts the number of occurrences
1126                         # of bold and italics mark-ups.
1127                         $i = 0;
1128                         $numbold = 0;
1129                         $numitalics = 0;
1130                         foreach ( $arr as $r )
1131                         {
1132                                 if ( ( $i % 2 ) == 1 )
1133                                 {
1134                                         # If there are ever four apostrophes, assume the first is supposed to
1135                                         # be text, and the remaining three constitute mark-up for bold text.
1136                                         if ( strlen( $arr[$i] ) == 4 )
1137                                         {
1138                                                 $arr[$i-1] .= "'";
1139                                                 $arr[$i] = "'''";
1140                                         }
1141                                         # If there are more than 5 apostrophes in a row, assume they're all
1142                                         # text except for the last 5.
1143                                         else if ( strlen( $arr[$i] ) > 5 )
1144                                         {
1145                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1146                                                 $arr[$i] = "'''''";
1147                                         }
1148                                         # Count the number of occurrences of bold and italics mark-ups.
1149                                         # We are not counting sequences of five apostrophes.
1150                                         if ( strlen( $arr[$i] ) == 2 )      { $numitalics++;             }
1151                                         else if ( strlen( $arr[$i] ) == 3 ) { $numbold++;                }
1152                                         else if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
1153                                 }
1154                                 $i++;
1155                         }
1156
1157                         # If there is an odd number of both bold and italics, it is likely
1158                         # that one of the bold ones was meant to be an apostrophe followed
1159                         # by italics. Which one we cannot know for certain, but it is more
1160                         # likely to be one that has a single-letter word before it.
1161                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1162                         {
1163                                 $i = 0;
1164                                 $firstsingleletterword = -1;
1165                                 $firstmultiletterword = -1;
1166                                 $firstspace = -1;
1167                                 foreach ( $arr as $r )
1168                                 {
1169                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1170                                         {
1171                                                 $x1 = substr ($arr[$i-1], -1);
1172                                                 $x2 = substr ($arr[$i-1], -2, 1);
1173                                                 if ($x1 === ' ') {
1174                                                         if ($firstspace == -1) $firstspace = $i;
1175                                                 } else if ($x2 === ' ') {
1176                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1177                                                 } else {
1178                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1179                                                 }
1180                                         }
1181                                         $i++;
1182                                 }
1183
1184                                 # If there is a single-letter word, use it!
1185                                 if ($firstsingleletterword > -1)
1186                                 {
1187                                         $arr [ $firstsingleletterword ] = "''";
1188                                         $arr [ $firstsingleletterword-1 ] .= "'";
1189                                 }
1190                                 # If not, but there's a multi-letter word, use that one.
1191                                 else if ($firstmultiletterword > -1)
1192                                 {
1193                                         $arr [ $firstmultiletterword ] = "''";
1194                                         $arr [ $firstmultiletterword-1 ] .= "'";
1195                                 }
1196                                 # ... otherwise use the first one that has neither.
1197                                 # (notice that it is possible for all three to be -1 if, for example,
1198                                 # there is only one pentuple-apostrophe in the line)
1199                                 else if ($firstspace > -1)
1200                                 {
1201                                         $arr [ $firstspace ] = "''";
1202                                         $arr [ $firstspace-1 ] .= "'";
1203                                 }
1204                         }
1205
1206                         # Now let's actually convert our apostrophic mush to HTML!
1207                         $output = '';
1208                         $buffer = '';
1209                         $state = '';
1210                         $i = 0;
1211                         foreach ($arr as $r)
1212                         {
1213                                 if (($i % 2) == 0)
1214                                 {
1215                                         if ($state === 'both')
1216                                                 $buffer .= $r;
1217                                         else
1218                                                 $output .= $r;
1219                                 }
1220                                 else
1221                                 {
1222                                         if (strlen ($r) == 2)
1223                                         {
1224                                                 if ($state === 'i')
1225                                                 { $output .= '</i>'; $state = ''; }
1226                                                 else if ($state === 'bi')
1227                                                 { $output .= '</i>'; $state = 'b'; }
1228                                                 else if ($state === 'ib')
1229                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1230                                                 else if ($state === 'both')
1231                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1232                                                 else # $state can be 'b' or ''
1233                                                 { $output .= '<i>'; $state .= 'i'; }
1234                                         }
1235                                         else if (strlen ($r) == 3)
1236                                         {
1237                                                 if ($state === 'b')
1238                                                 { $output .= '</b>'; $state = ''; }
1239                                                 else if ($state === 'bi')
1240                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1241                                                 else if ($state === 'ib')
1242                                                 { $output .= '</b>'; $state = 'i'; }
1243                                                 else if ($state === 'both')
1244                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1245                                                 else # $state can be 'i' or ''
1246                                                 { $output .= '<b>'; $state .= 'b'; }
1247                                         }
1248                                         else if (strlen ($r) == 5)
1249                                         {
1250                                                 if ($state === 'b')
1251                                                 { $output .= '</b><i>'; $state = 'i'; }
1252                                                 else if ($state === 'i')
1253                                                 { $output .= '</i><b>'; $state = 'b'; }
1254                                                 else if ($state === 'bi')
1255                                                 { $output .= '</i></b>'; $state = ''; }
1256                                                 else if ($state === 'ib')
1257                                                 { $output .= '</b></i>'; $state = ''; }
1258                                                 else if ($state === 'both')
1259                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1260                                                 else # ($state == '')
1261                                                 { $buffer = ''; $state = 'both'; }
1262                                         }
1263                                 }
1264                                 $i++;
1265                         }
1266                         # Now close all remaining tags.  Notice that the order is important.
1267                         if ($state === 'b' || $state === 'ib')
1268                                 $output .= '</b>';
1269                         if ($state === 'i' || $state === 'bi' || $state === 'ib')
1270                                 $output .= '</i>';
1271                         if ($state === 'bi')
1272                                 $output .= '</b>';
1273                         # There might be lonely ''''', so make sure we have a buffer
1274                         if ($state === 'both' && $buffer)
1275                                 $output .= '<b><i>'.$buffer.'</i></b>';
1276                         return $output;
1277                 }
1278         }
1279
1280         /**
1281          * Replace external links (REL)
1282          *
1283          * Note: this is all very hackish and the order of execution matters a lot.
1284          * Make sure to run maintenance/parserTests.php if you change this code.
1285          *
1286          * @private
1287          */
1288         function replaceExternalLinks( $text ) {
1289                 global $wgContLang;
1290                 wfProfileIn( __METHOD__ );
1291
1292                 $sk = $this->mOptions->getSkin();
1293
1294                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1295                 $s = array_shift( $bits );
1296
1297                 $i = 0;
1298                 while ( $i<count( $bits ) ) {
1299                         $url = $bits[$i++];
1300                         $protocol = $bits[$i++];
1301                         $text = $bits[$i++];
1302                         $trail = $bits[$i++];
1303
1304                         # The characters '<' and '>' (which were escaped by
1305                         # removeHTMLtags()) should not be included in
1306                         # URLs, per RFC 2396.
1307                         $m2 = array();
1308                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1309                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1310                                 $url = substr($url, 0, $m2[0][1]);
1311                         }
1312
1313                         # If the link text is an image URL, replace it with an <img> tag
1314                         # This happened by accident in the original parser, but some people used it extensively
1315                         $img = $this->maybeMakeExternalImage( $text );
1316                         if ( $img !== false ) {
1317                                 $text = $img;
1318                         }
1319
1320                         $dtrail = '';
1321
1322                         # Set linktype for CSS - if URL==text, link is essentially free
1323                         $linktype = ($text === $url) ? 'free' : 'text';
1324
1325                         # No link text, e.g. [http://domain.tld/some.link]
1326                         if ( $text == '' ) {
1327                                 # Autonumber if allowed. See bug #5918
1328                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1329                                         $langObj = $this->getFunctionLang();
1330                                         $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1331                                         $linktype = 'autonumber';
1332                                 } else {
1333                                         # Otherwise just use the URL
1334                                         $text = htmlspecialchars( $url );
1335                                         $linktype = 'free';
1336                                 }
1337                         } else {
1338                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1339                                 # Check for trail
1340                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1341                         }
1342
1343                         $text = $wgContLang->markNoConversion($text);
1344
1345                         $url = Sanitizer::cleanUrl( $url );
1346
1347                         # Use the encoded URL
1348                         # This means that users can paste URLs directly into the text
1349                         # Funny characters like &ouml; aren't valid in URLs anyway
1350                         # This was changed in August 2004
1351                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype,
1352                                 $this->getExternalLinkAttribs( $url ) ) . $dtrail . $trail;
1353
1354                         # Register link in the output object.
1355                         # Replace unnecessary URL escape codes with the referenced character
1356                         # This prevents spammers from hiding links from the filters
1357                         $pasteurized = self::replaceUnusualEscapes( $url );
1358                         $this->mOutput->addExternalLink( $pasteurized );
1359                 }
1360
1361                 wfProfileOut( __METHOD__ );
1362                 return $s;
1363         }
1364
1365         /**
1366          * Get an associative array of additional HTML attributes appropriate for a
1367          * particular external link.  This currently may include rel => nofollow
1368          * (depending on configuration, namespace, and the URL's domain) and/or a
1369          * target attribute (depending on configuration).
1370          *
1371          * @param string $url Optional URL, to extract the domain from for rel =>
1372          *   nofollow if appropriate
1373          * @return array Associative array of HTML attributes
1374          */
1375         function getExternalLinkAttribs( $url = false ) {
1376                 $attribs = array();
1377                 global $wgNoFollowLinks, $wgNoFollowNsExceptions;
1378                 $ns = $this->mTitle->getNamespace();
1379                 if( $wgNoFollowLinks && !in_array($ns, $wgNoFollowNsExceptions) ) {
1380                         $attribs['rel'] = 'nofollow';
1381
1382                         global $wgNoFollowDomainExceptions;
1383                         if ( $wgNoFollowDomainExceptions ) {
1384                                 $bits = wfParseUrl( $url );
1385                                 if ( is_array( $bits ) && isset( $bits['host'] ) ) {
1386                                         foreach ( $wgNoFollowDomainExceptions as $domain ) {
1387                                                 if( substr( $bits['host'], -strlen( $domain ) )
1388                                                 == $domain ) {
1389                                                         unset( $attribs['rel'] );
1390                                                         break;
1391                                                 }
1392                                         }
1393                                 }
1394                         }
1395                 }
1396                 if ( $this->mOptions->getExternalLinkTarget() ) {
1397                         $attribs['target'] = $this->mOptions->getExternalLinkTarget();
1398                 }
1399                 return $attribs;
1400         }
1401
1402
1403         /**
1404          * Replace unusual URL escape codes with their equivalent characters
1405          * @param string
1406          * @return string
1407          * @static
1408          * @todo  This can merge genuinely required bits in the path or query string,
1409          *        breaking legit URLs. A proper fix would treat the various parts of
1410          *        the URL differently; as a workaround, just use the output for
1411          *        statistical records, not for actual linking/output.
1412          */
1413         static function replaceUnusualEscapes( $url ) {
1414                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1415                         array( __CLASS__, 'replaceUnusualEscapesCallback' ), $url );
1416         }
1417
1418         /**
1419          * Callback function used in replaceUnusualEscapes().
1420          * Replaces unusual URL escape codes with their equivalent character
1421          * @static
1422          * @private
1423          */
1424         private static function replaceUnusualEscapesCallback( $matches ) {
1425                 $char = urldecode( $matches[0] );
1426                 $ord = ord( $char );
1427                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1428                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1429                         // No, shouldn't be escaped
1430                         return $char;
1431                 } else {
1432                         // Yes, leave it escaped
1433                         return $matches[0];
1434                 }
1435         }
1436
1437         /**
1438          * make an image if it's allowed, either through the global
1439          * option, through the exception, or through the on-wiki whitelist
1440          * @private
1441          */
1442         function maybeMakeExternalImage( $url ) {
1443                 $sk = $this->mOptions->getSkin();
1444                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1445                 $imagesexception = !empty($imagesfrom);
1446                 $text = false;
1447                 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1448                 if( $imagesexception && is_array( $imagesfrom ) ) {
1449                         $imagematch = false;
1450                         foreach( $imagesfrom as $match ) {
1451                                 if( strpos( $url, $match ) === 0 ) {
1452                                         $imagematch = true;
1453                                         break;
1454                                 }
1455                         }
1456                 } elseif( $imagesexception ) {
1457                         $imagematch = (strpos( $url, $imagesfrom ) === 0);
1458                 } else {
1459                         $imagematch = false;
1460                 }
1461                 if ( $this->mOptions->getAllowExternalImages()
1462                      || ( $imagesexception && $imagematch ) ) {
1463                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1464                                 # Image found
1465                                 $text = $sk->makeExternalImage( $url );
1466                         }
1467                 }
1468                 if( !$text && $this->mOptions->getEnableImageWhitelist()
1469                          && preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1470                         $whitelist = explode( "\n", wfMsgForContent( 'external_image_whitelist' ) );
1471                         foreach( $whitelist as $entry ) {
1472                                 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1473                                 if( strpos( $entry, '#' ) === 0 || $entry === '' )
1474                                         continue;
1475                                 if( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1476                                         # Image matches a whitelist entry
1477                                         $text = $sk->makeExternalImage( $url );
1478                                         break;
1479                                 }
1480                         }
1481                 }
1482                 return $text;
1483         }
1484
1485         /**
1486          * Process [[ ]] wikilinks
1487          * @return processed text
1488          *
1489          * @private
1490          */
1491         function replaceInternalLinks( $s ) {
1492                 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
1493                 return $s;
1494         }
1495
1496         /**
1497          * Process [[ ]] wikilinks (RIL)
1498          * @return LinkHolderArray
1499          *
1500          * @private
1501          */
1502         function replaceInternalLinks2( &$s ) {
1503                 global $wgContLang;
1504
1505                 wfProfileIn( __METHOD__ );
1506
1507                 wfProfileIn( __METHOD__.'-setup' );
1508                 static $tc = FALSE, $e1, $e1_img;
1509                 # the % is needed to support urlencoded titles as well
1510                 if ( !$tc ) {
1511                         $tc = Title::legalChars() . '#%';
1512                         # Match a link having the form [[namespace:link|alternate]]trail
1513                         $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
1514                         # Match cases where there is no "]]", which might still be images
1515                         $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
1516                 }
1517
1518                 $sk = $this->mOptions->getSkin();
1519                 $holders = new LinkHolderArray( $this );
1520
1521                 #split the entire text string on occurences of [[
1522                 $a = StringUtils::explode( '[[', ' ' . $s );
1523                 #get the first element (all text up to first [[), and remove the space we added
1524                 $s = $a->current();
1525                 $a->next();
1526                 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
1527                 $s = substr( $s, 1 );
1528
1529                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1530                 $e2 = null;
1531                 if ( $useLinkPrefixExtension ) {
1532                         # Match the end of a line for a word that's not followed by whitespace,
1533                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1534                         $e2 = wfMsgForContent( 'linkprefix' );
1535                 }
1536
1537                 if( is_null( $this->mTitle ) ) {
1538                         wfProfileOut( __METHOD__.'-setup' );
1539                         wfProfileOut( __METHOD__ );
1540                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1541                 }
1542                 $nottalk = !$this->mTitle->isTalkPage();
1543
1544                 if ( $useLinkPrefixExtension ) {
1545                         $m = array();
1546                         if ( preg_match( $e2, $s, $m ) ) {
1547                                 $first_prefix = $m[2];
1548                         } else {
1549                                 $first_prefix = false;
1550                         }
1551                 } else {
1552                         $prefix = '';
1553                 }
1554
1555                 if($wgContLang->hasVariants()) {
1556                         $selflink = $wgContLang->convertLinkToAllVariants($this->mTitle->getPrefixedText());
1557                 } else {
1558                         $selflink = array($this->mTitle->getPrefixedText());
1559                 }
1560                 $useSubpages = $this->areSubpagesAllowed();
1561                 wfProfileOut( __METHOD__.'-setup' );
1562
1563                 # Loop for each link
1564                 for ( ; $line !== false && $line !== null ; $a->next(), $line = $a->current() ) {
1565                         # Check for excessive memory usage
1566                         if ( $holders->isBig() ) {
1567                                 # Too big
1568                                 # Do the existence check, replace the link holders and clear the array
1569                                 $holders->replace( $s );
1570                                 $holders->clear();
1571                         }
1572
1573                         if ( $useLinkPrefixExtension ) {
1574                                 wfProfileIn( __METHOD__.'-prefixhandling' );
1575                                 if ( preg_match( $e2, $s, $m ) ) {
1576                                         $prefix = $m[2];
1577                                         $s = $m[1];
1578                                 } else {
1579                                         $prefix='';
1580                                 }
1581                                 # first link
1582                                 if($first_prefix) {
1583                                         $prefix = $first_prefix;
1584                                         $first_prefix = false;
1585                                 }
1586                                 wfProfileOut( __METHOD__.'-prefixhandling' );
1587                         }
1588
1589                         $might_be_img = false;
1590
1591                         wfProfileIn( __METHOD__."-e1" );
1592                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1593                                 $text = $m[2];
1594                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1595                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1596                                 # the real problem is with the $e1 regex
1597                                 # See bug 1300.
1598                                 #
1599                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1600                                 # and no image is in sight. See bug 2095.
1601                                 #
1602                                 if( $text !== '' &&
1603                                         substr( $m[3], 0, 1 ) === ']' &&
1604                                         strpos($text, '[') !== false
1605                                 )
1606                                 {
1607                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1608                                         $m[3] = substr( $m[3], 1 );
1609                                 }
1610                                 # fix up urlencoded title texts
1611                                 if( strpos( $m[1], '%' ) !== false ) {
1612                                         # Should anchors '#' also be rejected?
1613                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1614                                 }
1615                                 $trail = $m[3];
1616                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1617                                 $might_be_img = true;
1618                                 $text = $m[2];
1619                                 if ( strpos( $m[1], '%' ) !== false ) {
1620                                         $m[1] = urldecode($m[1]);
1621                                 }
1622                                 $trail = "";
1623                         } else { # Invalid form; output directly
1624                                 $s .= $prefix . '[[' . $line ;
1625                                 wfProfileOut( __METHOD__."-e1" );
1626                                 continue;
1627                         }
1628                         wfProfileOut( __METHOD__."-e1" );
1629                         wfProfileIn( __METHOD__."-misc" );
1630
1631                         # Don't allow internal links to pages containing
1632                         # PROTO: where PROTO is a valid URL protocol; these
1633                         # should be external links.
1634                         if ( preg_match( '/^\b(?:' . wfUrlProtocols() . ')/', $m[1] ) ) {
1635                                 $s .= $prefix . '[[' . $line ;
1636                                 wfProfileOut( __METHOD__."-misc" );
1637                                 continue;
1638                         }
1639
1640                         # Make subpage if necessary
1641                         if ( $useSubpages ) {
1642                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1643                         } else {
1644                                 $link = $m[1];
1645                         }
1646
1647                         $noforce = (substr( $m[1], 0, 1 ) !== ':');
1648                         if (!$noforce) {
1649                                 # Strip off leading ':'
1650                                 $link = substr( $link, 1 );
1651                         }
1652
1653                         wfProfileOut( __METHOD__."-misc" );
1654                         wfProfileIn( __METHOD__."-title" );
1655                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki( $link ) );
1656                         if ( $nt === null ) {
1657                                 $s .= $prefix . '[[' . $line;
1658                                 wfProfileOut( __METHOD__."-title" );
1659                                 continue;
1660                         }
1661
1662                         $ns = $nt->getNamespace();
1663                         $iw = $nt->getInterWiki();
1664                         wfProfileOut( __METHOD__."-title" );
1665
1666                         if ( $might_be_img ) { # if this is actually an invalid link
1667                                 wfProfileIn( __METHOD__."-might_be_img" );
1668                                 if ( $ns == NS_FILE && $noforce ) { #but might be an image
1669                                         $found = false;
1670                                         while ( true ) {
1671                                                 #look at the next 'line' to see if we can close it there
1672                                                 $a->next();
1673                                                 $next_line = $a->current();
1674                                                 if ( $next_line === false || $next_line === null ) {
1675                                                         break;
1676                                                 }
1677                                                 $m = explode( ']]', $next_line, 3 );
1678                                                 if ( count( $m ) == 3 ) {
1679                                                         # the first ]] closes the inner link, the second the image
1680                                                         $found = true;
1681                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1682                                                         $trail = $m[2];
1683                                                         break;
1684                                                 } elseif ( count( $m ) == 2 ) {
1685                                                         #if there's exactly one ]] that's fine, we'll keep looking
1686                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1687                                                 } else {
1688                                                         #if $next_line is invalid too, we need look no further
1689                                                         $text .= '[[' . $next_line;
1690                                                         break;
1691                                                 }
1692                                         }
1693                                         if ( !$found ) {
1694                                                 # we couldn't find the end of this imageLink, so output it raw
1695                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1696                                                 $holders->merge( $this->replaceInternalLinks2( $text ) );
1697                                                 $s .= "{$prefix}[[$link|$text";
1698                                                 # note: no $trail, because without an end, there *is* no trail
1699                                                 wfProfileOut( __METHOD__."-might_be_img" );
1700                                                 continue;
1701                                         }
1702                                 } else { #it's not an image, so output it raw
1703                                         $s .= "{$prefix}[[$link|$text";
1704                                         # note: no $trail, because without an end, there *is* no trail
1705                                         wfProfileOut( __METHOD__."-might_be_img" );
1706                                         continue;
1707                                 }
1708                                 wfProfileOut( __METHOD__."-might_be_img" );
1709                         }
1710
1711                         $wasblank = ( $text  == '' );
1712                         if ( $wasblank ) $text = $link;
1713
1714                         # Link not escaped by : , create the various objects
1715                         if ( $noforce ) {
1716
1717                                 # Interwikis
1718                                 wfProfileIn( __METHOD__."-interwiki" );
1719                                 if ( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1720                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1721                                         $s = rtrim($s . $prefix);
1722                                         $s .= trim($trail, "\n") == '' ? '': $prefix . $trail;
1723                                         wfProfileOut( __METHOD__."-interwiki" );
1724                                         continue;
1725                                 }
1726                                 wfProfileOut( __METHOD__."-interwiki" );
1727
1728                                 if ( $ns == NS_FILE ) {
1729                                         wfProfileIn( __METHOD__."-image" );
1730                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1731                                                 if ( $wasblank ) {
1732                                                         # if no parameters were passed, $text
1733                                                         # becomes something like "File:Foo.png",
1734                                                         # which we don't want to pass on to the
1735                                                         # image generator
1736                                                         $text = '';
1737                                                 } else {
1738                                                         # recursively parse links inside the image caption
1739                                                         # actually, this will parse them in any other parameters, too,
1740                                                         # but it might be hard to fix that, and it doesn't matter ATM
1741                                                         $text = $this->replaceExternalLinks($text);
1742                                                         $holders->merge( $this->replaceInternalLinks2( $text ) );
1743                                                 }
1744                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1745                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text, $holders ) ) . $trail;
1746                                         }
1747                                         $this->mOutput->addImage( $nt->getDBkey() );
1748                                         wfProfileOut( __METHOD__."-image" );
1749                                         continue;
1750
1751                                 }
1752
1753                                 if ( $ns == NS_CATEGORY ) {
1754                                         wfProfileIn( __METHOD__."-category" );
1755                                         $s = rtrim($s . "\n"); # bug 87
1756
1757                                         if ( $wasblank ) {
1758                                                 $sortkey = $this->getDefaultSort();
1759                                         } else {
1760                                                 $sortkey = $text;
1761                                         }
1762                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1763                                         $sortkey = str_replace( "\n", '', $sortkey );
1764                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1765                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1766
1767                                         /**
1768                                          * Strip the whitespace Category links produce, see bug 87
1769                                          * @todo We might want to use trim($tmp, "\n") here.
1770                                          */
1771                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1772
1773                                         wfProfileOut( __METHOD__."-category" );
1774                                         continue;
1775                                 }
1776                         }
1777
1778                         # Self-link checking
1779                         if( $nt->getFragment() === '' && $ns != NS_SPECIAL ) {
1780                                 if( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1781                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1782                                         continue;
1783                                 }
1784                         }
1785
1786                         # NS_MEDIA is a pseudo-namespace for linking directly to a file
1787                         # FIXME: Should do batch file existence checks, see comment below
1788                         if( $ns == NS_MEDIA ) {
1789                                 wfProfileIn( __METHOD__."-media" );
1790                                 # Give extensions a chance to select the file revision for us
1791                                 $skip = $time = false;
1792                                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$nt, &$skip, &$time ) );
1793                                 if ( $skip ) {
1794                                         $link = $sk->link( $nt );
1795                                 } else {
1796                                         $link = $sk->makeMediaLinkObj( $nt, $text, $time );
1797                                 }
1798                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1799                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1800                                 $this->mOutput->addImage( $nt->getDBkey() );
1801                                 wfProfileOut( __METHOD__."-media" );
1802                                 continue;
1803                         }
1804
1805                         wfProfileIn( __METHOD__."-always_known" );
1806                         # Some titles, such as valid special pages or files in foreign repos, should
1807                         # be shown as bluelinks even though they're not included in the page table
1808                         #
1809                         # FIXME: isAlwaysKnown() can be expensive for file links; we should really do
1810                         # batch file existence checks for NS_FILE and NS_MEDIA
1811                         if( $iw == '' && $nt->isAlwaysKnown() ) {
1812                                 $this->mOutput->addLink( $nt );
1813                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1814                         } else {
1815                                 # Links will be added to the output link list after checking
1816                                 $s .= $holders->makeHolder( $nt, $text, '', $trail, $prefix );
1817                         }
1818                         wfProfileOut( __METHOD__."-always_known" );
1819                 }
1820                 wfProfileOut( __METHOD__ );
1821                 return $holders;
1822         }
1823
1824         /**
1825          * Make a link placeholder. The text returned can be later resolved to a real link with
1826          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1827          * parsing of interwiki links, and secondly to allow all existence checks and
1828          * article length checks (for stub links) to be bundled into a single query.
1829          *
1830          * @deprecated
1831          */
1832         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1833                 return $this->mLinkHolders->makeHolder( $nt, $text, $query, $trail, $prefix );
1834         }
1835
1836         /**
1837          * Render a forced-blue link inline; protect against double expansion of
1838          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1839          * Since this little disaster has to split off the trail text to avoid
1840          * breaking URLs in the following text without breaking trails on the
1841          * wiki links, it's been made into a horrible function.
1842          *
1843          * @param Title $nt
1844          * @param string $text
1845          * @param string $query
1846          * @param string $trail
1847          * @param string $prefix
1848          * @return string HTML-wikitext mix oh yuck
1849          */
1850         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1851                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1852                 $sk = $this->mOptions->getSkin();
1853                 // FIXME: use link() instead of deprecated makeKnownLinkObj()
1854                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1855                 return $this->armorLinks( $link ) . $trail;
1856         }
1857
1858         /**
1859          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1860          * going to go through further parsing steps before inline URL expansion.
1861          *
1862          * Not needed quite as much as it used to be since free links are a bit
1863          * more sensible these days. But bracketed links are still an issue.
1864          *
1865          * @param string more-or-less HTML
1866          * @return string less-or-more HTML with NOPARSE bits
1867          */
1868         function armorLinks( $text ) {
1869                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1870                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1871         }
1872
1873         /**
1874          * Return true if subpage links should be expanded on this page.
1875          * @return bool
1876          */
1877         function areSubpagesAllowed() {
1878                 # Some namespaces don't allow subpages
1879                 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
1880         }
1881
1882         /**
1883          * Handle link to subpage if necessary
1884          * @param string $target the source of the link
1885          * @param string &$text the link text, modified as necessary
1886          * @return string the full name of the link
1887          * @private
1888          */
1889         function maybeDoSubpageLink($target, &$text) {
1890                 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
1891         }
1892
1893         /**#@+
1894          * Used by doBlockLevels()
1895          * @private
1896          */
1897         /* private */ function closeParagraph() {
1898                 $result = '';
1899                 if ( $this->mLastSection != '' ) {
1900                         $result = '</' . $this->mLastSection  . ">\n";
1901                 }
1902                 $this->mInPre = false;
1903                 $this->mLastSection = '';
1904                 return $result;
1905         }
1906         # getCommon() returns the length of the longest common substring
1907         # of both arguments, starting at the beginning of both.
1908         #
1909         /* private */ function getCommon( $st1, $st2 ) {
1910                 $fl = strlen( $st1 );
1911                 $shorter = strlen( $st2 );
1912                 if ( $fl < $shorter ) { $shorter = $fl; }
1913
1914                 for ( $i = 0; $i < $shorter; ++$i ) {
1915                         if ( $st1{$i} != $st2{$i} ) { break; }
1916                 }
1917                 return $i;
1918         }
1919         # These next three functions open, continue, and close the list
1920         # element appropriate to the prefix character passed into them.
1921         #
1922         /* private */ function openList( $char ) {
1923                 $result = $this->closeParagraph();
1924
1925                 if ( '*' === $char ) { $result .= '<ul><li>'; }
1926                 elseif ( '#' === $char ) { $result .= '<ol><li>'; }
1927                 elseif ( ':' === $char ) { $result .= '<dl><dd>'; }
1928                 elseif ( ';' === $char ) {
1929                         $result .= '<dl><dt>';
1930                         $this->mDTopen = true;
1931                 }
1932                 else { $result = '<!-- ERR 1 -->'; }
1933
1934                 return $result;
1935         }
1936
1937         /* private */ function nextItem( $char ) {
1938                 if ( '*' === $char || '#' === $char ) { return '</li><li>'; }
1939                 elseif ( ':' === $char || ';' === $char ) {
1940                         $close = '</dd>';
1941                         if ( $this->mDTopen ) { $close = '</dt>'; }
1942                         if ( ';' === $char ) {
1943                                 $this->mDTopen = true;
1944                                 return $close . '<dt>';
1945                         } else {
1946                                 $this->mDTopen = false;
1947                                 return $close . '<dd>';
1948                         }
1949                 }
1950                 return '<!-- ERR 2 -->';
1951         }
1952
1953         /* private */ function closeList( $char ) {
1954                 if ( '*' === $char ) { $text = '</li></ul>'; }
1955                 elseif ( '#' === $char ) { $text = '</li></ol>'; }
1956                 elseif ( ':' === $char ) {
1957                         if ( $this->mDTopen ) {
1958                                 $this->mDTopen = false;
1959                                 $text = '</dt></dl>';
1960                         } else {
1961                                 $text = '</dd></dl>';
1962                         }
1963                 }
1964                 else {  return '<!-- ERR 3 -->'; }
1965                 return $text."\n";
1966         }
1967         /**#@-*/
1968
1969         /**
1970          * Make lists from lines starting with ':', '*', '#', etc. (DBL)
1971          *
1972          * @param $linestart bool whether or not this is at the start of a line.
1973          * @private
1974          * @return string the lists rendered as HTML
1975          */
1976         function doBlockLevels( $text, $linestart ) {
1977                 wfProfileIn( __METHOD__ );
1978
1979                 # Parsing through the text line by line.  The main thing
1980                 # happening here is handling of block-level elements p, pre,
1981                 # and making lists from lines starting with * # : etc.
1982                 #
1983                 $textLines = StringUtils::explode( "\n", $text );
1984
1985                 $lastPrefix = $output = '';
1986                 $this->mDTopen = $inBlockElem = false;
1987                 $prefixLength = 0;
1988                 $paragraphStack = false;
1989
1990                 foreach ( $textLines as $oLine ) {
1991                         # Fix up $linestart
1992                         if ( !$linestart ) {
1993                                 $output .= $oLine;
1994                                 $linestart = true;
1995                                 continue;
1996                         }
1997                         // * = ul
1998                         // # = ol
1999                         // ; = dt
2000                         // : = dd
2001
2002                         $lastPrefixLength = strlen( $lastPrefix );
2003                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
2004                         $preOpenMatch = preg_match('/<pre/i', $oLine );
2005                         // If not in a <pre> element, scan for and figure out what prefixes are there.
2006                         if ( !$this->mInPre ) {
2007                                 # Multiple prefixes may abut each other for nested lists.
2008                                 $prefixLength = strspn( $oLine, '*#:;' );
2009                                 $prefix = substr( $oLine, 0, $prefixLength );
2010
2011                                 # eh?
2012                                 // ; and : are both from definition-lists, so they're equivalent
2013                                 //  for the purposes of determining whether or not we need to open/close
2014                                 //  elements.
2015                                 $prefix2 = str_replace( ';', ':', $prefix );
2016                                 $t = substr( $oLine, $prefixLength );
2017                                 $this->mInPre = (bool)$preOpenMatch;
2018                         } else {
2019                                 # Don't interpret any other prefixes in preformatted text
2020                                 $prefixLength = 0;
2021                                 $prefix = $prefix2 = '';
2022                                 $t = $oLine;
2023                         }
2024
2025                         # List generation
2026                         if( $prefixLength && $lastPrefix === $prefix2 ) {
2027                                 # Same as the last item, so no need to deal with nesting or opening stuff
2028                                 $output .= $this->nextItem( substr( $prefix, -1 ) );
2029                                 $paragraphStack = false;
2030
2031                                 if ( substr( $prefix, -1 ) === ';') {
2032                                         # The one nasty exception: definition lists work like this:
2033                                         # ; title : definition text
2034                                         # So we check for : in the remainder text to split up the
2035                                         # title and definition, without b0rking links.
2036                                         $term = $t2 = '';
2037                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2038                                                 $t = $t2;
2039                                                 $output .= $term . $this->nextItem( ':' );
2040                                         }
2041                                 }
2042                         } elseif( $prefixLength || $lastPrefixLength ) {
2043                                 // We need to open or close prefixes, or both.
2044
2045                                 # Either open or close a level...
2046                                 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2047                                 $paragraphStack = false;
2048
2049                                 // Close all the prefixes which aren't shared.
2050                                 while( $commonPrefixLength < $lastPrefixLength ) {
2051                                         $output .= $this->closeList( $lastPrefix[$lastPrefixLength-1] );
2052                                         --$lastPrefixLength;
2053                                 }
2054
2055                                 // Continue the current prefix if appropriate.
2056                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2057                                         $output .= $this->nextItem( $prefix[$commonPrefixLength-1] );
2058                                 }
2059
2060                                 // Open prefixes where appropriate.
2061                                 while ( $prefixLength > $commonPrefixLength ) {
2062                                         $char = substr( $prefix, $commonPrefixLength, 1 );
2063                                         $output .= $this->openList( $char );
2064
2065                                         if ( ';' === $char ) {
2066                                                 # FIXME: This is dupe of code above
2067                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2068                                                         $t = $t2;
2069                                                         $output .= $term . $this->nextItem( ':' );
2070                                                 }
2071                                         }
2072                                         ++$commonPrefixLength;
2073                                 }
2074                                 $lastPrefix = $prefix2;
2075                         }
2076
2077                         // If we have no prefixes, go to paragraph mode.
2078                         if( 0 == $prefixLength ) {
2079                                 wfProfileIn( __METHOD__."-paragraph" );
2080                                 # No prefix (not in list)--go to paragraph mode
2081                                 // XXX: use a stack for nestable elements like span, table and div
2082                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2083                                 $closematch = preg_match(
2084                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2085                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2086                                 if ( $openmatch or $closematch ) {
2087                                         $paragraphStack = false;
2088                                         # TODO bug 5718: paragraph closed
2089                                         $output .= $this->closeParagraph();
2090                                         if ( $preOpenMatch and !$preCloseMatch ) {
2091                                                 $this->mInPre = true;
2092                                         }
2093                                         if ( $closematch ) {
2094                                                 $inBlockElem = false;
2095                                         } else {
2096                                                 $inBlockElem = true;
2097                                         }
2098                                 } else if ( !$inBlockElem && !$this->mInPre ) {
2099                                         if ( ' ' == substr( $t, 0, 1 ) and ( $this->mLastSection === 'pre' or trim($t) != '' ) ) {
2100                                                 // pre
2101                                                 if ($this->mLastSection !== 'pre') {
2102                                                         $paragraphStack = false;
2103                                                         $output .= $this->closeParagraph().'<pre>';
2104                                                         $this->mLastSection = 'pre';
2105                                                 }
2106                                                 $t = substr( $t, 1 );
2107                                         } else {
2108                                                 // paragraph
2109                                                 if ( trim($t) == '' ) {
2110                                                         if ( $paragraphStack ) {
2111                                                                 $output .= $paragraphStack.'<br />';
2112                                                                 $paragraphStack = false;
2113                                                                 $this->mLastSection = 'p';
2114                                                         } else {
2115                                                                 if ($this->mLastSection !== 'p' ) {
2116                                                                         $output .= $this->closeParagraph();
2117                                                                         $this->mLastSection = '';
2118                                                                         $paragraphStack = '<p>';
2119                                                                 } else {
2120                                                                         $paragraphStack = '</p><p>';
2121                                                                 }
2122                                                         }
2123                                                 } else {
2124                                                         if ( $paragraphStack ) {
2125                                                                 $output .= $paragraphStack;
2126                                                                 $paragraphStack = false;
2127                                                                 $this->mLastSection = 'p';
2128                                                         } else if ($this->mLastSection !== 'p') {
2129                                                                 $output .= $this->closeParagraph().'<p>';
2130                                                                 $this->mLastSection = 'p';
2131                                                         }
2132                                                 }
2133                                         }
2134                                 }
2135                                 wfProfileOut( __METHOD__."-paragraph" );
2136                         }
2137                         // somewhere above we forget to get out of pre block (bug 785)
2138                         if($preCloseMatch && $this->mInPre) {
2139                                 $this->mInPre = false;
2140                         }
2141                         if ($paragraphStack === false) {
2142                                 $output .= $t."\n";
2143                         }
2144                 }
2145                 while ( $prefixLength ) {
2146                         $output .= $this->closeList( $prefix2[$prefixLength-1] );
2147                         --$prefixLength;
2148                 }
2149                 if ( $this->mLastSection != '' ) {
2150                         $output .= '</' . $this->mLastSection . '>';
2151                         $this->mLastSection = '';
2152                 }
2153
2154                 wfProfileOut( __METHOD__ );
2155                 return $output;
2156         }
2157
2158         /**
2159          * Split up a string on ':', ignoring any occurences inside tags
2160          * to prevent illegal overlapping.
2161          * @param string $str the string to split
2162          * @param string &$before set to everything before the ':'
2163          * @param string &$after set to everything after the ':'
2164          * return string the position of the ':', or false if none found
2165          */
2166         function findColonNoLinks($str, &$before, &$after) {
2167                 wfProfileIn( __METHOD__ );
2168
2169                 $pos = strpos( $str, ':' );
2170                 if( $pos === false ) {
2171                         // Nothing to find!
2172                         wfProfileOut( __METHOD__ );
2173                         return false;
2174                 }
2175
2176                 $lt = strpos( $str, '<' );
2177                 if( $lt === false || $lt > $pos ) {
2178                         // Easy; no tag nesting to worry about
2179                         $before = substr( $str, 0, $pos );
2180                         $after = substr( $str, $pos+1 );
2181                         wfProfileOut( __METHOD__ );
2182                         return $pos;
2183                 }
2184
2185                 // Ugly state machine to walk through avoiding tags.
2186                 $state = self::COLON_STATE_TEXT;
2187                 $stack = 0;
2188                 $len = strlen( $str );
2189                 for( $i = 0; $i < $len; $i++ ) {
2190                         $c = $str{$i};
2191
2192                         switch( $state ) {
2193                         // (Using the number is a performance hack for common cases)
2194                         case 0: // self::COLON_STATE_TEXT:
2195                                 switch( $c ) {
2196                                 case "<":
2197                                         // Could be either a <start> tag or an </end> tag
2198                                         $state = self::COLON_STATE_TAGSTART;
2199                                         break;
2200                                 case ":":
2201                                         if( $stack == 0 ) {
2202                                                 // We found it!
2203                                                 $before = substr( $str, 0, $i );
2204                                                 $after = substr( $str, $i + 1 );
2205                                                 wfProfileOut( __METHOD__ );
2206                                                 return $i;
2207                                         }
2208                                         // Embedded in a tag; don't break it.
2209                                         break;
2210                                 default:
2211                                         // Skip ahead looking for something interesting
2212                                         $colon = strpos( $str, ':', $i );
2213                                         if( $colon === false ) {
2214                                                 // Nothing else interesting
2215                                                 wfProfileOut( __METHOD__ );
2216                                                 return false;
2217                                         }
2218                                         $lt = strpos( $str, '<', $i );
2219                                         if( $stack === 0 ) {
2220                                                 if( $lt === false || $colon < $lt ) {
2221                                                         // We found it!
2222                                                         $before = substr( $str, 0, $colon );
2223                                                         $after = substr( $str, $colon + 1 );
2224                                                         wfProfileOut( __METHOD__ );
2225                                                         return $i;
2226                                                 }
2227                                         }
2228                                         if( $lt === false ) {
2229                                                 // Nothing else interesting to find; abort!
2230                                                 // We're nested, but there's no close tags left. Abort!
2231                                                 break 2;
2232                                         }
2233                                         // Skip ahead to next tag start
2234                                         $i = $lt;
2235                                         $state = self::COLON_STATE_TAGSTART;
2236                                 }
2237                                 break;
2238                         case 1: // self::COLON_STATE_TAG:
2239                                 // In a <tag>
2240                                 switch( $c ) {
2241                                 case ">":
2242                                         $stack++;
2243                                         $state = self::COLON_STATE_TEXT;
2244                                         break;
2245                                 case "/":
2246                                         // Slash may be followed by >?
2247                                         $state = self::COLON_STATE_TAGSLASH;
2248                                         break;
2249                                 default:
2250                                         // ignore
2251                                 }
2252                                 break;
2253                         case 2: // self::COLON_STATE_TAGSTART:
2254                                 switch( $c ) {
2255                                 case "/":
2256                                         $state = self::COLON_STATE_CLOSETAG;
2257                                         break;
2258                                 case "!":
2259                                         $state = self::COLON_STATE_COMMENT;
2260                                         break;
2261                                 case ">":
2262                                         // Illegal early close? This shouldn't happen D:
2263                                         $state = self::COLON_STATE_TEXT;
2264                                         break;
2265                                 default:
2266                                         $state = self::COLON_STATE_TAG;
2267                                 }
2268                                 break;
2269                         case 3: // self::COLON_STATE_CLOSETAG:
2270                                 // In a </tag>
2271                                 if( $c === ">" ) {
2272                                         $stack--;
2273                                         if( $stack < 0 ) {
2274                                                 wfDebug( __METHOD__.": Invalid input; too many close tags\n" );
2275                                                 wfProfileOut( __METHOD__ );
2276                                                 return false;
2277                                         }
2278                                         $state = self::COLON_STATE_TEXT;
2279                                 }
2280                                 break;
2281                         case self::COLON_STATE_TAGSLASH:
2282                                 if( $c === ">" ) {
2283                                         // Yes, a self-closed tag <blah/>
2284                                         $state = self::COLON_STATE_TEXT;
2285                                 } else {
2286                                         // Probably we're jumping the gun, and this is an attribute
2287                                         $state = self::COLON_STATE_TAG;
2288                                 }
2289                                 break;
2290                         case 5: // self::COLON_STATE_COMMENT:
2291                                 if( $c === "-" ) {
2292                                         $state = self::COLON_STATE_COMMENTDASH;
2293                                 }
2294                                 break;
2295                         case self::COLON_STATE_COMMENTDASH:
2296                                 if( $c === "-" ) {
2297                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2298                                 } else {
2299                                         $state = self::COLON_STATE_COMMENT;
2300                                 }
2301                                 break;
2302                         case self::COLON_STATE_COMMENTDASHDASH:
2303                                 if( $c === ">" ) {
2304                                         $state = self::COLON_STATE_TEXT;
2305                                 } else {
2306                                         $state = self::COLON_STATE_COMMENT;
2307                                 }
2308                                 break;
2309                         default:
2310                                 throw new MWException( "State machine error in " . __METHOD__ );
2311                         }
2312                 }
2313                 if( $stack > 0 ) {
2314                         wfDebug( __METHOD__.": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2315                         return false;
2316                 }
2317                 wfProfileOut( __METHOD__ );
2318                 return false;
2319         }
2320
2321         /**
2322          * Return value of a magic variable (like PAGENAME)
2323          *
2324          * @private
2325          */
2326         function getVariableValue( $index, $frame=false ) {
2327                 global $wgContLang, $wgSitename, $wgServer, $wgServerName;
2328                 global $wgScriptPath, $wgStylePath;
2329
2330                 /**
2331                  * Some of these require message or data lookups and can be
2332                  * expensive to check many times.
2333                  */
2334                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2335                         if ( isset( $this->mVarCache[$index] ) ) {
2336                                 return $this->mVarCache[$index];
2337                         }
2338                 }
2339
2340                 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2341                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2342
2343                 # Use the time zone
2344                 global $wgLocaltimezone;
2345                 if ( isset( $wgLocaltimezone ) ) {
2346                         $oldtz = date_default_timezone_get();
2347                         date_default_timezone_set( $wgLocaltimezone );
2348                 }
2349
2350                 $localTimestamp = date( 'YmdHis', $ts );
2351                 $localMonth = date( 'm', $ts );
2352                 $localMonth1 = date( 'n', $ts );
2353                 $localMonthName = date( 'n', $ts );
2354                 $localDay = date( 'j', $ts );
2355                 $localDay2 = date( 'd', $ts );
2356                 $localDayOfWeek = date( 'w', $ts );
2357                 $localWeek = date( 'W', $ts );
2358                 $localYear = date( 'Y', $ts );
2359                 $localHour = date( 'H', $ts );
2360                 if ( isset( $wgLocaltimezone ) ) {
2361                         date_default_timezone_set( $oldtz );
2362                 }
2363
2364                 switch ( $index ) {
2365                         case 'currentmonth':
2366                                 $value = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2367                                 break;
2368                         case 'currentmonth1':
2369                                 $value = $wgContLang->formatNum( gmdate( 'n', $ts ) );
2370                                 break;
2371                         case 'currentmonthname':
2372                                 $value = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2373                                 break;
2374                         case 'currentmonthnamegen':
2375                                 $value = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2376                                 break;
2377                         case 'currentmonthabbrev':
2378                                 $value = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2379                                 break;
2380                         case 'currentday':
2381                                 $value = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2382                                 break;
2383                         case 'currentday2':
2384                                 $value = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2385                                 break;
2386                         case 'localmonth':
2387                                 $value = $wgContLang->formatNum( $localMonth );
2388                                 break;
2389                         case 'localmonth1':
2390                                 $value = $wgContLang->formatNum( $localMonth1 );
2391                                 break;
2392                         case 'localmonthname':
2393                                 $value = $wgContLang->getMonthName( $localMonthName );
2394                                 break;
2395                         case 'localmonthnamegen':
2396                                 $value = $wgContLang->getMonthNameGen( $localMonthName );
2397                                 break;
2398                         case 'localmonthabbrev':
2399                                 $value = $wgContLang->getMonthAbbreviation( $localMonthName );
2400                                 break;
2401                         case 'localday':
2402                                 $value = $wgContLang->formatNum( $localDay );
2403                                 break;
2404                         case 'localday2':
2405                                 $value = $wgContLang->formatNum( $localDay2 );
2406                                 break;
2407                         case 'pagename':
2408                                 $value = wfEscapeWikiText( $this->mTitle->getText() );
2409                                 break;
2410                         case 'pagenamee':
2411                                 $value = $this->mTitle->getPartialURL();
2412                                 break;
2413                         case 'fullpagename':
2414                                 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2415                                 break;
2416                         case 'fullpagenamee':
2417                                 $value = $this->mTitle->getPrefixedURL();
2418                                 break;
2419                         case 'subpagename':
2420                                 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2421                                 break;
2422                         case 'subpagenamee':
2423                                 $value = $this->mTitle->getSubpageUrlForm();
2424                                 break;
2425                         case 'basepagename':
2426                                 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2427                                 break;
2428                         case 'basepagenamee':
2429                                 $value = wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2430                                 break;
2431                         case 'talkpagename':
2432                                 if( $this->mTitle->canTalk() ) {
2433                                         $talkPage = $this->mTitle->getTalkPage();
2434                                         $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2435                                 } else {
2436                                         $value = '';
2437                                 }
2438                                 break;
2439                         case 'talkpagenamee':
2440                                 if( $this->mTitle->canTalk() ) {
2441                                         $talkPage = $this->mTitle->getTalkPage();
2442                                         $value = $talkPage->getPrefixedUrl();
2443                                 } else {
2444                                         $value = '';
2445                                 }
2446                                 break;
2447                         case 'subjectpagename':
2448                                 $subjPage = $this->mTitle->getSubjectPage();
2449                                 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2450                                 break;
2451                         case 'subjectpagenamee':
2452                                 $subjPage = $this->mTitle->getSubjectPage();
2453                                 $value = $subjPage->getPrefixedUrl();
2454                                 break;
2455                         case 'revisionid':
2456                                 // Let the edit saving system know we should parse the page
2457                                 // *after* a revision ID has been assigned.
2458                                 $this->mOutput->setFlag( 'vary-revision' );
2459                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2460                                 $value = $this->mRevisionId;
2461                                 break;
2462                         case 'revisionday':
2463                                 // Let the edit saving system know we should parse the page
2464                                 // *after* a revision ID has been assigned. This is for null edits.
2465                                 $this->mOutput->setFlag( 'vary-revision' );
2466                                 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2467                                 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2468                                 break;
2469                         case 'revisionday2':
2470                                 // Let the edit saving system know we should parse the page
2471                                 // *after* a revision ID has been assigned. This is for null edits.
2472                                 $this->mOutput->setFlag( 'vary-revision' );
2473                                 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2474                                 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2475                                 break;
2476                         case 'revisionmonth':
2477                                 // Let the edit saving system know we should parse the page
2478                                 // *after* a revision ID has been assigned. This is for null edits.
2479                                 $this->mOutput->setFlag( 'vary-revision' );
2480                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2481                                 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2482                                 break;
2483                         case 'revisionyear':
2484                                 // Let the edit saving system know we should parse the page
2485                                 // *after* a revision ID has been assigned. This is for null edits.
2486                                 $this->mOutput->setFlag( 'vary-revision' );
2487                                 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2488                                 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2489                                 break;
2490                         case 'revisiontimestamp':
2491                                 // Let the edit saving system know we should parse the page
2492                                 // *after* a revision ID has been assigned. This is for null edits.
2493                                 $this->mOutput->setFlag( 'vary-revision' );
2494                                 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2495                                 $value = $this->getRevisionTimestamp();
2496                                 break;
2497                         case 'revisionuser':
2498                                 // Let the edit saving system know we should parse the page
2499                                 // *after* a revision ID has been assigned. This is for null edits.
2500                                 $this->mOutput->setFlag( 'vary-revision' );
2501                                 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-revision...\n" );
2502                                 $value = $this->getRevisionUser();
2503                                 break;
2504                         case 'namespace':
2505                                 $value = str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2506                                 break;
2507                         case 'namespacee':
2508                                 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2509                                 break;
2510                         case 'talkspace':
2511                                 $value = $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2512                                 break;
2513                         case 'talkspacee':
2514                                 $value = $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2515                                 break;
2516                         case 'subjectspace':
2517                                 $value = $this->mTitle->getSubjectNsText();
2518                                 break;
2519                         case 'subjectspacee':
2520                                 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2521                                 break;
2522                         case 'currentdayname':
2523                                 $value = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2524                                 break;
2525                         case 'currentyear':
2526                                 $value = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2527                                 break;
2528                         case 'currenttime':
2529                                 $value = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2530                                 break;
2531                         case 'currenthour':
2532                                 $value = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2533                                 break;
2534                         case 'currentweek':
2535                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2536                                 // int to remove the padding
2537                                 $value = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2538                                 break;
2539                         case 'currentdow':
2540                                 $value = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2541                                 break;
2542                         case 'localdayname':
2543                                 $value = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2544                                 break;
2545                         case 'localyear':
2546                                 $value = $wgContLang->formatNum( $localYear, true );
2547                                 break;
2548                         case 'localtime':
2549                                 $value = $wgContLang->time( $localTimestamp, false, false );
2550                                 break;
2551                         case 'localhour':
2552                                 $value = $wgContLang->formatNum( $localHour, true );
2553                                 break;
2554                         case 'localweek':
2555                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2556                                 // int to remove the padding
2557                                 $value = $wgContLang->formatNum( (int)$localWeek );
2558                                 break;
2559                         case 'localdow':
2560                                 $value = $wgContLang->formatNum( $localDayOfWeek );
2561                                 break;
2562                         case 'numberofarticles':
2563                                 $value = $wgContLang->formatNum( SiteStats::articles() );
2564                                 break;
2565                         case 'numberoffiles':
2566                                 $value = $wgContLang->formatNum( SiteStats::images() );
2567                                 break;
2568                         case 'numberofusers':
2569                                 $value = $wgContLang->formatNum( SiteStats::users() );
2570                                 break;
2571                         case 'numberofactiveusers':
2572                                 $value = $wgContLang->formatNum( SiteStats::activeUsers() );
2573                                 break;
2574                         case 'numberofpages':
2575                                 $value = $wgContLang->formatNum( SiteStats::pages() );
2576                                 break;
2577                         case 'numberofadmins':
2578                                 $value = $wgContLang->formatNum( SiteStats::numberingroup('sysop') );
2579                                 break;
2580                         case 'numberofedits':
2581                                 $value = $wgContLang->formatNum( SiteStats::edits() );
2582                                 break;
2583                         case 'numberofviews':
2584                                 $value = $wgContLang->formatNum( SiteStats::views() );
2585                                 break;
2586                         case 'currenttimestamp':
2587                                 $value = wfTimestamp( TS_MW, $ts );
2588                                 break;
2589                         case 'localtimestamp':
2590                                 $value = $localTimestamp;
2591                                 break;
2592                         case 'currentversion':
2593                                 $value = SpecialVersion::getVersion();
2594                                 break;
2595                         case 'sitename':
2596                                 return $wgSitename;
2597                         case 'server':
2598                                 return $wgServer;
2599                         case 'servername':
2600                                 return $wgServerName;
2601                         case 'scriptpath':
2602                                 return $wgScriptPath;
2603                         case 'stylepath':
2604                                 return $wgStylePath;
2605                         case 'directionmark':
2606                                 return $wgContLang->getDirMark();
2607                         case 'contentlanguage':
2608                                 global $wgContLanguageCode;
2609                                 return $wgContLanguageCode;
2610                         default:
2611                                 $ret = null;
2612                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret, &$frame ) ) )
2613                                         return $ret;
2614                                 else
2615                                         return null;
2616                 }
2617
2618                 if ( $index )
2619                         $this->mVarCache[$index] = $value;
2620
2621                 return $value;
2622         }
2623
2624         /**
2625          * initialise the magic variables (like CURRENTMONTHNAME)
2626          *
2627          * @private
2628          */
2629         function initialiseVariables() {
2630                 wfProfileIn( __METHOD__ );
2631                 $variableIDs = MagicWord::getVariableIDs();
2632
2633                 $this->mVariables = new MagicWordArray( $variableIDs );
2634                 wfProfileOut( __METHOD__ );
2635         }
2636
2637         /**
2638          * Preprocess some wikitext and return the document tree.
2639          * This is the ghost of replace_variables().
2640          *
2641          * @param string $text The text to parse
2642          * @param integer flags Bitwise combination of:
2643          *          self::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
2644          *                                     included. Default is to assume a direct page view.
2645          *
2646          * The generated DOM tree must depend only on the input text and the flags.
2647          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2648          *
2649          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2650          * change in the DOM tree for a given text, must be passed through the section identifier
2651          * in the section edit link and thus back to extractSections().
2652          *
2653          * The output of this function is currently only cached in process memory, but a persistent
2654          * cache may be implemented at a later date which takes further advantage of these strict
2655          * dependency requirements.
2656          *
2657          * @private
2658          */
2659         function preprocessToDom ( $text, $flags = 0 ) {
2660                 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2661                 return $dom;
2662         }
2663
2664         /*
2665          * Return a three-element array: leading whitespace, string contents, trailing whitespace
2666          */
2667         public static function splitWhitespace( $s ) {
2668                 $ltrimmed = ltrim( $s );
2669                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2670                 $trimmed = rtrim( $ltrimmed );
2671                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2672                 if ( $diff > 0 ) {
2673                         $w2 = substr( $ltrimmed, -$diff );
2674                 } else {
2675                         $w2 = '';
2676                 }
2677                 return array( $w1, $trimmed, $w2 );
2678         }
2679
2680         /**
2681          * Replace magic variables, templates, and template arguments
2682          * with the appropriate text. Templates are substituted recursively,
2683          * taking care to avoid infinite loops.
2684          *
2685          * Note that the substitution depends on value of $mOutputType:
2686          *  self::OT_WIKI: only {{subst:}} templates
2687          *  self::OT_PREPROCESS: templates but not extension tags
2688          *  self::OT_HTML: all templates and extension tags
2689          *
2690          * @param string $tex The text to transform
2691          * @param PPFrame $frame Object describing the arguments passed to the template.
2692          *        Arguments may also be provided as an associative array, as was the usual case before MW1.12.
2693          *        Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
2694          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2695          * @private
2696          */
2697         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2698                 # Is there any text? Also, Prevent too big inclusions!
2699                 if ( strlen( $text ) < 1 || strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2700                         return $text;
2701                 }
2702                 wfProfileIn( __METHOD__ );
2703
2704                 if ( $frame === false ) {
2705                         $frame = $this->getPreprocessor()->newFrame();
2706                 } elseif ( !( $frame instanceof PPFrame ) ) {
2707                         wfDebug( __METHOD__." called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
2708                         $frame = $this->getPreprocessor()->newCustomFrame($frame);
2709                 }
2710
2711                 $dom = $this->preprocessToDom( $text );
2712                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2713                 $text = $frame->expand( $dom, $flags );
2714
2715                 wfProfileOut( __METHOD__ );
2716                 return $text;
2717         }
2718
2719         /// Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2720         static function createAssocArgs( $args ) {
2721                 $assocArgs = array();
2722                 $index = 1;
2723                 foreach( $args as $arg ) {
2724                         $eqpos = strpos( $arg, '=' );
2725                         if ( $eqpos === false ) {
2726                                 $assocArgs[$index++] = $arg;
2727                         } else {
2728                                 $name = trim( substr( $arg, 0, $eqpos ) );
2729                                 $value = trim( substr( $arg, $eqpos+1 ) );
2730                                 if ( $value === false ) {
2731                                         $value = '';
2732                                 }
2733                                 if ( $name !== false ) {
2734                                         $assocArgs[$name] = $value;
2735                                 }
2736                         }
2737                 }
2738
2739                 return $assocArgs;
2740         }
2741
2742         /**
2743          * Warn the user when a parser limitation is reached
2744          * Will warn at most once the user per limitation type
2745          *
2746          * @param string $limitationType, should be one of:
2747          *   'expensive-parserfunction' (corresponding messages: 'expensive-parserfunction-warning', 'expensive-parserfunction-category')
2748          *   'post-expand-template-argument' (corresponding messages: 'post-expand-template-argument-warning', 'post-expand-template-argument-category')
2749          *   'post-expand-template-inclusion' (corresponding messages: 'post-expand-template-inclusion-warning', 'post-expand-template-inclusion-category')
2750          * @params int $current, $max When an explicit limit has been
2751          *       exceeded, provide the values (optional)
2752          */
2753         function limitationWarn( $limitationType, $current=null, $max=null) {
2754                 //does no harm if $current and $max are present but are unnecessary for the message
2755                 $warning = wfMsgExt( "$limitationType-warning", array( 'parsemag', 'escape' ), $current, $max );
2756                 $this->mOutput->addWarning( $warning );
2757                 $this->addTrackingCategory( "$limitationType-category" );
2758         }
2759
2760         /**
2761          * Return the text of a template, after recursively
2762          * replacing any variables or templates within the template.
2763          *
2764          * @param array $piece The parts of the template
2765          *  $piece['title']: the title, i.e. the part before the |
2766          *  $piece['parts']: the parameter array
2767          *  $piece['lineStart']: whether the brace was at the start of a line
2768          * @param PPFrame The current frame, contains template arguments
2769          * @return string the text of the template
2770          * @private
2771          */
2772         function braceSubstitution( $piece, $frame ) {
2773                 global $wgContLang, $wgNonincludableNamespaces;
2774                 wfProfileIn( __METHOD__ );
2775                 wfProfileIn( __METHOD__.'-setup' );
2776
2777                 # Flags
2778                 $found = false;             # $text has been filled
2779                 $nowiki = false;            # wiki markup in $text should be escaped
2780                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2781                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2782                 $isChildObj = false;        # $text is a DOM node needing expansion in a child frame
2783                 $isLocalObj = false;        # $text is a DOM node needing expansion in the current frame
2784
2785                 # Title object, where $text came from
2786                 $title = null;
2787
2788                 # $part1 is the bit before the first |, and must contain only title characters.
2789                 # Various prefixes will be stripped from it later.
2790                 $titleWithSpaces = $frame->expand( $piece['title'] );
2791                 $part1 = trim( $titleWithSpaces );
2792                 $titleText = false;
2793
2794                 # Original title text preserved for various purposes
2795                 $originalTitle = $part1;
2796
2797                 # $args is a list of argument nodes, starting from index 0, not including $part1
2798                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2799                 wfProfileOut( __METHOD__.'-setup' );
2800
2801                 # SUBST
2802                 wfProfileIn( __METHOD__.'-modifiers' );
2803                 if ( !$found ) {
2804                         $mwSubst = MagicWord::get( 'subst' );
2805                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor $this->ot['wiki'] ) {
2806                                 # One of two possibilities is true:
2807                                 # 1) Found SUBST but not in the PST phase
2808                                 # 2) Didn't find SUBST and in the PST phase
2809                                 # In either case, return without further processing
2810                                 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2811                                 $isLocalObj = true;
2812                                 $found = true;
2813                         }
2814                 }
2815
2816                 # Variables
2817                 if ( !$found && $args->getLength() == 0 ) {
2818                         $id = $this->mVariables->matchStartToEnd( $part1 );
2819                         if ( $id !== false ) {
2820                                 $text = $this->getVariableValue( $id, $frame );
2821                                 if (MagicWord::getCacheTTL($id)>-1)
2822                                         $this->mOutput->mContainsOldMagic = true;
2823                                 $found = true;
2824                         }
2825                 }
2826
2827                 # MSG, MSGNW and RAW
2828                 if ( !$found ) {
2829                         # Check for MSGNW:
2830                         $mwMsgnw = MagicWord::get( 'msgnw' );
2831                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2832                                 $nowiki = true;
2833                         } else {
2834                                 # Remove obsolete MSG:
2835                                 $mwMsg = MagicWord::get( 'msg' );
2836                                 $mwMsg->matchStartAndRemove( $part1 );
2837                         }
2838
2839                         # Check for RAW:
2840                         $mwRaw = MagicWord::get( 'raw' );
2841                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2842                                 $forceRawInterwiki = true;
2843                         }
2844                 }
2845                 wfProfileOut( __METHOD__.'-modifiers' );
2846
2847                 # Parser functions
2848                 if ( !$found ) {
2849                         wfProfileIn( __METHOD__ . '-pfunc' );
2850
2851                         $colonPos = strpos( $part1, ':' );
2852                         if ( $colonPos !== false ) {
2853                                 # Case sensitive functions
2854                                 $function = substr( $part1, 0, $colonPos );
2855                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
2856                                         $function = $this->mFunctionSynonyms[1][$function];
2857                                 } else {
2858                                         # Case insensitive functions
2859                                         $function = $wgContLang->lc( $function );
2860                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
2861                                                 $function = $this->mFunctionSynonyms[0][$function];
2862                                         } else {
2863                                                 $function = false;
2864                                         }
2865                                 }
2866                                 if ( $function ) {
2867                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
2868                                         $initialArgs = array( &$this );
2869                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
2870                                         if ( $flags & SFH_OBJECT_ARGS ) {
2871                                                 # Add a frame parameter, and pass the arguments as an array
2872                                                 $allArgs = $initialArgs;
2873                                                 $allArgs[] = $frame;
2874                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2875                                                         $funcArgs[] = $args->item( $i );
2876                                                 }
2877                                                 $allArgs[] = $funcArgs;
2878                                         } else {
2879                                                 # Convert arguments to plain text
2880                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2881                                                         $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
2882                                                 }
2883                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
2884                                         }
2885
2886                                         # Workaround for PHP bug 35229 and similar
2887                                         if ( !is_callable( $callback ) ) {
2888                                                 wfProfileOut( __METHOD__ . '-pfunc' );
2889                                                 wfProfileOut( __METHOD__ );
2890                                                 throw new MWException( "Tag hook for $function is not callable\n" );
2891                                         }
2892                                         $result = call_user_func_array( $callback, $allArgs );
2893                                         $found = true;
2894                                         $noparse = true;
2895                                         $preprocessFlags = 0;
2896
2897                                         if ( is_array( $result ) ) {
2898                                                 if ( isset( $result[0] ) ) {
2899                                                         $text = $result[0];
2900                                                         unset( $result[0] );
2901                                                 }
2902
2903                                                 // Extract flags into the local scope
2904                                                 // This allows callers to set flags such as nowiki, found, etc.
2905                                                 extract( $result );
2906                                         } else {
2907                                                 $text = $result;
2908                                         }
2909                                         if ( !$noparse ) {
2910                                                 $text = $this->preprocessToDom( $text, $preprocessFlags );
2911                                                 $isChildObj = true;
2912                                         }
2913                                 }
2914                         }
2915                         wfProfileOut( __METHOD__ . '-pfunc' );
2916                 }
2917
2918                 # Finish mangling title and then check for loops.
2919                 # Set $title to a Title object and $titleText to the PDBK
2920                 if ( !$found ) {
2921                         $ns = NS_TEMPLATE;
2922                         # Split the title into page and subpage
2923                         $subpage = '';
2924                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
2925                         if ($subpage !== '') {
2926                                 $ns = $this->mTitle->getNamespace();
2927                         }
2928                         $title = Title::newFromText( $part1, $ns );
2929                         if ( $title ) {
2930                                 $titleText = $title->getPrefixedText();
2931                                 # Check for language variants if the template is not found
2932                                 if($wgContLang->hasVariants() && $title->getArticleID() == 0){
2933                                         $wgContLang->findVariantLink( $part1, $title, true );
2934                                 }
2935                                 # Do recursion depth check
2936                                 $limit = $this->mOptions->getMaxTemplateDepth();
2937                                 if ( $frame->depth >= $limit ) {
2938                                         $found = true;
2939                                         $text = '<span class="error">' . wfMsgForContent( 'parser-template-recursion-depth-warning', $limit ) . '</span>';
2940                                 }
2941                         }
2942                 }
2943
2944                 # Load from database
2945                 if ( !$found && $title ) {
2946                         wfProfileIn( __METHOD__ . '-loadtpl' );
2947                         if ( !$title->isExternal() ) {
2948                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
2949                                         $text = SpecialPage::capturePath( $title );
2950                                         if ( is_string( $text ) ) {
2951                                                 $found = true;
2952                                                 $isHTML = true;
2953                                                 $this->disableCache();
2954                                         }
2955                                 } else if ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
2956                                         $found = false; //access denied
2957                                         wfDebug( __METHOD__.": template inclusion denied for " . $title->getPrefixedDBkey() );
2958                                 } else {
2959                                         list( $text, $title ) = $this->getTemplateDom( $title );
2960                                         if ( $text !== false ) {
2961                                                 $found = true;
2962                                                 $isChildObj = true;
2963                                         }
2964                                 }
2965
2966                                 # If the title is valid but undisplayable, make a link to it
2967                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
2968                                         $text = "[[:$titleText]]";
2969                                         $found = true;
2970                                 }
2971                         } elseif ( $title->isTrans() ) {
2972                                 // Interwiki transclusion
2973                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
2974                                         $text = $this->interwikiTransclude( $title, 'render' );
2975                                         $isHTML = true;
2976                                 } else {
2977                                         $text = $this->interwikiTransclude( $title, 'raw' );
2978                                         // Preprocess it like a template
2979                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
2980                                         $isChildObj = true;
2981                                 }
2982                                 $found = true;
2983                         }
2984
2985                         # Do infinite loop check
2986                         # This has to be done after redirect resolution to avoid infinite loops via redirects
2987                         if ( !$frame->loopCheck( $title ) ) {
2988                                 $found = true;
2989                                 $text = '<span class="error">' . wfMsgForContent( 'parser-template-loop-warning', $titleText ) . '</span>';
2990                                 wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
2991                         }
2992                         wfProfileOut( __METHOD__ . '-loadtpl' );
2993                 }
2994
2995                 # If we haven't found text to substitute by now, we're done
2996                 # Recover the source wikitext and return it
2997                 if ( !$found ) {
2998                         $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2999                         wfProfileOut( __METHOD__ );
3000                         return array( 'object' => $text );
3001                 }
3002
3003                 # Expand DOM-style return values in a child frame
3004                 if ( $isChildObj ) {
3005                         # Clean up argument array
3006                         $newFrame = $frame->newChild( $args, $title );
3007
3008                         if ( $nowiki ) {
3009                                 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3010                         } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3011                                 # Expansion is eligible for the empty-frame cache
3012                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
3013                                         $text = $this->mTplExpandCache[$titleText];
3014                                 } else {
3015                                         $text = $newFrame->expand( $text );
3016                                         $this->mTplExpandCache[$titleText] = $text;
3017                                 }
3018                         } else {
3019                                 # Uncached expansion
3020                                 $text = $newFrame->expand( $text );
3021                         }
3022                 }
3023                 if ( $isLocalObj && $nowiki ) {
3024                         $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3025                         $isLocalObj = false;
3026                 }
3027
3028                 # Replace raw HTML by a placeholder
3029                 # Add a blank line preceding, to prevent it from mucking up
3030                 # immediately preceding headings
3031                 if ( $isHTML ) {
3032                         $text = "\n\n" . $this->insertStripItem( $text );
3033                 }
3034                 # Escape nowiki-style return values
3035                 elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3036                         $text = wfEscapeWikiText( $text );
3037                 }
3038                 # Bug 529: if the template begins with a table or block-level
3039                 # element, it should be treated as beginning a new line.
3040                 # This behaviour is somewhat controversial.
3041                 elseif ( is_string( $text ) && !$piece['lineStart'] && preg_match('/^(?:{\\||:|;|#|\*)/', $text)) /*}*/{
3042                         $text = "\n" . $text;
3043                 }
3044
3045                 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3046                         # Error, oversize inclusion
3047                         $text = "[[$originalTitle]]" .
3048                                 $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3049                         $this->limitationWarn( 'post-expand-template-inclusion' );
3050                 }
3051
3052                 if ( $isLocalObj ) {
3053                         $ret = array( 'object' => $text );
3054                 } else {
3055                         $ret = array( 'text' => $text );
3056                 }
3057
3058                 wfProfileOut( __METHOD__ );
3059                 return $ret;
3060         }
3061
3062         /**
3063          * Get the semi-parsed DOM representation of a template with a given title,
3064          * and its redirect destination title. Cached.
3065          */
3066         function getTemplateDom( $title ) {
3067                 $cacheTitle = $title;
3068                 $titleText = $title->getPrefixedDBkey();
3069
3070                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3071                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3072                         $title = Title::makeTitle( $ns, $dbk );
3073                         $titleText = $title->getPrefixedDBkey();
3074                 }
3075                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3076                         return array( $this->mTplDomCache[$titleText], $title );
3077                 }
3078
3079                 // Cache miss, go to the database
3080                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3081
3082                 if ( $text === false ) {
3083                         $this->mTplDomCache[$titleText] = false;
3084                         return array( false, $title );
3085                 }
3086
3087                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3088                 $this->mTplDomCache[ $titleText ] = $dom;
3089
3090                 if (! $title->equals($cacheTitle)) {
3091                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3092                                 array( $title->getNamespace(),$cdb = $title->getDBkey() );
3093                 }
3094
3095                 return array( $dom, $title );
3096         }
3097
3098         /**
3099          * Fetch the unparsed text of a template and register a reference to it.
3100          */
3101         function fetchTemplateAndTitle( $title ) {
3102                 $templateCb = $this->mOptions->getTemplateCallback();
3103                 $stuff = call_user_func( $templateCb, $title, $this );
3104                 $text = $stuff['text'];
3105                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3106                 if ( isset( $stuff['deps'] ) ) {
3107                         foreach ( $stuff['deps'] as $dep ) {
3108                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3109                         }
3110                 }
3111                 return array($text,$finalTitle);
3112         }
3113
3114         function fetchTemplate( $title ) {
3115                 $rv = $this->fetchTemplateAndTitle($title);
3116                 return $rv[0];
3117         }
3118
3119         /**
3120          * Static function to get a template
3121          * Can be overridden via ParserOptions::setTemplateCallback().
3122          */
3123         static function statelessFetchTemplate( $title, $parser=false ) {
3124                 $text = $skip = false;
3125                 $finalTitle = $title;
3126                 $deps = array();
3127
3128                 // Loop to fetch the article, with up to 1 redirect
3129                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3130                         # Give extensions a chance to select the revision instead
3131                         $id = false; // Assume current
3132                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( $parser, &$title, &$skip, &$id ) );
3133
3134                         if( $skip ) {
3135                                 $text = false;
3136                                 $deps[] = array(
3137                                         'title' => $title,
3138                                         'page_id' => $title->getArticleID(),
3139                                         'rev_id' => null );
3140                                 break;
3141                         }
3142                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3143                         $rev_id = $rev ? $rev->getId() : 0;
3144                         // If there is no current revision, there is no page
3145                         if( $id === false && !$rev ) {
3146                                 $linkCache = LinkCache::singleton();
3147                                 $linkCache->addBadLinkObj( $title );
3148                         }
3149
3150                         $deps[] = array(
3151                                 'title' => $title,
3152                                 'page_id' => $title->getArticleID(),
3153                                 'rev_id' => $rev_id );
3154
3155                         if( $rev ) {
3156                                 $text = $rev->getText();
3157                         } elseif( $title->getNamespace() == NS_MEDIAWIKI ) {
3158                                 global $wgContLang;
3159                                 $message = $wgContLang->lcfirst( $title->getText() );
3160                                 $text = wfMsgForContentNoTrans( $message );
3161                                 if( wfEmptyMsg( $message, $text ) ) {
3162                                         $text = false;
3163                                         break;
3164                                 }
3165                         } else {
3166                                 break;
3167                         }
3168                         if ( $text === false ) {
3169                                 break;
3170                         }
3171                         // Redirect?
3172                         $finalTitle = $title;
3173                         $title = Title::newFromRedirect( $text );
3174                 }
3175                 return array(
3176                         'text' => $text,
3177                         'finalTitle' => $finalTitle,
3178                         'deps' => $deps );
3179         }
3180
3181         /**
3182          * Transclude an interwiki link.
3183          */
3184         function interwikiTransclude( $title, $action ) {
3185                 global $wgEnableScaryTranscluding;
3186
3187                 if (!$wgEnableScaryTranscluding)
3188                         return wfMsg('scarytranscludedisabled');
3189
3190                 $url = $title->getFullUrl( "action=$action" );
3191
3192                 if (strlen($url) > 255)
3193                         return wfMsg('scarytranscludetoolong');
3194                 return $this->fetchScaryTemplateMaybeFromCache($url);
3195         }
3196
3197         function fetchScaryTemplateMaybeFromCache($url) {
3198                 global $wgTranscludeCacheExpiry;
3199                 $dbr = wfGetDB(DB_SLAVE);
3200                 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
3201                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3202                                 array('tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ) );
3203                 if ($obj) {
3204                         return $obj->tc_contents;
3205                 }
3206
3207                 $text = Http::get($url);
3208                 if (!$text)
3209                         return wfMsg('scarytranscludefailed', $url);
3210
3211                 $dbw = wfGetDB(DB_MASTER);
3212                 $dbw->replace('transcache', array('tc_url'), array(
3213                         'tc_url' => $url,
3214                         'tc_time' => $dbw->timestamp( time() ),
3215                         'tc_contents' => $text));
3216                 return $text;
3217         }
3218
3219
3220         /**
3221          * Triple brace replacement -- used for template arguments
3222          * @private
3223          */
3224         function argSubstitution( $piece, $frame ) {
3225                 wfProfileIn( __METHOD__ );
3226
3227                 $error = false;
3228                 $parts = $piece['parts'];
3229                 $nameWithSpaces = $frame->expand( $piece['title'] );
3230                 $argName = trim( $nameWithSpaces );
3231                 $object = false;
3232                 $text = $frame->getArgument( $argName );
3233                 if (  $text === false && $parts->getLength() > 0
3234                   && (
3235                     $this->ot['html']
3236                     || $this->ot['pre']
3237                     || ( $this->ot['wiki'] && $frame->isTemplate() )
3238                   )
3239                 ) {
3240                         # No match in frame, use the supplied default
3241                         $object = $parts->item( 0 )->getChildren();
3242                 }
3243                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3244                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3245                         $this->limitationWarn( 'post-expand-template-argument' );
3246                 }
3247
3248                 if ( $text === false && $object === false ) {
3249                         # No match anywhere
3250                         $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3251                 }
3252                 if ( $error !== false ) {
3253                         $text .= $error;
3254                 }
3255                 if ( $object !== false ) {
3256                         $ret = array( 'object' => $object );
3257                 } else {
3258                         $ret = array( 'text' => $text );
3259                 }
3260
3261                 wfProfileOut( __METHOD__ );
3262                 return $ret;
3263         }
3264
3265         /**
3266          * Return the text to be used for a given extension tag.
3267          * This is the ghost of strip().
3268          *
3269          * @param array $params Associative array of parameters:
3270          *     name       PPNode for the tag name
3271          *     attr       PPNode for unparsed text where tag attributes are thought to be
3272          *     attributes Optional associative array of parsed attributes
3273          *     inner      Contents of extension element
3274          *     noClose    Original text did not have a close tag
3275          * @param PPFrame $frame
3276          */
3277         function extensionSubstitution( $params, $frame ) {
3278                 global $wgRawHtml, $wgContLang;
3279
3280                 $name = $frame->expand( $params['name'] );
3281                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3282                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3283
3284                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf('%08X', $this->mMarkerIndex++) . self::MARKER_SUFFIX;
3285
3286                 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower($name)] ) &&
3287                         ( $this->ot['html'] || $this->ot['pre'] );
3288                 if ( $this->ot['html'] || $isFunctionTag ) {
3289                         $name = strtolower( $name );
3290                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3291                         if ( isset( $params['attributes'] ) ) {
3292                                 $attributes = $attributes + $params['attributes'];
3293                         }
3294                         switch ( $name ) {
3295                                 case 'html':
3296                                         if( $wgRawHtml ) {
3297                                                 $output = $content;
3298                                                 break;
3299                                         } else {
3300                                                 throw new MWException( '<html> extension tag encountered unexpectedly' );
3301                                         }
3302                                 case 'nowiki':
3303                                         $content = strtr($content, array('-{' => '-&#123;', '}-' => '&#125;-'));
3304                                         $output = Xml::escapeTagsOnly( $content );
3305                                         break;
3306                                 case 'gallery':
3307                                         $output = $this->renderImageGallery( $content, $attributes );
3308                                         break;
3309                                 case 'a':
3310                                         $output = $this->renderHyperlink( $content, $attributes, $frame );
3311                                         break;
3312                                 case 'math':
3313                                         if ( $this->mOptions->getUseTeX() ) {
3314                                                 $output = $wgContLang->armourMath(
3315                                                         MathRenderer::renderMath( $content, $attributes ) );
3316                                                 break;
3317                                         }
3318                                         /* else let a tag hook handle it (bug 21222) */
3319                                 default:
3320                                         if( isset( $this->mTagHooks[$name] ) ) {
3321                                                 # Workaround for PHP bug 35229 and similar
3322                                                 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3323                                                         throw new MWException( "Tag hook for $name is not callable\n" );
3324                                                 }
3325                                                 $output = call_user_func_array( $this->mTagHooks[$name],
3326                                                         array( $content, $attributes, $this, $frame ) );
3327                                         } elseif( isset( $this->mFunctionTagHooks[$name] ) ) {
3328                                                 list( $callback, $flags ) = $this->mFunctionTagHooks[$name];
3329                                                 if( !is_callable( $callback ) )
3330                                                         throw new MWException( "Tag hook for $name is not callable\n" );
3331
3332                                                 $output = call_user_func_array( $callback,
3333                                                         array( &$this, $frame, $content, $attributes ) );
3334                                         } else {
3335                                                 $output = '<span class="error">Invalid tag extension name: ' .
3336                                                         htmlspecialchars( $name ) . '</span>';
3337                                         }
3338                         }
3339                 } else {
3340                         if ( is_null( $attrText ) ) {
3341                                 $attrText = '';
3342                         }
3343                         if ( isset( $params['attributes'] ) ) {
3344                                 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3345                                         $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3346                                                 htmlspecialchars( $attrValue ) . '"';
3347                                 }
3348                         }
3349                         if ( $content === null ) {
3350                                 $output = "<$name$attrText/>";
3351                         } else {
3352                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3353                                 $output = "<$name$attrText>$content$close";
3354                         }
3355                 }
3356
3357                 if( $isFunctionTag ) {
3358                         return $output;
3359                 } elseif ( $name === 'html' || $name === 'nowiki' ) {
3360                         $this->mStripState->nowiki->setPair( $marker, $output );
3361                 } else {
3362                         $this->mStripState->general->setPair( $marker, $output );
3363                 }
3364                 return $marker;
3365         }
3366
3367         /**
3368          * Increment an include size counter
3369          *
3370          * @param string $type The type of expansion
3371          * @param integer $size The size of the text
3372          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3373          */
3374         function incrementIncludeSize( $type, $size ) {
3375                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize( $type ) ) {
3376                         return false;
3377                 } else {
3378                         $this->mIncludeSizes[$type] += $size;
3379                         return true;
3380                 }
3381         }
3382
3383         /**
3384          * Increment the expensive function count
3385          *
3386          * @return boolean False if the limit has been exceeded
3387          */
3388         function incrementExpensiveFunctionCount() {
3389                 global $wgExpensiveParserFunctionLimit;
3390                 $this->mExpensiveFunctionCount++;
3391                 if($this->mExpensiveFunctionCount <= $wgExpensiveParserFunctionLimit) {
3392                         return true;
3393                 }
3394                 return false;
3395         }
3396
3397         /**
3398          * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3399          * Fills $this->mDoubleUnderscores, returns the modified text
3400          */
3401         function doDoubleUnderscore( $text ) {
3402                 global $wgDisableLangConversion;
3403                 wfProfileIn( __METHOD__ );
3404
3405                 // The position of __TOC__ needs to be recorded
3406                 $mw = MagicWord::get( 'toc' );
3407                 if( $mw->match( $text ) ) {
3408                         $this->mShowToc = true;
3409                         $this->mForceTocPosition = true;
3410
3411                         // Set a placeholder. At the end we'll fill it in with the TOC.
3412                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3413
3414                         // Only keep the first one.
3415                         $text = $mw->replace( '', $text );
3416                 }
3417
3418                 // Now match and remove the rest of them
3419                 $mwa = MagicWord::getDoubleUnderscoreArray();
3420                 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3421
3422                 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3423                         $this->mOutput->mNoGallery = true;
3424                 }
3425                 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3426                         $this->mShowToc = false;
3427                 }
3428                 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
3429                         $this->mOutput->setProperty( 'hiddencat', 'y' );
3430                         $this->addTrackingCategory( 'hidden-category-category' );
3431                 }
3432                 # (bug 8068) Allow control over whether robots index a page.
3433                 #
3434                 # FIXME (bug 14899): __INDEX__ always overrides __NOINDEX__ here!  This
3435                 # is not desirable, the last one on the page should win.
3436                 if( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
3437                         $this->mOutput->setIndexPolicy( 'noindex' );
3438                         $this->addTrackingCategory( 'noindex-category' );
3439                 }
3440                 if( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ){
3441                         $this->mOutput->setIndexPolicy( 'index' );
3442                         $this->addTrackingCategory( 'index-category' );
3443                 }
3444
3445                 if ( !$wgDisableLangConversion ) {
3446                         if( isset( $this->mDoubleUnderscores['notitleconvert'] ) ){
3447                                 $this->mDoTitleConvert = false;
3448                         }
3449
3450                         // Don't convert talk pages
3451                         if( isset( $this->mDoubleUnderscores['nocontentconvert'] )
3452                                 && !$this->mTitle->isTalkPage() ){
3453                                 $this->mDoContentConvert = false;
3454                         }
3455                 }
3456
3457                 wfProfileOut( __METHOD__ );
3458                 return $text;
3459         }
3460
3461         /**
3462          * Add a tracking category, getting the title from a system message,
3463          * or print a debug message if the title is invalid.
3464          * @param $msg String message key
3465          * @return Bool whether the addition was successful
3466          */
3467         protected function addTrackingCategory( $msg ){
3468                 $cat = wfMsgForContent( $msg );
3469
3470                 # Allow tracking categories to be disabled by setting them to "-"
3471                 if( $cat === '-' ) return false;
3472
3473                 $containerCategory = Title::makeTitleSafe( NS_CATEGORY, $cat );
3474                 if ( $containerCategory ) {
3475                         $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3476                         return true;
3477                 } else {
3478                         wfDebug( __METHOD__.": [[MediaWiki:$msg]] is not a valid title!\n" );
3479                         return false;
3480                 }
3481         }
3482
3483         /**
3484          * This function accomplishes several tasks:
3485          * 1) Auto-number headings if that option is enabled
3486          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3487          * 3) Add a Table of contents on the top for users who have enabled the option
3488          * 4) Auto-anchor headings
3489          *
3490          * It loops through all headlines, collects the necessary data, then splits up the
3491          * string and re-inserts the newly formatted headlines.
3492          *
3493          * @param string $text
3494          * @param string $origText Original, untouched wikitext
3495          * @param boolean $isMain
3496          * @private
3497          */
3498         function formatHeadings( $text, $origText, $isMain=true ) {
3499                 global $wgMaxTocLevel, $wgContLang, $wgEnforceHtmlIds;
3500
3501                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3502                 $showEditLink = $this->mOptions->getEditSection();
3503
3504                 // Do not call quickUserCan unless necessary
3505                 if( $showEditLink && !$this->mTitle->quickUserCan( 'edit' ) ) {
3506                         $showEditLink = 0;
3507                 }
3508
3509                 # Inhibit editsection links if requested in the page
3510                 if ( isset( $this->mDoubleUnderscores['noeditsection'] )  || $this->mOptions->getIsPrintable() ) {
3511                         $showEditLink = 0;
3512                 }
3513
3514                 # Get all headlines for numbering them and adding funky stuff like [edit]
3515                 # links - this is for later, but we need the number of headlines right now
3516                 $matches = array();
3517                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3518
3519                 # if there are fewer than 4 headlines in the article, do not show TOC
3520                 # unless it's been explicitly enabled.
3521                 $enoughToc = $this->mShowToc &&
3522                         (($numMatches >= 4) || $this->mForceTocPosition);
3523
3524                 # Allow user to stipulate that a page should have a "new section"
3525                 # link added via __NEWSECTIONLINK__
3526                 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
3527                         $this->mOutput->setNewSection( true );
3528                 }
3529
3530                 # Allow user to remove the "new section"
3531                 # link via __NONEWSECTIONLINK__
3532                 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
3533                         $this->mOutput->hideNewSection( true );
3534                 }
3535
3536                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3537                 # override above conditions and always show TOC above first header
3538                 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
3539                         $this->mShowToc = true;
3540                         $enoughToc = true;
3541                 }
3542
3543                 # We need this to perform operations on the HTML
3544                 $sk = $this->mOptions->getSkin();
3545
3546                 # headline counter
3547                 $headlineCount = 0;
3548                 $numVisible = 0;
3549
3550                 # Ugh .. the TOC should have neat indentation levels which can be
3551                 # passed to the skin functions. These are determined here
3552                 $toc = '';
3553                 $full = '';
3554                 $head = array();
3555                 $sublevelCount = array();
3556                 $levelCount = array();
3557                 $toclevel = 0;
3558                 $level = 0;
3559                 $prevlevel = 0;
3560                 $toclevel = 0;
3561                 $prevtoclevel = 0;
3562                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
3563                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3564                 $oldType = $this->mOutputType;
3565                 $this->setOutputType( self::OT_WIKI );
3566                 $frame = $this->getPreprocessor()->newFrame();
3567                 $root = $this->preprocessToDom( $origText );
3568                 $node = $root->getFirstChild();
3569                 $byteOffset = 0;
3570                 $tocraw = array();
3571
3572                 foreach( $matches[3] as $headline ) {
3573                         $isTemplate = false;
3574                         $titleText = false;
3575                         $sectionIndex = false;
3576                         $numbering = '';
3577                         $markerMatches = array();
3578                         if (preg_match("/^$markerRegex/", $headline, $markerMatches)) {
3579                                 $serial = $markerMatches[1];
3580                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3581                                 $isTemplate = ($titleText != $baseTitleText);
3582                                 $headline = preg_replace("/^$markerRegex/", "", $headline);
3583                         }
3584
3585                         if( $toclevel ) {
3586                                 $prevlevel = $level;
3587                                 $prevtoclevel = $toclevel;
3588                         }
3589                         $level = $matches[1][$headlineCount];
3590
3591                         if ( $level > $prevlevel ) {
3592                                 # Increase TOC level
3593                                 $toclevel++;
3594                                 $sublevelCount[$toclevel] = 0;
3595                                 if( $toclevel<$wgMaxTocLevel ) {
3596                                         $prevtoclevel = $toclevel;
3597                                         $toc .= $sk->tocIndent();
3598                                         $numVisible++;
3599                                 }
3600                         }
3601                         elseif ( $level < $prevlevel && $toclevel > 1 ) {
3602                                 # Decrease TOC level, find level to jump to
3603
3604                                 for ($i = $toclevel; $i > 0; $i--) {
3605                                         if ( $levelCount[$i] == $level ) {
3606                                                 # Found last matching level
3607                                                 $toclevel = $i;
3608                                                 break;
3609                                         }
3610                                         elseif ( $levelCount[$i] < $level ) {
3611                                                 # Found first matching level below current level
3612                                                 $toclevel = $i + 1;
3613                                                 break;
3614                                         }
3615                                 }
3616                                 if( $i == 0 ) $toclevel = 1;
3617                                 if( $toclevel<$wgMaxTocLevel ) {
3618                                         if($prevtoclevel < $wgMaxTocLevel) {
3619                                                 # Unindent only if the previous toc level was shown :p
3620                                                 $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3621                                                 $prevtoclevel = $toclevel;
3622                                         } else {
3623                                                 $toc .= $sk->tocLineEnd();
3624                                         }
3625                                 }
3626                         }
3627                         else {
3628                                 # No change in level, end TOC line
3629                                 if( $toclevel<$wgMaxTocLevel ) {
3630                                         $toc .= $sk->tocLineEnd();
3631                                 }
3632                         }
3633
3634                         $levelCount[$toclevel] = $level;
3635
3636                         # count number of headlines for each level
3637                         @$sublevelCount[$toclevel]++;
3638                         $dot = 0;
3639                         for( $i = 1; $i <= $toclevel; $i++ ) {
3640                                 if( !empty( $sublevelCount[$i] ) ) {
3641                                         if( $dot ) {
3642                                                 $numbering .= '.';
3643                                         }
3644                                         $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3645                                         $dot = 1;
3646                                 }
3647                         }
3648
3649                         # The safe header is a version of the header text safe to use for links
3650                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3651                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
3652
3653                         # Remove link placeholders by the link text.
3654                         #     <!--LINK number-->
3655                         # turns into
3656                         #     link text with suffix
3657                         $safeHeadline = $this->replaceLinkHoldersText( $safeHeadline );
3658
3659                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3660                         $tocline = preg_replace(
3661                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3662                                 array( '',                          '<$1>'),
3663                                 $safeHeadline
3664                         );
3665                         $tocline = trim( $tocline );
3666
3667                         # For the anchor, strip out HTML-y stuff period
3668                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
3669                         $safeHeadline = preg_replace( '/[ _]+/', ' ', $safeHeadline );
3670                         $safeHeadline = trim( $safeHeadline );
3671
3672                         # Save headline for section edit hint before it's escaped
3673                         $headlineHint = $safeHeadline;
3674
3675                         if ( $wgEnforceHtmlIds ) {
3676                                 $legacyHeadline = false;
3677                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
3678                                         'noninitial' );
3679                         } else {
3680                                 # For reverse compatibility, provide an id that's
3681                                 # HTML4-compatible, like we used to.
3682                                 #
3683                                 # It may be worth noting, academically, that it's possible for
3684                                 # the legacy anchor to conflict with a non-legacy headline
3685                                 # anchor on the page.  In this case likely the "correct" thing
3686                                 # would be to either drop the legacy anchors or make sure
3687                                 # they're numbered first.  However, this would require people
3688                                 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
3689                                 # manually, so let's not bother worrying about it.
3690                                 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
3691                                         'noninitial' );
3692                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline, 'xml' );
3693
3694                                 if ( $legacyHeadline == $safeHeadline ) {
3695                                         # No reason to have both (in fact, we can't)
3696                                         $legacyHeadline = false;
3697                                 } elseif ( $legacyHeadline != Sanitizer::escapeId(
3698                                 $legacyHeadline, 'xml' ) ) {
3699                                         # The legacy id is invalid XML.  We used to allow this, but
3700                                         # there's no reason to do so anymore.  Backward
3701                                         # compatibility will fail slightly in this case, but it's
3702                                         # no big deal.
3703                                         $legacyHeadline = false;
3704                                 }
3705                         }
3706
3707                         # HTML names must be case-insensitively unique (bug 10721).  FIXME:
3708                         # Does this apply to Unicode characters?  Because we aren't
3709                         # handling those here.
3710                         $arrayKey = strtolower( $safeHeadline );
3711                         if ( $legacyHeadline === false ) {
3712                                 $legacyArrayKey = false;
3713                         } else {
3714                                 $legacyArrayKey = strtolower( $legacyHeadline );
3715                         }
3716
3717                         # count how many in assoc. array so we can track dupes in anchors
3718                         if ( isset( $refers[$arrayKey] ) ) {
3719                                 $refers[$arrayKey]++;
3720                         } else {
3721                                 $refers[$arrayKey] = 1;
3722                         }
3723                         if ( isset( $refers[$legacyArrayKey] ) ) {
3724                                 $refers[$legacyArrayKey]++;
3725                         } else {
3726                                 $refers[$legacyArrayKey] = 1;
3727                         }
3728
3729                         # Don't number the heading if it is the only one (looks silly)
3730                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3731                                 # the two are different if the line contains a link
3732                                 $headline=$numbering . ' ' . $headline;
3733                         }
3734
3735                         # Create the anchor for linking from the TOC to the section
3736                         $anchor = $safeHeadline;
3737                         $legacyAnchor = $legacyHeadline;
3738                         if ( $refers[$arrayKey] > 1 ) {
3739                                 $anchor .= '_' . $refers[$arrayKey];
3740                         }
3741                         if ( $legacyHeadline !== false && $refers[$legacyArrayKey] > 1 ) {
3742                                 $legacyAnchor .= '_' . $refers[$legacyArrayKey];
3743                         }
3744                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3745                                 $toc .= $sk->tocLine($anchor, $tocline,
3746                                         $numbering, $toclevel, ($isTemplate ? false : $sectionIndex));
3747                         }
3748
3749                         # Add the section to the section tree
3750                         # Find the DOM node for this header
3751                         while ( $node && !$isTemplate ) {
3752                                 if ( $node->getName() === 'h' ) {
3753                                         $bits = $node->splitHeading();
3754                                         if ( $bits['i'] == $sectionIndex )
3755                                                 break;
3756                                 }
3757                                 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
3758                                         $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
3759                                 $node = $node->getNextSibling();
3760                         }
3761                         $tocraw[] = array(
3762                                 'toclevel' => $toclevel,
3763                                 'level' => $level,
3764                                 'line' => $tocline,
3765                                 'number' => $numbering,
3766                                 'index' => ($isTemplate ? 'T-' : '' ) . $sectionIndex,
3767                                 'fromtitle' => $titleText,
3768                                 'byteoffset' => ( $isTemplate ? null : $byteOffset ),
3769                                 'anchor' => $anchor,
3770                         );
3771
3772                         # give headline the correct <h#> tag
3773                         if( $showEditLink && $sectionIndex !== false ) {
3774                                 if( $isTemplate ) {
3775                                         # Put a T flag in the section identifier, to indicate to extractSections()
3776                                         # that sections inside <includeonly> should be counted.
3777                                         $editlink = $sk->doEditSectionLink(Title::newFromText( $titleText ), "T-$sectionIndex");
3778                                 } else {
3779                                         $editlink = $sk->doEditSectionLink($this->mTitle, $sectionIndex, $headlineHint);
3780                                 }
3781                         } else {
3782                                 $editlink = '';
3783                         }
3784                         $head[$headlineCount] = $sk->makeHeadline( $level,
3785                                 $matches['attrib'][$headlineCount], $anchor, $headline,
3786                                 $editlink, $legacyAnchor );
3787
3788                         $headlineCount++;
3789                 }
3790
3791                 $this->setOutputType( $oldType );
3792
3793                 # Never ever show TOC if no headers
3794                 if( $numVisible < 1 ) {
3795                         $enoughToc = false;
3796                 }
3797
3798                 if( $enoughToc ) {
3799                         if( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
3800                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
3801                         }
3802                         $toc = $sk->tocList( $toc );
3803                         $this->mOutput->setTOCHTML( $toc );
3804                 }
3805
3806                 if ( $isMain ) {
3807                         $this->mOutput->setSections( $tocraw );
3808                 }
3809
3810                 # split up and insert constructed headlines
3811
3812                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3813                 $i = 0;
3814
3815                 foreach( $blocks as $block ) {
3816                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block !== "\n" ) {
3817                                 # This is the [edit] link that appears for the top block of text when
3818                                 # section editing is enabled
3819
3820                                 # Disabled because it broke block formatting
3821                                 # For example, a bullet point in the top line
3822                                 # $full .= $sk->editSectionLink(0);
3823                         }
3824                         $full .= $block;
3825                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3826                                 # Top anchor now in skin
3827                                 $full = $full.$toc;
3828                         }
3829
3830                         if( !empty( $head[$i] ) ) {
3831                                 $full .= $head[$i];
3832                         }
3833                         $i++;
3834                 }
3835                 if( $this->mForceTocPosition ) {
3836                         return str_replace( '<!--MWTOC-->', $toc, $full );
3837                 } else {
3838                         return $full;
3839                 }
3840         }
3841
3842         /**
3843          * Merge $tree2 into $tree1 by replacing the section with index
3844          * $section in $tree1 and its descendants with the sections in $tree2.
3845          * Note that in the returned section tree, only the 'index' and
3846          * 'byteoffset' fields are guaranteed to be correct.
3847          * @param $tree1 array Section tree from ParserOutput::getSectons()
3848         *  @param $tree2 array Section tree
3849          * @param $section int Section index
3850          * @param $title Title Title both section trees come from
3851          * @param $len2 int Length of the original wikitext for $tree2
3852          * @return array Merged section tree
3853          */
3854         public static function mergeSectionTrees( $tree1, $tree2, $section, $title, $len2 ) {
3855                 global $wgContLang;
3856                 $newTree = array();
3857                 $targetLevel = false;
3858                 $merged = false;
3859                 $lastLevel = 1;
3860                 $nextIndex = 1;
3861                 $numbering = array( 0 );
3862                 $titletext = $title->getPrefixedDBkey();
3863                 foreach ( $tree1 as $s ) {
3864                         if ( $targetLevel !== false ) {
3865                                 if ( $s['level'] <= $targetLevel )
3866                                         // We've skipped enough
3867                                         $targetLevel = false;
3868                                 else
3869                                         continue;
3870                         }
3871                         if ( $s['index'] != $section ||
3872                                         $s['fromtitle'] != $titletext ) {
3873                                 self::incrementNumbering( $numbering,
3874                                         $s['toclevel'], $lastLevel );
3875
3876                                 // Rewrite index, byteoffset and number
3877                                 if ( $s['fromtitle'] == $titletext ) {
3878                                         $s['index'] = $nextIndex++;
3879                                         if ( $merged )
3880                                                 $s['byteoffset'] += $len2;
3881                                 }
3882                                 $s['number']  = implode( '.', array_map(
3883                                         array( $wgContLang, 'formatnum' ),
3884                                         $numbering ) );
3885                                 $lastLevel = $s['toclevel'];
3886                                 $newTree[] = $s;
3887                         } else {
3888                                 // We're at $section
3889                                 // Insert sections from $tree2 here
3890                                 foreach ( $tree2 as $s2 ) {
3891                                         // Rewrite the fields in $s2
3892                                         // before inserting it
3893                                         $s2['toclevel'] += $s['toclevel'] - 1;
3894                                         $s2['level'] += $s['level'] - 1;
3895                                         $s2['index'] = $nextIndex++;
3896                                         $s2['byteoffset'] += $s['byteoffset'];
3897
3898                                         self::incrementNumbering( $numbering,
3899                                                 $s2['toclevel'], $lastLevel );
3900                                         $s2['number']  = implode( '.', array_map(
3901                                                 array( $wgContLang, 'formatnum' ),
3902                                                 $numbering ) );
3903                                         $lastLevel = $s2['toclevel'];
3904                                         $newTree[] = $s2;
3905                                 }
3906                                 // Skip all descendants of $section in $tree1
3907                                 $targetLevel = $s['level'];
3908                                 $merged = true;
3909                         }
3910                 }
3911                 return $newTree;
3912         }
3913
3914         /**
3915          * Increment a section number. Helper function for mergeSectionTrees()
3916          * @param $number array Array representing a section number
3917          * @param $level int Current TOC level (depth)
3918          * @param $lastLevel int Level of previous TOC entry
3919          */
3920         private static function incrementNumbering( &$number, $level, $lastLevel ) {
3921                 if ( $level > $lastLevel )
3922                         $number[$level - 1] = 1;
3923                 else if ( $level < $lastLevel ) {
3924                         foreach ( $number as $key => $unused )
3925                                 if ( $key >= $level )
3926                                         unset( $number[$key] );
3927                         $number[$level - 1]++;
3928                 } else
3929                         $number[$level - 1]++;
3930         }
3931
3932         /**
3933          * Transform wiki markup when saving a page by doing \r\n -> \n
3934          * conversion, substitting signatures, {{subst:}} templates, etc.
3935          *
3936          * @param string $text the text to transform
3937          * @param Title &$title the Title object for the current article
3938          * @param User $user the User object describing the current user
3939          * @param ParserOptions $options parsing options
3940          * @param bool $clearState whether to clear the parser state first
3941          * @return string the altered wiki markup
3942          * @public
3943          */
3944         function preSaveTransform( $text, Title $title, $user, $options, $clearState = true ) {
3945                 $this->mOptions = $options;
3946                 $this->setTitle( $title );
3947                 $this->setOutputType( self::OT_WIKI );
3948
3949                 if ( $clearState ) {
3950                         $this->clearState();
3951                 }
3952
3953                 $pairs = array(
3954                         "\r\n" => "\n",
3955                 );
3956                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3957                 $text = $this->pstPass2( $text, $user );
3958                 $text = $this->mStripState->unstripBoth( $text );
3959                 return $text;
3960         }
3961
3962         /**
3963          * Pre-save transform helper function
3964          * @private
3965          */
3966         function pstPass2( $text, $user ) {
3967                 global $wgContLang, $wgLocaltimezone;
3968
3969                 /* Note: This is the timestamp saved as hardcoded wikitext to
3970                  * the database, we use $wgContLang here in order to give
3971                  * everyone the same signature and use the default one rather
3972                  * than the one selected in each user's preferences.
3973                  *
3974                  * (see also bug 12815)
3975                  */
3976                 $ts = $this->mOptions->getTimestamp();
3977                 if ( isset( $wgLocaltimezone ) ) {
3978                         $tz = $wgLocaltimezone;
3979                 } else {
3980                         $tz = date_default_timezone_get();
3981                 }
3982
3983                 $unixts = wfTimestamp( TS_UNIX, $ts );
3984                 $oldtz = date_default_timezone_get();
3985                 date_default_timezone_set( $tz );
3986                 $ts = date( 'YmdHis', $unixts );
3987                 $tzMsg = date( 'T', $unixts );  # might vary on DST changeover!
3988
3989                 /* Allow translation of timezones trough wiki. date() can return
3990                  * whatever crap the system uses, localised or not, so we cannot
3991                  * ship premade translations.
3992                  */
3993                 $key = 'timezone-' . strtolower( trim( $tzMsg ) );
3994                 $value = wfMsgForContent( $key );
3995                 if ( !wfEmptyMsg( $key, $value ) ) $tzMsg = $value;
3996
3997                 date_default_timezone_set( $oldtz );
3998
3999                 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4000
4001                 # Variable replacement
4002                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4003                 $text = $this->replaceVariables( $text );
4004
4005                 # Signatures
4006                 $sigText = $this->getUserSig( $user );
4007                 $text = strtr( $text, array(
4008                         '~~~~~' => $d,
4009                         '~~~~' => "$sigText $d",
4010                         '~~~' => $sigText
4011                 ) );
4012
4013                 # Context links: [[|name]] and [[name (context)|]]
4014                 #
4015                 global $wgLegalTitleChars;
4016                 $tc = "[$wgLegalTitleChars]";
4017                 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4018
4019                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
4020                 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)(（$tc+）)\\|]]/";             # [[ns:page（context）|]]
4021                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
4022                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
4023
4024                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4025                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4026                 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4027                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4028
4029                 $t = $this->mTitle->getText();
4030                 $m = array();
4031                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4032                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4033                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4034                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4035                 } else {
4036                         # if there's no context, don't bother duplicating the title
4037                         $text = preg_replace( $p2, '[[\\1]]', $text );
4038                 }
4039
4040                 # Trim trailing whitespace
4041                 $text = rtrim( $text );
4042
4043                 return $text;
4044         }
4045
4046         /**
4047          * Fetch the user's signature text, if any, and normalize to
4048          * validated, ready-to-insert wikitext.
4049          * If you have pre-fetched the nickname or the fancySig option, you can
4050          * specify them here to save a database query.
4051          *
4052          * @param User $user
4053          * @return string
4054          */
4055         function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4056                 global $wgMaxSigChars;
4057
4058                 $username = $user->getName();
4059
4060                 // If not given, retrieve from the user object.
4061                 if ( $nickname === false )
4062                         $nickname = $user->getOption( 'nickname' );
4063
4064                 if ( is_null( $fancySig ) )
4065                         $fancySig = $user->getBoolOption( 'fancysig' );
4066
4067                 $nickname = $nickname == null ? $username : $nickname;
4068
4069                 if( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4070                         $nickname = $username;
4071                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4072                 } elseif( $fancySig !== false ) {
4073                         # Sig. might contain markup; validate this
4074                         if( $this->validateSig( $nickname ) !== false ) {
4075                                 # Validated; clean up (if needed) and return it
4076                                 return $this->cleanSig( $nickname, true );
4077                         } else {
4078                                 # Failed to validate; fall back to the default
4079                                 $nickname = $username;
4080                                 wfDebug( __METHOD__.": $username has bad XML tags in signature.\n" );
4081                         }
4082                 }
4083
4084                 // Make sure nickname doesnt get a sig in a sig
4085                 $nickname = $this->cleanSigInSig( $nickname );
4086
4087                 # If we're still here, make it a link to the user page
4088                 $userText = wfEscapeWikiText( $username );
4089                 $nickText = wfEscapeWikiText( $nickname );
4090                 if ( $user->isAnon() )  {
4091                         return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
4092                 } else {
4093                         return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
4094                 }
4095         }
4096
4097         /**
4098          * Check that the user's signature contains no bad XML
4099          *
4100          * @param string $text
4101          * @return mixed An expanded string, or false if invalid.
4102          */
4103         function validateSig( $text ) {
4104                 return( Xml::isWellFormedXmlFragment( $text ) ? $text : false );
4105         }
4106
4107         /**
4108          * Clean up signature text
4109          *
4110          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
4111          * 2) Substitute all transclusions
4112          *
4113          * @param string $text
4114          * @param $parsing Whether we're cleaning (preferences save) or parsing
4115          * @return string Signature text
4116          */
4117         function cleanSig( $text, $parsing = false ) {
4118                 if ( !$parsing ) {
4119                         global $wgTitle;
4120                         $this->clearState();
4121                         $this->setTitle( $wgTitle );
4122                         $this->mOptions = new ParserOptions;
4123                         $this->setOutputType = self::OT_PREPROCESS;
4124                 }
4125
4126                 # Option to disable this feature
4127                 if ( !$this->mOptions->getCleanSignatures() ) {
4128                         return $text;
4129                 }
4130
4131                 # FIXME: regex doesn't respect extension tags or nowiki
4132                 #  => Move this logic to braceSubstitution()
4133                 $substWord = MagicWord::get( 'subst' );
4134                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4135                 $substText = '{{' . $substWord->getSynonym( 0 );
4136
4137                 $text = preg_replace( $substRegex, $substText, $text );
4138                 $text = $this->cleanSigInSig( $text );
4139                 $dom = $this->preprocessToDom( $text );
4140                 $frame = $this->getPreprocessor()->newFrame();
4141                 $text = $frame->expand( $dom );
4142
4143                 if ( !$parsing ) {
4144                         $text = $this->mStripState->unstripBoth( $text );
4145                 }
4146
4147                 return $text;
4148         }
4149
4150         /**
4151          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4152          * @param string $text
4153          * @return string Signature text with /~{3,5}/ removed
4154          */
4155         function cleanSigInSig( $text ) {
4156                 $text = preg_replace( '/~{3,5}/', '', $text );
4157                 return $text;
4158         }
4159
4160         /**
4161          * Set up some variables which are usually set up in parse()
4162          * so that an external function can call some class members with confidence
4163          * @public
4164          */
4165         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
4166                 $this->setTitle( $title );
4167                 $this->mOptions = $options;
4168                 $this->setOutputType( $outputType );
4169                 if ( $clearState ) {
4170                         $this->clearState();
4171                 }
4172         }
4173
4174         /**
4175          * Wrapper for preprocess()
4176          *
4177          * @param string $text the text to preprocess
4178          * @param ParserOptions $options  options
4179          * @return string
4180          * @public
4181          */
4182         function transformMsg( $text, $options ) {
4183                 global $wgTitle;
4184                 static $executing = false;
4185
4186                 # Guard against infinite recursion
4187                 if ( $executing ) {
4188                         return $text;
4189                 }
4190                 $executing = true;
4191
4192                 wfProfileIn(__METHOD__);
4193                 $text = $this->preprocess( $text, $wgTitle, $options );
4194
4195                 $executing = false;
4196                 wfProfileOut(__METHOD__);
4197                 return $text;
4198         }
4199
4200         /**
4201          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
4202          * The callback should have the following form:
4203          *    function myParserHook( $text, $params, &$parser ) { ... }
4204          *
4205          * Transform and return $text. Use $parser for any required context, e.g. use
4206          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4207          *
4208          * @public
4209          *
4210          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
4211          * @param mixed $callback The callback function (and object) to use for the tag
4212          *
4213          * @return The old value of the mTagHooks array associated with the hook
4214          */
4215         function setHook( $tag, $callback ) {
4216                 $tag = strtolower( $tag );
4217                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4218                 $this->mTagHooks[$tag] = $callback;
4219                 if( !in_array( $tag, $this->mStripList ) ) {
4220                         $this->mStripList[] = $tag;
4221                 }
4222
4223                 return $oldVal;
4224         }
4225
4226         function setTransparentTagHook( $tag, $callback ) {
4227                 $tag = strtolower( $tag );
4228                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4229                 $this->mTransparentTagHooks[$tag] = $callback;
4230
4231                 return $oldVal;
4232         }
4233
4234         /**
4235          * Remove all tag hooks
4236          */
4237         function clearTagHooks() {
4238                 $this->mTagHooks = array();
4239                 $this->mStripList = $this->mDefaultStripList;
4240         }
4241
4242         /**
4243          * Create a function, e.g. {{sum:1|2|3}}
4244          * The callback function should have the form:
4245          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4246          *
4247          * Or with SFH_OBJECT_ARGS:
4248          *    function myParserFunction( $parser, $frame, $args ) { ... }
4249          *
4250          * The callback may either return the text result of the function, or an array with the text
4251          * in element 0, and a number of flags in the other elements. The names of the flags are
4252          * specified in the keys. Valid flags are:
4253          *   found                     The text returned is valid, stop processing the template. This
4254          *                             is on by default.
4255          *   nowiki                    Wiki markup in the return value should be escaped
4256          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4257          *
4258          * @public
4259          *
4260          * @param string $id The magic word ID
4261          * @param mixed $callback The callback function (and object) to use
4262          * @param integer $flags a combination of the following flags:
4263          *     SFH_NO_HASH   No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4264          *
4265          *     SFH_OBJECT_ARGS   Pass the template arguments as PPNode objects instead of text. This
4266          *     allows for conditional expansion of the parse tree, allowing you to eliminate dead
4267          *     branches and thus speed up parsing. It is also possible to analyse the parse tree of
4268          *     the arguments, and to control the way they are expanded.
4269          *
4270          *     The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4271          *     arguments, for instance:
4272          *         $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4273          *
4274          *     For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4275          *     future versions. Please call $frame->expand() on it anyway so that your code keeps
4276          *     working if/when this is changed.
4277          *
4278          *     If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4279          *     expansion.
4280          *
4281          *     Please read the documentation in includes/parser/Preprocessor.php for more information
4282          *     about the methods available in PPFrame and PPNode.
4283          *
4284          * @return The old callback function for this name, if any
4285          */
4286         function setFunctionHook( $id, $callback, $flags = 0 ) {
4287                 global $wgContLang;
4288
4289                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4290                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4291
4292                 # Add to function cache
4293                 $mw = MagicWord::get( $id );
4294                 if( !$mw )
4295                         throw new MWException( __METHOD__.'() expecting a magic word identifier.' );
4296
4297                 $synonyms = $mw->getSynonyms();
4298                 $sensitive = intval( $mw->isCaseSensitive() );
4299
4300                 foreach ( $synonyms as $syn ) {
4301                         # Case
4302                         if ( !$sensitive ) {
4303                                 $syn = $wgContLang->lc( $syn );
4304                         }
4305                         # Add leading hash
4306                         if ( !( $flags & SFH_NO_HASH ) ) {
4307                                 $syn = '#' . $syn;
4308                         }
4309                         # Remove trailing colon
4310                         if ( substr( $syn, -1, 1 ) === ':' ) {
4311                                 $syn = substr( $syn, 0, -1 );
4312                         }
4313                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4314                 }
4315                 return $oldVal;
4316         }
4317
4318         /**
4319          * Get all registered function hook identifiers
4320          *
4321          * @return array
4322          */
4323         function getFunctionHooks() {
4324                 return array_keys( $this->mFunctionHooks );
4325         }
4326
4327         /**
4328          * Create a tag function, e.g. <test>some stuff</test>.
4329          * Unlike tag hooks, tag functions are parsed at preprocessor level.
4330          * Unlike parser functions, their content is not preprocessed.
4331          */
4332         function setFunctionTagHook( $tag, $callback, $flags ) {
4333                 $tag = strtolower( $tag );
4334                 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
4335                         $this->mFunctionTagHooks[$tag] : null;
4336                 $this->mFunctionTagHooks[$tag] = array( $callback, $flags );
4337
4338                 if( !in_array( $tag, $this->mStripList ) ) {
4339                         $this->mStripList[] = $tag;
4340                 }
4341
4342                 return $old;
4343         }
4344
4345         /**
4346          * FIXME: update documentation. makeLinkObj() is deprecated.
4347          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4348          * Placeholders created in Skin::makeLinkObj()
4349          * Returns an array of link CSS classes, indexed by PDBK.
4350          */
4351         function replaceLinkHolders( &$text, $options = 0 ) {
4352                 return $this->mLinkHolders->replace( $text );
4353         }
4354
4355         /**
4356          * Replace <!--LINK--> link placeholders with plain text of links
4357          * (not HTML-formatted).
4358          * @param string $text
4359          * @return string
4360          */
4361         function replaceLinkHoldersText( $text ) {
4362                 return $this->mLinkHolders->replaceText( $text );
4363         }
4364
4365         /**
4366          * Tag hook handler for 'pre'.
4367          */
4368         function renderPreTag( $text, $attribs ) {
4369                 // Backwards-compatibility hack
4370                 $content = StringUtils::delimiterReplace( '<nowiki>', '</nowiki>', '$1', $text, 'i' );
4371
4372                 $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
4373                 return Xml::openElement( 'pre', $attribs ) .
4374                         Xml::escapeTagsOnly( $content ) .
4375                         '</pre>';
4376         }
4377
4378         /**
4379         * Tag hook handler for 'a'. Renders a HTML &lt;a&gt; tag, allowing most attributes, filtering href against
4380         * allowed protocols and spam blacklist.
4381         **/
4382         function renderHyperlink( $text, $params, $frame = false ) {
4383                 foreach ( $params as $name => $value ) {
4384                         $params[ $name ] = $this->replaceVariables( $value, $frame );
4385                 }
4386
4387                 $whitelist = Sanitizer::attributeWhitelist( 'a' );
4388                 $params = Sanitizer::validateAttributes( $params, $whitelist );
4389
4390                 $content = $this->recursiveTagParse( trim( $text ), $frame );
4391
4392                 if ( isset( $params[ 'href' ] ) ) {
4393                         $href = $params[ 'href' ];
4394                         $this->mOutput->addExternalLink( $href );
4395                         unset( $params[ 'href' ] );
4396                 } else {
4397                         # Non-link <a> tag
4398                         return Xml::openElement( 'a', $params ) . $content . Xml::closeElement( 'a' );
4399                 }
4400
4401                 $sk = $this->mOptions->getSkin();
4402                 $html = $sk->makeExternalLink( $href, $content, false, '', $params );
4403
4404                 return $html;
4405         }
4406
4407         /**
4408          * Renders an image gallery from a text with one line per image.
4409          * text labels may be given by using |-style alternative text. E.g.
4410          *   Image:one.jpg|The number "1"
4411          *   Image:tree.jpg|A tree
4412          * given as text will return the HTML of a gallery with two images,
4413          * labeled 'The number "1"' and
4414          * 'A tree'.
4415          */
4416         function renderImageGallery( $text, $params ) {
4417                 $ig = new ImageGallery();
4418                 $ig->setContextTitle( $this->mTitle );
4419                 $ig->setShowBytes( false );
4420                 $ig->setShowFilename( false );
4421                 $ig->setParser( $this );
4422                 $ig->setHideBadImages();
4423                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4424                 $ig->useSkin( $this->mOptions->getSkin() );
4425                 $ig->mRevisionId = $this->mRevisionId;
4426
4427                 if( isset( $params['caption'] ) ) {
4428                         $caption = $params['caption'];
4429                         $caption = htmlspecialchars( $caption );
4430                         $caption = $this->replaceInternalLinks( $caption );
4431                         $ig->setCaptionHtml( $caption );
4432                 }
4433                 if( isset( $params['perrow'] ) ) {
4434                         $ig->setPerRow( $params['perrow'] );
4435                 }
4436                 if( isset( $params['widths'] ) ) {
4437                         $ig->setWidths( $params['widths'] );
4438                 }
4439                 if( isset( $params['heights'] ) ) {
4440                         $ig->setHeights( $params['heights'] );
4441                 }
4442
4443                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4444
4445                 $lines = StringUtils::explode( "\n", $text );
4446                 foreach ( $lines as $line ) {
4447                         # match lines like these:
4448                         # Image:someimage.jpg|This is some image
4449                         $matches = array();
4450                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4451                         # Skip empty lines
4452                         if ( count( $matches ) == 0 ) {
4453                                 continue;
4454                         }
4455
4456                         if ( strpos( $matches[0], '%' ) !== false )
4457                                 $matches[1] = urldecode( $matches[1] );
4458                         $tp = Title::newFromText( $matches[1]/*, NS_FILE*/ );
4459                         $nt =& $tp;
4460                         if( is_null( $nt ) ) {
4461                                 # Bogus title. Ignore these so we don't bomb out later.
4462                                 continue;
4463                         }
4464                         if ( isset( $matches[3] ) ) {
4465                                 $label = $matches[3];
4466                         } else {
4467                                 $label = '';
4468                         }
4469
4470                         $html = $this->recursiveTagParse( trim( $label ) );
4471
4472                         $ig->add( $nt, $html );
4473
4474                         # Only add real images (bug #5586)
4475                         if ( $nt->getNamespace() == NS_FILE ) {
4476                                 $this->mOutput->addImage( $nt->getDBkey() );
4477                         }
4478                 }
4479                 return $ig->toHTML();
4480         }
4481
4482         function getImageParams( $handler ) {
4483                 if ( $handler ) {
4484                         $handlerClass = get_class( $handler );
4485                 } else {
4486                         $handlerClass = '';
4487                 }
4488                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4489                         // Initialise static lists
4490                         static $internalParamNames = array(
4491                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4492                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4493                                         'bottom', 'text-bottom' ),
4494                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4495                                         'upright', 'border', 'link', 'alt' ),
4496                         );
4497                         static $internalParamMap;
4498                         if ( !$internalParamMap ) {
4499                                 $internalParamMap = array();
4500                                 foreach ( $internalParamNames as $type => $names ) {
4501                                         foreach ( $names as $name ) {
4502                                                 $magicName = str_replace( '-', '_', "img_$name" );
4503                                                 $internalParamMap[$magicName] = array( $type, $name );
4504                                         }
4505                                 }
4506                         }
4507
4508                         // Add handler params
4509                         $paramMap = $internalParamMap;
4510                         if ( $handler ) {
4511                                 $handlerParamMap = $handler->getParamMap();
4512                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4513                                         $paramMap[$magic] = array( 'handler', $paramName );
4514                                 }
4515                         }
4516                         $this->mImageParams[$handlerClass] = $paramMap;
4517                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4518                 }
4519                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4520         }
4521
4522         /**
4523          * Parse image options text and use it to make an image
4524          * @param Title $title
4525          * @param string $options
4526          * @param LinkHolderArray $holders
4527          */
4528         function makeImage( $title, $options, $holders = false ) {
4529                 # Check if the options text is of the form "options|alt text"
4530                 # Options are:
4531                 #  * thumbnail  make a thumbnail with enlarge-icon and caption, alignment depends on lang
4532                 #  * left       no resizing, just left align. label is used for alt= only
4533                 #  * right      same, but right aligned
4534                 #  * none       same, but not aligned
4535                 #  * ___px      scale to ___ pixels width, no aligning. e.g. use in taxobox
4536                 #  * center     center the image
4537                 #  * frame      Keep original image size, no magnify-button.
4538                 #  * framed     Same as "frame"
4539                 #  * frameless  like 'thumb' but without a frame. Keeps user preferences for width
4540                 #  * upright    reduce width for upright images, rounded to full __0 px
4541                 #  * border     draw a 1px border around the image
4542                 #  * alt        Text for HTML alt attribute (defaults to empty)
4543                 #  * link       Set the target of the image link. Can be external, interwiki, or local
4544                 # vertical-align values (no % or length right now):
4545                 #  * baseline
4546                 #  * sub
4547                 #  * super
4548                 #  * top
4549                 #  * text-top
4550                 #  * middle
4551                 #  * bottom
4552                 #  * text-bottom
4553
4554                 $parts = StringUtils::explode( "|", $options );
4555                 $sk = $this->mOptions->getSkin();
4556
4557                 # Give extensions a chance to select the file revision for us
4558                 $skip = $time = $descQuery = false;
4559                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time, &$descQuery ) );
4560
4561                 if ( $skip ) {
4562                         return $sk->link( $title );
4563                 }
4564
4565                 # Get the file
4566                 $imagename = $title->getDBkey();
4567                 $file = wfFindFile( $title, array( 'time' => $time ) );
4568                 # Get parameter map
4569                 $handler = $file ? $file->getHandler() : false;
4570
4571                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4572
4573                 # Process the input parameters
4574                 $caption = '';
4575                 $params = array( 'frame' => array(), 'handler' => array(),
4576                         'horizAlign' => array(), 'vertAlign' => array() );
4577                 foreach( $parts as $part ) {
4578                         $part = trim( $part );
4579                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4580                         $validated = false;
4581                         if( isset( $paramMap[$magicName] ) ) {
4582                                 list( $type, $paramName ) = $paramMap[$magicName];
4583
4584                                 // Special case; width and height come in one variable together
4585                                 if( $type === 'handler' && $paramName === 'width' ) {
4586                                         $m = array();
4587                                         # (bug 13500) In both cases (width/height and width only),
4588                                         # permit trailing "px" for backward compatibility.
4589                                         if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
4590                                                 $width = intval( $m[1] );
4591                                                 $height = intval( $m[2] );
4592                                                 if ( $handler->validateParam( 'width', $width ) ) {
4593                                                         $params[$type]['width'] = $width;
4594                                                         $validated = true;
4595                                                 }
4596                                                 if ( $handler->validateParam( 'height', $height ) ) {
4597                                                         $params[$type]['height'] = $height;
4598                                                         $validated = true;
4599                                                 }
4600                                         } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
4601                                                 $width = intval( $value );
4602                                                 if ( $handler->validateParam( 'width', $width ) ) {
4603                                                         $params[$type]['width'] = $width;
4604                                                         $validated = true;
4605                                                 }
4606                                         } // else no validation -- bug 13436
4607                                 } else {
4608                                         if ( $type === 'handler' ) {
4609                                                 # Validate handler parameter
4610                                                 $validated = $handler->validateParam( $paramName, $value );
4611                                         } else {
4612                                                 # Validate internal parameters
4613                                                 switch( $paramName ) {
4614                                                 case 'manualthumb':
4615                                                 case 'alt':
4616                                                         // @todo Fixme: possibly check validity here for
4617                                                         // manualthumb? downstream behavior seems odd with
4618                                                         // missing manual thumbs.
4619                                                         $validated = true;
4620                                                         $value = $this->stripAltText( $value, $holders );
4621                                                         break;
4622                                                 case 'link':
4623                                                         $chars = self::EXT_LINK_URL_CLASS;
4624                                                         $prots = $this->mUrlProtocols;
4625                                                         if ( $value === '' ) {
4626                                                                 $paramName = 'no-link';
4627                                                                 $value = true;
4628                                                                 $validated = true;
4629                                                         } elseif ( preg_match( "/^$prots/", $value ) ) {
4630                                                                 if ( preg_match( "/^($prots)$chars+$/", $value, $m ) ) {
4631                                                                         $paramName = 'link-url';
4632                                                                         $this->mOutput->addExternalLink( $value );
4633                                                                         $validated = true;
4634                                                                 }
4635                                                         } else {
4636                                                                 $linkTitle = Title::newFromText( $value );
4637                                                                 if ( $linkTitle ) {
4638                                                                         $paramName = 'link-title';
4639                                                                         $value = $linkTitle;
4640                                                                         $this->mOutput->addLink( $linkTitle );
4641                                                                         $validated = true;
4642                                                                 }
4643                                                         }
4644                                                         break;
4645                                                 default:
4646                                                         // Most other things appear to be empty or numeric...
4647                                                         $validated = ( $value === false || is_numeric( trim( $value ) ) );
4648                                                 }
4649                                         }
4650
4651                                         if ( $validated ) {
4652                                                 $params[$type][$paramName] = $value;
4653                                         }
4654                                 }
4655                         }
4656                         if ( !$validated ) {
4657                                 $caption = $part;
4658                         }
4659                 }
4660
4661                 # Process alignment parameters
4662                 if ( $params['horizAlign'] ) {
4663                         $params['frame']['align'] = key( $params['horizAlign'] );
4664                 }
4665                 if ( $params['vertAlign'] ) {
4666                         $params['frame']['valign'] = key( $params['vertAlign'] );
4667                 }
4668
4669                 $params['frame']['caption'] = $caption;
4670
4671                 # Will the image be presented in a frame, with the caption below?
4672                 $imageIsFramed = isset( $params['frame']['frame'] ) ||
4673                                  isset( $params['frame']['framed'] ) ||
4674                                  isset( $params['frame']['thumbnail'] ) ||
4675                                  isset( $params['frame']['manualthumb'] );
4676
4677                 # In the old days, [[Image:Foo|text...]] would set alt text.  Later it
4678                 # came to also set the caption, ordinary text after the image -- which
4679                 # makes no sense, because that just repeats the text multiple times in
4680                 # screen readers.  It *also* came to set the title attribute.
4681                 #
4682                 # Now that we have an alt attribute, we should not set the alt text to
4683                 # equal the caption: that's worse than useless, it just repeats the
4684                 # text.  This is the framed/thumbnail case.  If there's no caption, we
4685                 # use the unnamed parameter for alt text as well, just for the time be-
4686                 # ing, if the unnamed param is set and the alt param is not.
4687                 #
4688                 # For the future, we need to figure out if we want to tweak this more,
4689                 # e.g., introducing a title= parameter for the title; ignoring the un-
4690                 # named parameter entirely for images without a caption; adding an ex-
4691                 # plicit caption= parameter and preserving the old magic unnamed para-
4692                 # meter for BC; ...
4693                 if ( $imageIsFramed ) { # Framed image
4694                         if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
4695                                 # No caption or alt text, add the filename as the alt text so
4696                                 # that screen readers at least get some description of the image
4697                                 $params['frame']['alt'] = $title->getText();
4698                         }
4699                         # Do not set $params['frame']['title'] because tooltips don't make sense
4700                         # for framed images
4701                 } else { # Inline image
4702                         if ( !isset( $params['frame']['alt'] ) ) {
4703                                 # No alt text, use the "caption" for the alt text
4704                                 if ( $caption !== '') {
4705                                         $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
4706                                 } else {
4707                                         # No caption, fall back to using the filename for the
4708                                         # alt text
4709                                         $params['frame']['alt'] = $title->getText();
4710                                 }
4711                         }
4712                         # Use the "caption" for the tooltip text
4713                         $params['frame']['title'] = $this->stripAltText( $caption, $holders );
4714                 }
4715
4716                 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params ) );
4717
4718                 # Linker does the rest
4719                 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'], $time, $descQuery );
4720
4721                 # Give the handler a chance to modify the parser object
4722                 if ( $handler ) {
4723                         $handler->parserTransformHook( $this, $file );
4724                 }
4725
4726                 return $ret;
4727         }
4728
4729         protected function stripAltText( $caption, $holders ) {
4730                 # Strip bad stuff out of the title (tooltip).  We can't just use
4731                 # replaceLinkHoldersText() here, because if this function is called
4732                 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
4733                 if ( $holders ) {
4734                         $tooltip = $holders->replaceText( $caption );
4735                 } else {
4736                         $tooltip = $this->replaceLinkHoldersText( $caption );
4737                 }
4738
4739                 # make sure there are no placeholders in thumbnail attributes
4740                 # that are later expanded to html- so expand them now and
4741                 # remove the tags
4742                 $tooltip = $this->mStripState->unstripBoth( $tooltip );
4743                 $tooltip = Sanitizer::stripAllTags( $tooltip );
4744
4745                 return $tooltip;
4746         }
4747
4748         /**
4749          * Set a flag in the output object indicating that the content is dynamic and
4750          * shouldn't be cached.
4751          */
4752         function disableCache() {
4753                 wfDebug( "Parser output marked as uncacheable.\n" );
4754                 $this->mOutput->mCacheTime = -1;
4755         }
4756
4757         /**#@+
4758          * Callback from the Sanitizer for expanding items found in HTML attribute
4759          * values, so they can be safely tested and escaped.
4760          * @param string $text
4761          * @param PPFrame $frame
4762          * @return string
4763          * @private
4764          */
4765         function attributeStripCallback( &$text, $frame = false ) {
4766                 $text = $this->replaceVariables( $text, $frame );
4767                 $text = $this->mStripState->unstripBoth( $text );
4768                 return $text;
4769         }
4770
4771         /**#@-*/
4772
4773         /**#@+
4774          * Accessor/mutator
4775          */
4776         function Title( $x = null ) { return wfSetVar( $this->mTitle, $x ); }
4777         function Options( $x = null ) { return wfSetVar( $this->mOptions, $x ); }
4778         function OutputType( $x = null ) { return wfSetVar( $this->mOutputType, $x ); }
4779         /**#@-*/
4780
4781         /**#@+
4782          * Accessor
4783          */
4784         function getTags() { return array_merge( array_keys($this->mTransparentTagHooks), array_keys( $this->mTagHooks ) ); }
4785         /**#@-*/
4786
4787
4788         /**
4789          * Break wikitext input into sections, and either pull or replace
4790          * some particular section's text.
4791          *
4792          * External callers should use the getSection and replaceSection methods.
4793          *
4794          * @param string $text Page wikitext
4795          * @param string $section A section identifier string of the form:
4796          *   <flag1> - <flag2> - ... - <section number>
4797          *
4798          * Currently the only recognised flag is "T", which means the target section number
4799          * was derived during a template inclusion parse, in other words this is a template
4800          * section edit link. If no flags are given, it was an ordinary section edit link.
4801          * This flag is required to avoid a section numbering mismatch when a section is
4802          * enclosed by <includeonly> (bug 6563).
4803          *
4804          * The section number 0 pulls the text before the first heading; other numbers will
4805          * pull the given section along with its lower-level subsections. If the section is
4806          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
4807          *
4808          * @param string $mode One of "get" or "replace"
4809          * @param string $newText Replacement text for section data.
4810          * @return string for "get", the extracted section text.
4811          *                for "replace", the whole page with the section replaced.
4812          */
4813         private function extractSections( $text, $section, $mode, $newText='' ) {
4814                 global $wgTitle;
4815                 $this->clearState();
4816                 $this->setTitle( $wgTitle ); // not generally used but removes an ugly failure mode
4817                 $this->mOptions = new ParserOptions;
4818                 $this->setOutputType( self::OT_WIKI );
4819                 $outText = '';
4820                 $frame = $this->getPreprocessor()->newFrame();
4821
4822                 // Process section extraction flags
4823                 $flags = 0;
4824                 $sectionParts = explode( '-', $section );
4825                 $sectionIndex = array_pop( $sectionParts );
4826                 foreach ( $sectionParts as $part ) {
4827                         if ( $part === 'T' ) {
4828                                 $flags |= self::PTD_FOR_INCLUSION;
4829                         }
4830                 }
4831                 // Preprocess the text
4832                 $root = $this->preprocessToDom( $text, $flags );
4833
4834                 // <h> nodes indicate section breaks
4835                 // They can only occur at the top level, so we can find them by iterating the root's children
4836                 $node = $root->getFirstChild();
4837
4838                 // Find the target section
4839                 if ( $sectionIndex == 0 ) {
4840                         // Section zero doesn't nest, level=big
4841                         $targetLevel = 1000;
4842                 } else {
4843             while ( $node ) {
4844                 if ( $node->getName() === 'h' ) {
4845                     $bits = $node->splitHeading();
4846                                         if ( $bits['i'] == $sectionIndex ) {
4847                                         $targetLevel = $bits['level'];
4848                                                 break;
4849                                         }
4850                                 }
4851                                 if ( $mode === 'replace' ) {
4852                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4853                                 }
4854                                 $node = $node->getNextSibling();
4855                         }
4856                 }
4857
4858                 if ( !$node ) {
4859                         // Not found
4860                         if ( $mode === 'get' ) {
4861                                 return $newText;
4862                         } else {
4863                                 return $text;
4864                         }
4865                 }
4866
4867                 // Find the end of the section, including nested sections
4868                 do {
4869                         if ( $node->getName() === 'h' ) {
4870                                 $bits = $node->splitHeading();
4871                                 $curLevel = $bits['level'];
4872                                 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
4873                                         break;
4874                                 }
4875                         }
4876                         if ( $mode === 'get' ) {
4877                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4878                         }
4879                         $node = $node->getNextSibling();
4880                 } while ( $node );
4881
4882                 // Write out the remainder (in replace mode only)
4883                 if ( $mode === 'replace' ) {
4884                         // Output the replacement text
4885                         // Add two newlines on -- trailing whitespace in $newText is conventionally
4886                         // stripped by the editor, so we need both newlines to restore the paragraph gap
4887                         // Only add trailing whitespace if there is newText
4888                         if($newText != "") {
4889                                 $outText .= $newText . "\n\n";
4890                         }
4891
4892                         while ( $node ) {
4893                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4894                                 $node = $node->getNextSibling();
4895                         }
4896                 }
4897
4898                 if ( is_string( $outText ) ) {
4899                         // Re-insert stripped tags
4900                         $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
4901                 }
4902
4903                 return $outText;
4904         }
4905
4906         /**
4907          * This function returns the text of a section, specified by a number ($section).
4908          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4909          * the first section before any such heading (section 0).
4910          *
4911          * If a section contains subsections, these are also returned.
4912          *
4913          * @param string $text text to look in
4914          * @param string $section section identifier
4915          * @param string $deftext default to return if section is not found
4916          * @return string text of the requested section
4917          */
4918         public function getSection( $text, $section, $deftext='' ) {
4919                 return $this->extractSections( $text, $section, "get", $deftext );
4920         }
4921
4922         public function replaceSection( $oldtext, $section, $text ) {
4923                 return $this->extractSections( $oldtext, $section, "replace", $text );
4924         }
4925
4926         /**
4927          * Get the timestamp associated with the current revision, adjusted for
4928          * the default server-local timestamp
4929          */
4930         function getRevisionTimestamp() {
4931                 if ( is_null( $this->mRevisionTimestamp ) ) {
4932                         wfProfileIn( __METHOD__ );
4933                         global $wgContLang;
4934                         $dbr = wfGetDB( DB_SLAVE );
4935                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
4936                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
4937
4938                         // Normalize timestamp to internal MW format for timezone processing.
4939                         // This has the added side-effect of replacing a null value with
4940                         // the current time, which gives us more sensible behavior for
4941                         // previews.
4942                         $timestamp = wfTimestamp( TS_MW, $timestamp );
4943
4944                         // The cryptic '' timezone parameter tells to use the site-default
4945                         // timezone offset instead of the user settings.
4946                         //
4947                         // Since this value will be saved into the parser cache, served
4948                         // to other users, and potentially even used inside links and such,
4949                         // it needs to be consistent for all visitors.
4950                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
4951
4952                         wfProfileOut( __METHOD__ );
4953                 }
4954                 return $this->mRevisionTimestamp;
4955         }
4956
4957         /**
4958          * Get the name of the user that edited the last revision
4959          */
4960         function getRevisionUser() {
4961                 // if this template is subst: the revision id will be blank,
4962                 // so just use the current user's name
4963                 if( $this->mRevisionId ) {
4964                         $revision = Revision::newFromId( $this->mRevisionId );
4965                         $revuser = $revision->getUserText();
4966                 } else {
4967                         global $wgUser;
4968                         $revuser = $wgUser->getName();
4969                 }
4970                 return $revuser;
4971         }
4972
4973         /**
4974          * Mutator for $mDefaultSort
4975          *
4976          * @param $sort New value
4977          */
4978         public function setDefaultSort( $sort ) {
4979                 $this->mDefaultSort = $sort;
4980         }
4981
4982         /**
4983          * Accessor for $mDefaultSort
4984          * Will use the title/prefixed title if none is set
4985          *
4986          * @return string
4987          */
4988         public function getDefaultSort() {
4989                 global $wgCategoryPrefixedDefaultSortkey;
4990                 if( $this->mDefaultSort !== false ) {
4991                         return $this->mDefaultSort;
4992                 } elseif ($this->mTitle->getNamespace() == NS_CATEGORY ||
4993                         !$wgCategoryPrefixedDefaultSortkey) {
4994                         return $this->mTitle->getText();
4995                 } else {
4996                         return $this->mTitle->getPrefixedText();
4997                 }
4998         }
4999
5000         /**
5001          * Accessor for $mDefaultSort
5002          * Unlike getDefaultSort(), will return false if none is set
5003          *
5004          * @return string or false
5005          */
5006         public function getCustomDefaultSort() {
5007                 return $this->mDefaultSort;
5008         }
5009
5010         /**
5011          * Try to guess the section anchor name based on a wikitext fragment
5012          * presumably extracted from a heading, for example "Header" from
5013          * "== Header ==".
5014          */
5015         public function guessSectionNameFromWikiText( $text ) {
5016                 # Strip out wikitext links(they break the anchor)
5017                 $text = $this->stripSectionName( $text );
5018                 $headline = Sanitizer::decodeCharReferences( $text );
5019                 # strip out HTML
5020                 $headline = StringUtils::delimiterReplace( '<', '>', '', $headline );
5021                 $headline = trim( $headline );
5022                 $sectionanchor = '#' . urlencode( str_replace( ' ', '_', $headline ) );
5023                 $replacearray = array(
5024                         '%3A' => ':',
5025                         '%' => '.'
5026                 );
5027                 return str_replace(
5028                         array_keys( $replacearray ),
5029                         array_values( $replacearray ),
5030                         $sectionanchor );
5031         }
5032
5033         /**
5034          * Strips a text string of wikitext for use in a section anchor
5035          *
5036          * Accepts a text string and then removes all wikitext from the
5037          * string and leaves only the resultant text (i.e. the result of
5038          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5039          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5040          * to create valid section anchors by mimicing the output of the
5041          * parser when headings are parsed.
5042          *
5043          * @param $text string Text string to be stripped of wikitext
5044          * for use in a Section anchor
5045          * @return Filtered text string
5046          */
5047         public function stripSectionName( $text ) {
5048                 # Strip internal link markup
5049                 $text = preg_replace('/\[\[:?([^[|]+)\|([^[]+)\]\]/','$2',$text);
5050                 $text = preg_replace('/\[\[:?([^[]+)\|?\]\]/','$1',$text);
5051
5052                 # Strip external link markup (FIXME: Not Tolerant to blank link text
5053                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
5054                 # on how many empty links there are on the page - need to figure that out.
5055                 $text = preg_replace('/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/','$2',$text);
5056
5057                 # Parse wikitext quotes (italics & bold)
5058                 $text = $this->doQuotes($text);
5059
5060                 # Strip HTML tags
5061                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5062                 return $text;
5063         }
5064
5065         function srvus( $text ) {
5066                 return $this->testSrvus( $text, $this->mOutputType );
5067         }
5068
5069         /**
5070          * strip/replaceVariables/unstrip for preprocessor regression testing
5071          */
5072         function testSrvus( $text, $title, $options, $outputType = self::OT_HTML ) {
5073                 $this->clearState();
5074                 if ( ! ( $title instanceof Title ) ) {
5075                         $title = Title::newFromText( $title );
5076                 }
5077                 $this->mTitle = $title;
5078                 $this->mOptions = $options;
5079                 $this->setOutputType( $outputType );
5080                 $text = $this->replaceVariables( $text );
5081                 $text = $this->mStripState->unstripBoth( $text );
5082                 $text = Sanitizer::removeHTMLtags( $text );
5083                 return $text;
5084         }
5085
5086         function testPst( $text, $title, $options ) {
5087                 global $wgUser;
5088                 if ( ! ( $title instanceof Title ) ) {
5089                         $title = Title::newFromText( $title );
5090                 }
5091                 return $this->preSaveTransform( $text, $title, $wgUser, $options );
5092         }
5093
5094         function testPreprocess( $text, $title, $options ) {
5095                 if ( ! ( $title instanceof Title ) ) {
5096                         $title = Title::newFromText( $title );
5097                 }
5098                 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
5099         }
5100
5101         function markerSkipCallback( $s, $callback ) {
5102                 $i = 0;
5103                 $out = '';
5104                 while ( $i < strlen( $s ) ) {
5105                         $markerStart = strpos( $s, $this->mUniqPrefix, $i );
5106                         if ( $markerStart === false ) {
5107                                 $out .= call_user_func( $callback, substr( $s, $i ) );
5108                                 break;
5109                         } else {
5110                                 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5111                                 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
5112                                 if ( $markerEnd === false ) {
5113                                         $out .= substr( $s, $markerStart );
5114                                         break;
5115                                 } else {
5116                                         $markerEnd += strlen( self::MARKER_SUFFIX );
5117                                         $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5118                                         $i = $markerEnd;
5119                                 }
5120                         }
5121                 }
5122                 return $out;
5123         }
5124
5125         function serialiseHalfParsedText( $text ) {
5126                 $data = array();
5127                 $data['text'] = $text;
5128
5129                 // First, find all strip markers, and store their
5130                 //  data in an array.
5131                 $stripState = new StripState;
5132                 $pos = 0;
5133                 while( ( $start_pos = strpos( $text, $this->mUniqPrefix, $pos ) ) && ( $end_pos = strpos( $text, self::MARKER_SUFFIX, $pos ) ) ) {
5134                         $end_pos += strlen( self::MARKER_SUFFIX );
5135                         $marker = substr( $text, $start_pos, $end_pos-$start_pos );
5136
5137                         if ( !empty( $this->mStripState->general->data[$marker] ) ) {
5138                                 $replaceArray = $stripState->general;
5139                                 $stripText = $this->mStripState->general->data[$marker];
5140                         } elseif ( !empty( $this->mStripState->nowiki->data[$marker] ) ) {
5141                                 $replaceArray = $stripState->nowiki;
5142                                 $stripText = $this->mStripState->nowiki->data[$marker];
5143                         } else {
5144                                 throw new MWException( "Hanging strip marker: '$marker'." );
5145                         }
5146
5147                         $replaceArray->setPair( $marker, $stripText );
5148                         $pos = $end_pos;
5149                 }
5150                 $data['stripstate'] = $stripState;
5151
5152                 // Now, find all of our links, and store THEIR
5153                 //  data in an array! :)
5154                 $links = array( 'internal' => array(), 'interwiki' => array() );
5155                 $pos = 0;
5156
5157                 // Internal links
5158                 while( ( $start_pos = strpos( $text, '<!--LINK ', $pos ) ) ) {
5159                         list( $ns, $trail ) = explode( ':', substr( $text, $start_pos + strlen( '<!--LINK ' ) ), 2 );
5160
5161                         $ns = trim($ns);
5162                         if (empty( $links['internal'][$ns] )) {
5163                                 $links['internal'][$ns] = array();
5164                         }
5165
5166                         $key = trim( substr( $trail, 0, strpos( $trail, '-->' ) ) );
5167                         $links['internal'][$ns][] = $this->mLinkHolders->internals[$ns][$key];
5168                         $pos = $start_pos + strlen( "<!--LINK $ns:$key-->" );
5169                 }
5170
5171                 $pos = 0;
5172
5173                 // Interwiki links
5174                 while( ( $start_pos = strpos( $text, '<!--IWLINK ', $pos ) ) ) {
5175                         $data = substr( $text, $start_pos );
5176                         $key = trim( substr( $data, 0, strpos( $data, '-->' ) ) );
5177                         $links['interwiki'][] = $this->mLinkHolders->interwiki[$key];
5178                         $pos = $start_pos + strlen( "<!--IWLINK $key-->" );
5179                 }
5180
5181                 $data['linkholder'] = $links;
5182
5183                 return $data;
5184         }
5185
5186         function unserialiseHalfParsedText( $data, $intPrefix = null /* Unique identifying prefix */ ) {
5187                 if (!$intPrefix)
5188                         $intPrefix = $this->getRandomString();
5189
5190                 // First, extract the strip state.
5191                 $stripState = $data['stripstate'];
5192                 $this->mStripState->general->merge( $stripState->general );
5193                 $this->mStripState->nowiki->merge( $stripState->nowiki );
5194
5195                 // Now, extract the text, and renumber links
5196                 $text = $data['text'];
5197                 $links = $data['linkholder'];
5198
5199                 // Internal...
5200                 foreach( $links['internal'] as $ns => $nsLinks ) {
5201                         foreach( $nsLinks as $key => $entry ) {
5202                                 $newKey = $intPrefix . '-' . $key;
5203                                 $this->mLinkHolders->internals[$ns][$newKey] = $entry;
5204
5205                                 $text = str_replace( "<!--LINK $ns:$key-->", "<!--LINK $ns:$newKey-->", $text );
5206                         }
5207                 }
5208
5209                 // Interwiki...
5210                 foreach( $links['interwiki'] as $key => $entry ) {
5211                         $newKey = "$intPrefix-$key";
5212                         $this->mLinkHolders->interwikis[$newKey] = $entry;
5213
5214                         $text = str_replace( "<!--IWLINK $key-->", "<!--IWLINK $newKey-->", $text );
5215                 }
5216
5217                 // Should be good to go.
5218                 return $text;
5219         }
5220 }
5221
5222 /**
5223  * @todo document, briefly.
5224  * @ingroup Parser
5225  */
5226 class StripState {
5227         var $general, $nowiki;
5228
5229         function __construct() {
5230                 $this->general = new ReplacementArray;
5231                 $this->nowiki = new ReplacementArray;
5232         }
5233
5234         function unstripGeneral( $text ) {
5235                 wfProfileIn( __METHOD__ );
5236                 do {
5237                         $oldText = $text;
5238                         $text = $this->general->replace( $text );
5239                 } while ( $text !== $oldText );
5240                 wfProfileOut( __METHOD__ );
5241                 return $text;
5242         }
5243
5244         function unstripNoWiki( $text ) {
5245                 wfProfileIn( __METHOD__ );
5246                 do {
5247                         $oldText = $text;
5248                         $text = $this->nowiki->replace( $text );
5249                 } while ( $text !== $oldText );
5250                 wfProfileOut( __METHOD__ );
5251                 return $text;
5252         }
5253
5254         function unstripBoth( $text ) {
5255                 wfProfileIn( __METHOD__ );
5256                 do {
5257                         $oldText = $text;
5258                         $text = $this->general->replace( $text );
5259                         $text = $this->nowiki->replace( $text );
5260                 } while ( $text !== $oldText );
5261                 wfProfileOut( __METHOD__ );
5262                 return $text;
5263         }
5264 }
5265
5266 /**
5267  * @todo document, briefly.
5268  * @ingroup Parser
5269  */
5270 class OnlyIncludeReplacer {
5271         var $output = '';
5272
5273         function replace( $matches ) {
5274                 if ( substr( $matches[1], -1 ) === "\n" ) {
5275                         $this->output .= substr( $matches[1], 0, -1 );
5276                 } else {
5277                         $this->output .= $matches[1];
5278                 }
5279         }
5280 }